Ecological Informatics 64 (2021) 101357

Contents lists available at ScienceDirect

Ecological Informatics
journal homepage: www.elsevier.com/locate/ecolinf

Potential distribution of piscivores across the Atlantic Forest: From bats and
marsupials to large-bodied mammals under a trophic-guild viewpoint
Juliano A. Bogoni a, b, c, *, Mario Muniz Tagliari a, d, *
a
Programa de Pós-Graduação em Ecologia, Universidade Federal de Santa Catarina, Florianópolis, SC, Brazil
b
Universidade de São Paulo (USP), Escola Superior de Agricultura “Luiz de Queiroz” (ESALQ), Departamento de Ciências Florestais, Laboratório de Ecologia, Manejo e
Conservação de Fauna Silvestre (LEMaC), Piracicaba, SP 13418-900, Brazil
c
School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, United Kingdom
d
Faculdade Municipal do Meio Ambiente, Clevelândia, PR 85530-000, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: Biodiversity across the Neotropics is a result of selective forces along a myriad of ecosystems. The Brazilian
Tropical forest Atlantic Forest biome embodies 321 mammal species; only a small parcel of the species is considered piscivore.
Bats The Atlantic Forest biota is threatened by multiple factors, from habitat loss to climate changes. We aimed to
Semi-aquatic mammals
understand the current distribution of the six extant piscivore mammals and the distribution of this trophic guild,
Species distribution models
predicting their distribution under climate change across the Atlantic Forest and identifying areas for the trophic
Climate change
guild conservation prioritizations. We used data on species occurrence based on the GBIF database and literature,
and climatic data from WorldClim. We perform Species Distribution Modelling (SDM) to derive the piscivores
distribution and the presence of this trophic guild. Our results reveal that piscivores distribution is already
restricted in the present (average 8.84%), even the presence of this trophic guild (6.42%). Trophic guild dis­
tribution in the short-term under climate change may represent less than 6.5% across the Atlantic Forest, will be
contracting by − 30.0%. We conclude that combinations of factors inductors of biodiversity loss will be
empowered by climate changes, leading to more restricted species distribution. SDM is fundamental to under­
stand the species distribution to prioritize conservation efforts from local to pan-continental scales, being
paramount to slow down the consequences of climate change.

1. Introduction ca. 8.0% of the world's biodiversity (Myers et al., 2000) and is extremely
threatened due to chronic anthropogenic disturbances, such as defor­
The biodiversity across the tropics is high due to evolutionary history estation, fragmentation, urban expansion, defaunation, resources over­
imposed by a myriad of ecosystem types across a large variety of ecor­ exploitation, and climate changes (Bellard et al., 2014; Tabarelli et al.,
egions, which provide partitioned environments for the species devel­ 2010). These impacts affect negatively several group species, including
opment that tends to radiation avoiding competition (Carrillo et al., mammals (Bogoni et al., 2018; Galetti et al., 2013) and native fishes that
2014; Eisenberg, 1981). Despite this evolutionary history with cumu­ compose the prey base for piscivore vertebrates (Barletta et al., 2016;
lative speciation that increases biodiversity, currently, several taxa in Lucifora et al., 2017).
natural habitats worldwide are suffering human-induced species loss Among 321 mammal species distributed across the Atlantic Forest,
and population decline (Davidson et al., 2009; Garmendia et al., 2013). according to Graipel et al. (2017) (but see also Quintela et al. (2020) and
These modern population collapses are overwhelming across the Neo­ Abreu-Jr et al. (2021)) there are only six (ca. 2.0%) terrestrial, volant or
tropics, especially in the historically degraded Atlantic Forest biome semi-aquatic species that are considered piscivore: Noctilio albiventris
(Bogoni et al., 2018, 2020a; Navarro et al., 2021). The Atlantic Forest of Desmarest, 1818, Noctilio leporinus (Linnaeus, 1758), Chironectes mini­
South America (originally spanning over 1,310,299 km2) encompasses mus, (Zimmermann, 1780), Lutreolina crassicaudata (Desmarest, 1804),

* Corresponding authors at: Universidade de São Paulo (USP), Escola Superior de Agricultura “Luiz de Queiroz” (ESALQ), Departamento de Ciências Florestais,
Laboratório de Ecologia, Manejo e Conservação de Fauna Silvestre (LEMaC), Piracicaba, SP 13418-900, Brazil & Programa de Pós-Graduação em Ecologia, Uni­
versidade Federal de Santa Catarina, Florianópolis, SC, Brazil.
E-mail addresses: [email protected] (J.A. Bogoni), [email protected] (M.M. Tagliari).

https://doi.org/10.1016/j.ecoinf.2021.101357
Received 5 April 2021; Received in revised form 20 June 2021; Accepted 20 June 2021
Available online 25 June 2021
1574-9541/© 2021 Elsevier B.V. All rights reserved.
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

Lontra longicaudis (Olfers, 1818), and Pteronura brasiliensis (Gmelin, We hypothesized that: (i) the current distribution of each piscivore
1788). Among these piscivore mammals, L. longicaudis (NT) and species ranges from 5% to 25% of the Atlantic Forest territory,
P. brasiliensis (EN) are under threat according to the International Union congruently to the IUCN distribution polygons; (ii) the potential distri­
for Conservation of Nature (IUCN) Red List (IUCN, 2021). Despite the bution based on the realized climatic niche of the trophic guild (i.e. the
P. brasiliensis was listed according to Paglia et al. (2012) for the Atlantic averaged presence of at least one of the six extant piscivore species of the
Forest, this species is virtually extinct for the entire biome (Bogoni et al., Atlantic Forest) currently encompasses an area larger than 10% of the
2018) and practically not represented in the Atlantic Forest according to original Atlantic Forest territory; (iii) under future short-term climate
IUCN distribution maps (IUCN, 2021). Thus, this trophic guild might be change scenarios the distribution of species and consequently the dis­
the rarest, extinction-proneness and a species-poor group among the tribution of the trophic guild will be reduced from 30% to 40%,
entire mammalian biota in the Neotropics and consequently across the restricted to areas with higher altitudes, extreme-south of latitude, or
Atlantic Forest. coastal zones (due to the major precipitation in future climate change
These aforementioned piscivore mammals are characterized by scenarios); and (iv) prioritization areas coincide with the trophic guild
feeding on freshwater fishes across the Neotropical ecosystems (Hood distribution hotspots but also with the most heavily modified landscapes
and Jones, 1984; Hood and Pitocchelli, 1983; Larivière, 1999; Marshall, across the Atlantic Forest biome.
1978a, 1978b). Noctilio leporinus also prey on fishes in mangroves and
marines ecosystems (Hood and Jones, 1984; Bordignon, 2006). The fish- 2. Material and methods
based diet among these species varies largely according to their body
size and digestive systems (Cabral et al., 2010; Cáceres et al., 2002; 2.1. Data on species occurrence and climatic niche
Sousa et al., 2012). Despite highly adherent to a fish-based diet, these
piscivore mammals also feed on other alimentary items (e.g. Gastro­ We obtained georeferenced data for the six piscivore mammal
poda, birds, small-bodied mammals, and invertebrates; Quadros and occurrence across the Neotropics based on the Global Biodiversity In­
Monteiro-Filho, 2001; Facure and Ramos, 2011; Sousa et al., 2012; formation Facility (GBIF; http://data.gbif.org). Moreover, we also used
Groenendijk et al., 2015). Even this mild diet flexibility can lead the georeferenced information derived from recent data papers of mammal
species to thrive both in environments with high diversity or seasonally distribution across the Atlantic Forest and the Neotropics (Bovendorp
stressing ecosystems (Dunlop et al., 2017; Spencer et al., 2017), once et al., 2017; Figueiredo et al., 2017; Lima et al., 2017; Muylaert et al.,
restricted diet and body size are underlying causes of high extinction risk 2017; Nagy-Reis et al., 2020) and species distribution data obtained
(Cardillo et al., 2005, 2008; Davidson et al., 2009). from Aguiar et al. (2019) and Bogoni et al. (2017, 2020a, unpublished
The species extinction risk depends on functional traits and may data). We also searched for the species alpha taxonomy to further
increase due to climate change (Pacifici et al., 2015, 2017). Several improve the occurrence points using the main search engine, such as
combined factors (e.g. defaunation, habitat loss, functional traits, and Web of Science, Scopus, and Google Scholar. Our initial dataset con­
climate change) have been promoting a higher decline in species pop­ tained 4657 occurrence points among the six piscivore species distrib­
ulation (Bogoni et al., 2016; Pacifici et al., 2017; Peres, 1990; Peres uted along the entire Neotropical realm (Supporting Information S1).
et al., 2016). Moreover, the prey-basis decline also influences the pop­ Despite virtually extinct from the Atlantic Forest biome (Bogoni et al.,
ulation dynamics of mammals (Spencer et al., 2017). For native fish 2020a), we do not discarded P. brasiliensis from our analysis due to our
species in the Atlantic Forest, hydroelectric damming, water pollution, prior modelling-scale approach. To do so, we performed the SDM for all
agricultural practices, and overfishing are some examples that cause species across the entire Neotropics extent (Fig. 1A). The produced SDM
high levels of threat and fast decline (Reis et al., 2003, 2016). Conse­ outputs generated for the Neotropics were a posteriori rearranged
quently, as fishes are the basis of the trophic web for piscivores mam­ within the Atlantic Forest extent. Consequently, we also were able to
mals, disturbances affecting both groups could promote cascade-effects identified P. brasiliensis climatic niche within the Atlantic Forest. To
and influence several other ecological processes (Bovo et al., 2021; model the “trophic guild” we combined all occurrence points within the
Sobral et al., 2017). Neotropics extent of the six extant piscivores species (N = 4657) and
Cascade effects also can be promoted by climate change since the classified these points as a single organism (i.e. the “trophic guild”;
species distribution can be strongly altered promoting potential new despite being considered an “artificial” biological entity, we aimed to
interactions among species (Dyer and Letorneau, 2013; Schmitz, 2003). reveal core areas for species sympatry). We thus removed duplicated
The climate niche provides theoretical assumptions that are responsible points within the Atlantic Forest for every species and the trophic guild.
for defining the species distribution in a perspective of the classic Finally, we only accounted for our modelling approach occurrence
fundamental and realized niche (Colwell and Rangel, 2009; Franklin, points with complete climatic information (i.e. the bioclimatic variables
2009; Hutchinson, 1957). Future projections had shown that many selected to perform the SDM). Modelling approaches require a minimum
species could have their distribution reduced severally, only remaining of three occurrence points to perform a satisfactory correlative niche
in refuge areas (e.g. Levinsky et al., 2007; Porto et al., 2012). Yet, other modelling (see van Proosdij et al., 2016). Consequently, we could
species — mainly invasive — may present an increase in their distri­ project satisfactorily the current and future climatic niche of our target
bution range (e.g. Ficetola et al., 2007). Species distribution models species and their trophic guild.
(SDMs) are an important tool to understand the species distribution
dynamics under future climate change scenarios (Elith et al., 2011; Elith 2.2. Current and future climatic data
and Leathwick, 2009). Climate change scenarios — fueled by anthropic
activities — can lead to an increase in temperature from 1.8 ◦ C to 4.8 ◦ C We collected current and future climatic conditions from WordClim
for the 2100 s (Harris et al., 2014; Parmesan, 2006). Global Climate Database version 2.1 (https://www.worldclim.org/da
Considering the practically irreversible scenarios of climate change ta/worldclim21.html). Current climatic conditions were generated by
increasing the threat to one of the most important tropical biodiversity interpolated climatic data from 1970 to 2000 (Fick and Hijmans, 2017).
hotspots worldwide — i.e., the Atlantic Forest of South America — Future climatic data were downloaded from the Coupled Model Inter­
under a realized climatic niche approach we aimed: (1) to understand comparison Project Phase 6 (CMIP6), using WorldClim 2.1 as the
the current distribution of the six (quasi)extant piscivore mammals and baseline climate between 2021 and 2040 (2030 average), under three
the potential distribution of this mammalian trophic guild across the distinct SSPs: SSP 126, SSP 245, and SSP 585. We selected SSPs because
Atlantic Forest; and (2) to predict the distributions under distinct their future projections encompass climatic data from Representative
climate change scenarios via Shared Socioeconomic Pathways (SSP; Concentration Pathways — commonly used in SDM approaches — with
Riahi et al., 2017) which were projected to the short-term (i.e., 2030). socioeconomic data, such as economic growth, population change,

2
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

A B

N Noctilio albiventris

Noctilio leporinus

Chironectes minimus

Lutreolina crassicaudata

Noctilio albiventris
Lontra longicaudis

Noctilio leporinus
Chironectes minimum
Lutreolina crassicaudata

Lontra longicaudis
Pteronura brasiliensis

Pteronura brasiliensis

Fig. 1. (A) Spatial distribution of six extant piscivore mammals post hoc filtered presence-points (N = 4657) across the Neotropical realm. The background map
(gray-scale only for Atlantic Forest) are the elevation gradient derived from ASTER-GDEM (https://asterweb.jpl.nasa.gov/gdem.asp) and the major Pan-American
rivers (blue) according to GR-UNESCO, World-wide Hydrogeological Mapping and Assessment Programme (WHYMAP: www.whymap.org). (B) Distribution
ranges of species according to IUCN (2021) polygons, highlighting the Atlantic Forest biome. Blue lines represents the major rivers according to WHYMAP. (For
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

urbanization, and air pollution (Riahi et al., 2017). We created a 2.3. Species Distribution Modelling: calibrating, forecasting, validating,
framework using different SSPs: (i) optimistic scenario with low chal­ and zonation classification
lenges to mitigation and adaptation to climate change (SSP 126); (ii) a
realistic scenario according to middle challenges to mitigation and To perform our Species Distribution Modelling (SDM) and to assess
adaptation (SSP 245), and (iii) a “fossil-fueled development” scenario the realized climatic niche (sensu Hutchinson, 1957; but see Wiens et al.,
(SSP 585), assuming high challenges to mitigation and adaptation (Riahi 2009) of the piscivore mammals within the Atlantic Forest and their
et al., 2017). The WorldClim downloaded data consists of the standard trophic guild, we combined four statistical algorithms: random forests
19 bioclimatic variables, generated from monthly temperature and (RF); artificial neural network (ANN); generalized linear models (GLM);
rainfall values. Both current and future data were downloaded at 2.5 and generalized additive model (GAM). We aimed to perform a gradient
arc-minutes (grids of 5 km2) and computed over the Neotropical realm from robust algorithms (GAM and GLM) to complex algorithms (ANN
extent (Fig. 1). and RF), following Elith and Graham (2009). We followed the ensemble
We selected three Global Circulation Model (GCM) to simulate future modelling approach (Araújo and New, 2007), as a strategy to produce a
climate in 2030: Centre National de Recherches Météorologique — CNRM- consensus among several outputs produced by the SDM. Due to the
CM6–1; the Beijing Climate Center Climate System Model — BBC-CSM2- combination of two regression models (GAM and GLM) and another two
MR; and the Model for Interdisciplinary Research on Climate — MIROC6 machine learning approaches (i.e. RF and ANN) in the SDM, it is advised
(all available at https://www.worldclim.org/data/cmip6/cmip6_clim the inclusion of pseudo-absences to obtain more accurate results
2.5m.html). As we aimed in this study to describe the realized climatic (Barbet-Massin et al., 2012). Consequently, we added 10,000 pseudo-
niche of the six extant piscivore mammals across the Atlantic Forest, as absences throughout the study area. We split our new presence/
well as their trophic guild, we selected four bioclimatic variables that pseudo-absence dataset using 70% for model calibration and 30% for
usually encompass the majority of 19 Worldclim variables (Fick and model validation and repeated a cross-validation procedure 5 times.
Hijmans, 2017): (1) annual mean temperature (◦ C); (2) mean diurnal Data split for calibration and validation significantly reduces sampling
range (mean of monthly [maximal temperature – minimum tempera­ bias as most of the data (70%) were used for calibration/fitting, whereas
ture]); (3) temperature seasonality (standard deviation × 100); and it also reduces variance as another 30% of the data were used to validate
annual precipitation (mm.y− 1). We also selected (4) altitude variable our models. This simple step leaning on calibrated and validated data
despite its potential correlation with temperature, as we aimed to assess seeks to add the efficiency of the modelling approach and avoid sam­
species redistribution to higher or lower elevations, and how this vari­ pling bias (Crisci et al., 2012). The performance to predict species'
able is important for their potential range changes. presence-absence and the ensemble modelling were evaluated via three

3
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

co-related metrics: True Skills Statistics (TSS); Sensitivity, or the prob­ on a maximum of 2693 (“trophic guild” case) points (see Supporting
ability of properly predicting the species presences; and Specificity, Information S1 for all species). Still, we have satisfactory projection
which indicates the probability of correcting predicting the species maps that represent their distributions according to IUCN maps
absence (Liu et al., 2011). The usage of these metrics is largely employed (Fig. 1B), and we were backed by satisfactorily accurate predictions
in SDM methods due to their easy-to-measure methods (Araújo et al., (Supporting Information S2). Species distribution currently can cover on
2019) and they complement each other (TSS = Sensitivity + Specificity average 8.84% (±6.2 sd; range 2.8% to 20.2%) of the Atlantic Forest
− 1; Allouche et al., 2006; Lawson et al., 2014; Tagliari et al., in review). biome. The species with the most prolific current distribution was L.
We set a quality threshold of 0.5 to evaluate algorithms prediction ca­ longicaudis, reaching 264,590 km2, whereas P. brasiliensis could be
pacity for the ensemble modelling according to the TSS metric, a potentially distributed across 37,335 km2 of the Atlantic Forest,
threshold-dependent parameter (Liu et al., 2011). Models where pre­ following the coastline (Fig. 2; Table 1). Our model of the potential
dictions were initially TSS < 0.5 were removed prior to ensemble distribution of the “trophic guild” reached currently 84,180 km2, mainly
modelling. Models where TSS ≥ 0.5 were selected for the binary trans­ concentrated in two distinct regions: at the coastline of the Atlantic
formation to the committee averaging step and the ensemble forecasting Forest and the western portion of interior plateaus of the Atlantic Forest
(Thuiller et al., 2009). (Fig. 2). Both piscivores bat species had current distribution from
For the committee averaging we selected a minimum of two out of 50,000 km2 to 105,000 km2 across the Atlantic Forest, presenting fail­
the four algorithms that indicated the species presence for current dis­ ures across the subtropical portion of the biome, especially adapted to
tribution (i.e. at least 50% of algorithms predicted the species occur­ lower elevations when compared with the other non-volant piscivore
rence). Thus, over these suitable areas, we calculated the area within the mammals (Fig. 2; Table 1). Chironectes minimus have its current distri­
Atlantic Forest (in km2) with complete climatic/environmental infor­ bution virtually cumpled at the southeastern coastline. Differently,
mation available. Within these areas, we also randomly sampled 1000 L. crassicaudata showed potential current distribution at the interior
points to collect data to describe the species' realized climatic niche (see plateaus, especially the Brazilian southern plateaus (Fig. 2), reaching
Tagliari et al., in review; Vieilledent et al., 2013). To predict species 141,835 km2 (Table 1).
future distribution we followed the ensemble forecasting approach Based on the Limited Dispersion (LD) scenario, the species most
under committee averaging (Araújo and New, 2007) by combining our penalized due to the projection of climate change was P. brasiliensis
four previously selected algorithms with the tree Global Circulation which potential climatic niche might decrease by ≌–14.5% (only 31,995
Models mentioned above. We thus obtained for each species, as well for km2 of climatically suitable area), followed by L. longicaudis, decreasing
the whole trophic guild, under each SSP for 2030, 12 combined maps its original area by − 7.5% at the worst-case and realistic scenarios (i.e.,
that indicate the trophic guild and the five species' future suitable SSP-585 and SSP 245, respectively). In addition, the potential trophic
occurrence areas. The projected occurrence in 2030 was set by defining guild is expected to reduce its climatically suitable areas by − 31.8% (i.e.
at least 6 out of 12 models (i.e. minimum 50%) that correctly predicted a 26,735 km2) under the worst-case scenario (Table 1; Fig. 2). Our models
climatically suitable area. We also calculated the area (i.e. in km2) over under the Non-Limited Dispersion (NLD) approach showed that some
these climatically suitable occurrence areas in the future. Finally, aiming piscivore species could increase their current distribution, depending on
to perform a more “realistic” correlative SDM, we added two dispersal the climate change scenario (see Table 1). For instance, the bat
scenarios to future forecasting: the Limited-Dispersion (LD) and Non- N. leporinus might increase their climatic range by +63.4% (i.e. 66,575
Limited Dispersion (NLD). On one hand, under the LD scenario, spe­ km2), considering milder climate change scenarios (e.g. SSP-245;
cies future distribution cannot surpass the area calculated for the current Table 1). The “trophic guild”, even under all Non-Limited Dispersion
baseline. On the other hand, in the NLD scenario, we consider the spe­ scenarios, indicated a potential expansion-contraction of the climatic
cies' capacity to expand their climatic niche beyond their predicted niche from +0.2% to − 6.63% compared to the current prediction of
occurrence area in the present (Tagliari et al., in review; Vieilledent 84,180 km2. Each SDM performed for piscivore mammal species across
et al., 2013). the Neotropics indicate — beyond the Atlantic Forest — parts of the
We aimed to identify a rank of priority areas for conservation using Amazon forest, Pantanal wetlands, Cerrado and Meso-America as hot­
the spatial planning tool Zonation (version 4.0; Moilanen et al., 2014). spot of the species distribution (except for L. crassicaudata concentrated
We selected the SDM outputs within the Atlantic Forest produced for the majoritarian at Southern Brazil, Uruguay and Northern Argentina; these
current and the worst-case climate change SSP in 2030 (SSP 585) under 56 files are available upon request).
both LD and NLD dispersal scenarios. We thus defined a ranking of high Due to potential redistribution of the climatic niche, we identified an
priority areas for the merged maps of the six piscivore mammals of the upslope range shift for N. neporinus (from 276 m up to 396 m), C. min­
study. These priority areas were generated by iteratively removing cells imus (from 449 m up to 555 m), and L. longicaudis (531 m up to 548 m).
with low species potential occurrence (Spiers et al., 2018). We selected Other species are expected to remain stable in their elevational gradient
the ‘core area zonation’ algorithm with the ‘distribution smoothing’ option. compared to current predictions, such as N. albiventris (from 140 m to an
We did not select the ‘habitat condition’ and ‘boundary layer’ options interval ranging from 128 to 139 m, according to climatic scenarios) and
because we modeled species' climatic niche, aiming to identify potential L. crassicaudata (from 557 m to 553–558 m). Elevational range for
occurrence areas within the Atlantic Forest, disregarding potential P. brasiliensis is expected to decrease up to 11.4% (101 m) under climate
geographic limitation for the species' occurrences. All modelling ap­ change when compared to its current elevational gradient of 114 m. The
proaches, data collection, and figure creation (besides Zonation outputs) trophic guild showed a higher elevation gradient (i.e. 501 m) which is
were generated via R (R Core Team, 2020) using the package biomod2 expected to remain stable (maximum elevation upward change is +6.9%
(Thuiller et al., 2009). Others relevant R packages selected were rgdal compared to current mean elevation). Curiously, the trophic guild in the
(Bivand et al., 2020); raster (Hijmans, 2020); viridis (Garnier, 2018); and present averages 501 m, similar to C. minimus (449 m), L. crassicaudata
sp (Pebesma and Bivand, 2005). The full reproducible code is available (557 m), and L. longicaudis (531 m), presumably influenced by the
upon request. dominance of presence points of L. longicaudis. However, both
piscivores-bats are adapted, on average, at lower elevations, i.e. 140 m
3. Results for N. albiventris and 276 m for N. leporinus. In 2030, due to a downward
re-distributional range, L. crassicaudata might overlap with Lontra
Initially, we gathered 4657 presence points distributed among the six longicaudis and C. minimus bioclimatic niche and trophic guild (see
extant piscivore mammals across the Neotropical realm (Fig. 1A). Table 1), whilst piscivore-bats N. albiventris and P. brasiliensis could co-
However, according to our presence-points filtering with complete occur in similar elevational gradient in short-term climate change.
environmental data, we projected the species distribution models based As a consequence of the identification of the current distribution of

4
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

Fig. 2. Current and future (2030s, mild SSP-126 and worst-case SSP-585 under limited- dispersal scenario) species distribution models for the six extant piscivore
mammals and the potential trophic guild presence (i.e., at least one species) across the Atlantic Forest of South America.

5
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

Table 1
Area calculated in the present and under short-term climate change in 2030 (according to three different SSPs and two distinct dispersal scenarios: Limited Dispersion
and Non-Limited Dispersion).
Species Area current Current SSP Dispersal scenarios: Limited (LD) Area future % of area Future elevation (m) and % of
(km2) elevation (m) Scenario and Non-Limited (NLD) (km2) change elevation change

Noctilio albiventris 57,560 140 126 LD 54,905 − 4.6 135 | –3.5%

126 NLD 58,140 +1.0 139 | –1.0%
245 LD 54,220 − 5.8 133 | –5.0%
245 NLD 58,060 +0.9 138 | –1.4%
585 LD 52,955 − 8.0 128 | –8.5%
585 NLD 56,255 − 2.3 132 | –5.7%
Noctilio leporinus 104,995 276 126 LD 104,995 0.0 276 | 0.0%
126 NLD 168,585 +60.6 371 | +34.4%
245 LD 104,995 0.0 276 | 0.0%
245 NLD 171,570 +63.4 377 | +36.5%
585 LD 104,995 0.0 276 | 0.0%
585 NLD 181,105 +72.5 396 | 43.5%
Chironectes 89,035 449 126 LD 85,575 − 3.9 549 | +22.2%
minimus 126 NLD 122,015 +37.0 555 | +23.6%
245 LD 84,930 − 4.6 462 | +2.9%
245 NLD 113,945 +28.0 530 | +18.0%
585 LD 81,630 − 8.3 464 | +3.3%
585 NLD 113,885 +27.9 545 | +21.3%
Lutreolina 141,835 557 126 LD 135,115 − 4.7 554 | –0.5%
crassicaudata 126 NLD 136,960 − 3.4 554 | –0.5%
245 LD 136,250 − 3.9 558 | +0.1%
245 NLD 137,230 − 3.2 558 | –0.1%
585 LD 125,505 − 11.5 553 | –0.7%
585 NLD 126,080 − 11.1 553 | –0.7%
Lontra longicaudis 264,590 531 126 LD 246,455 − 6.9 544 | +2.4%
126 NLD 247,475 − 6.5 546 | +2.8%
245 LD 247,235 − 6.6 544 | +2.4%
245 NLD 248,195 − 6.2 546 | +2.8%
585 LD 243,200 − 8.1 546 | +2.8%
585 NLD 244,240 − 7.7 548 | +3.2%
Pteronura 37,335 114 126 LD 32,895 − 11.9 105 | –7.9%
brasiliensis 126 NLD 34,760 − 6.9 112 | –1.7%
245 LD 32,040 − 14.2 103 | –9.6%
245 NLD 33,570 − 10.1 108 | –5.2%
585 LD 30,930 − 17.2 101 | –11.4%
585 NLD 32,030 − 14.2 104 | –8.7%
“Trophic guild” 84,180 501 126 LD 60,015 − 28.7 523 | +4.4%
126 NLD 84,350 +0.2 535 | +6.7%
245 LD 59,500 − 29.3 511 | +1.9%
245 NLD 78,605 − 6.6 514 | +2.6%
585 LD 57,445 − 31.8 511 | +1.9%
585 NLD 84,420 +0.3 536 | +6.9%

These areas were generated firstly to the entire Neotropical realm, then we cropped for the Atlantic Forest area to calculate the percentages of change. We indicated the
percentage of area (in km2) and elevation (in m) change in the distribution of the six extant piscivores mammals and for the potential persistence of the trophic guild
across the Atlantic Forest of South America.

the six extant piscivore mammals of the Atlantic Forest, we identified Forest; (4) Bahia interior forests, and (5) Pernambuco Endemism Cen­
their current bioclimatic niche. The average annual mean temperature ter (Fig. 4), being similar between the current and future distribution of
of all six species ranges from 16 ◦ C to 28 ◦ C (Fig. 3). The potential tro­ piscivore species across the Atlantic Forest. The prioritization by
phic guild averages 19.6 ◦ C. Temperature seasonality is similar among merging each species along the scenarios (i.e., current and future)
lower-elevation adapted species (N. albiventris, N. leporinus, and coincide with the result presented in Fig. 4 and can be accessed in the
P. brasiliensis: 1.7 to 2.0 ◦ C, whereas C. minimuns, L. longicaudis, and Supporting Information S3–S8.
L. crassicaudata showed average temperature seasonality from 2.5 to
3.1 ◦ C, which might indicates a correlation of temperature seasonality 4. Discussion
and elevation. Mean diurnal range showed variation between 8.0 to
~12.0 ◦ C, which is a similar climatic niche among all species. Annual Understanding species distribution in space and time is a former
mean precipitation is also lower (but still similar) for both bat species challenge in ecology (Franklin, 2009; Hutchinson, 1957). In this study,
(1383.9 and 1279.2 mm.y− 1N. albiventris and N. leporinus, respectively we showed that piscivore mammal distribution across the Atlantic
and 1371.5 2 mm.y− 1 for P. brasiliensis). The other three piscivore Forest is already restricted in the present (from 2.8% to 20.2%; average
mammals: C. minimus, L. longicaudis, and L. crassicaudata showed 8.84; ±6.2 sd), even the potential presence of this trophic guild, repre­
respectively 1526.3, 1563.09, and 1456.8 2 mm.y− 1; Fig. 3). Our pri­ sented for at least one of any six species presence. Considering that re­
oritization approach — by merging a posteriori each species model in a mains only 12–15% of the original Atlantic Forest, the area of 8.84%
single portfolio for each climate scenario — revealed that the areas that weighted by the vegetal remnants represents only 17,374.6 km2 (1.4%)
should be prioritized for trophic guild conservation coincides with our of suitable habitat across the entire biome. The potential distribution of
prior modelling of “trophic guild”. These areas (see hot-colors in Fig. 4) the trophic guild in the short-term under climate change may contract
are concentrated in five major regions: (1) southern interior plateaus, by − 30.0% (58,987 km2) on average across the Atlantic Forest original
(2) “Serra do Mar” kniferidges, (3) western transition Cerrado-Atlantic territory, contracting by − 31.2% under the Limited Dispersion scenario

6
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

Fig. 3. Density plots indicating the climatic niche of the six piscivores mammals and their trophic guild across the Atlantic Forest biome.

and SSP 585 in comparison with the current distribution. From mild to are associated with a lower ability to exploit alternative resources, adapt
worst-case scenarios of climate change (i.e. from SSP 126 to SSP 585), to new environmental conditions and selective forces (Pacifici et al.,
the species' bioclimatic niche contraction ranged from 0.0% to − 17.0%, 2017) and are highly passive to temporal pulses of food availability
constraining the distributions coincidences mainly to coastal knifer­ (Bogoni et al., 2020b).
idges. Specialized diets (e.g. herbivores, piscivores, and carnivores) in The species with the largest decline in territorial distribution were
mammals are associated with a greater probability of negative responses P. brasiliensis (− 11.9% at SSP-126; and − 17.2% at SSP-585),
to climate changes (Pacifici et al., 2017). Our results are in agreement L. crassicaudata (− 4.7%; − 11.5%, respectively) and L. longicaudis
with prior studies on the influences of climate and habitat change upon (− 6.9%; − 8.1%). Lutreolina crassicaudata and C. minimus beyond the
functional groups or guilds (see Navarro et al., 2021; Pacifici et al., decline have a partially narrowed distribution area from current to
2017). Although the six extant piscivore mammal species of the Atlantic short-term climate change scenarios. The presumable rarity of the
Forest have a spectrum of prey that include other groups beyond fish, C. minimus indicated by its reduced distribution area may be due to — at
fishes presumably have the greatest cost-benefit for these species at the least in part — their secretive nocturnal habits and the inaccessibility of
optimal foraging theory viewpoint (MacArthur and Pianka, 1966), given their habitat (Walker, 1968). Chironectes minimus is intimately associ­
that is a high-energetic food. Thus, species with narrower food extents ated with freshwater streams and lakes, and in some areas, its

7
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

Fig. 4. Post hoc prioritization areas for piscivore mammal species under climate change scenarios across the Atlantic Forest biome. Where: (1) southern interior
plateaus, (2) “Serra do Mar” kniferidges, (3) western transition Cerrado-Atlantic Forest; (4) Bahia interior forests, and (5) Pernambuco Endemism Center.

individuals are found at considerable elevations (>1500 m above sea Associated with the decline of suitable areas for piscivore mammals
level) along mountain rivers (Walker, 1968; Marshall, 1978a, 1978b). occurrence, is presumable the decline of their prey. Empirical evidence
These environments are common in remnants of the Atlantic Forest so far indicates that fishes will be decreased under current and future
biome (Ribeiro et al., 2009), but generally, these species refugia tend to threats (e.g. Reis et al., 2003, 2016). Thus, future studies might seek to
be subsampled. Such as C. minimus, L. crassicaudata presents a disjunc­ identify the main fish species preyed on by these mammals to perform a
tive and fragmented distribution in South America, presenting similar co-occurrence modelling (e.g. Pollock et al., 2014). Other great groups
habits and habitats of C. minimus (Graipel et al., 2001) and thus, con­ that are part of the diet of these species are the insects. Despite insects
fronting the same restrictions. The only truly extant large-bodied (i.e. have the greatest terrestrial animal biomass, more than 75% percent
adult body size >1 kg) piscivore mammal across the Atlantic Forest — declined over 27-years (Hallmann et al., 2017). Loss of insects is certain
L. longicaudis — is a versatile species (Larivière, 1999), notwithstanding to have adverse effects on ecosystem functioning, as insects play a
could be a distribution shrinkage until 2030 by 10% of its current dis­ central role in providing a food source for higher trophic levels such as
tribution. Thus, the local disappearance of L. longicaudis can interrupt birds, mammals, and amphibians (Hallmann et al., 2017).
some ecological processes — once also performed by P. brasiliensis —, Our results, which showed a reduced species distribution already in
such as the top-down control of medium- to large-bodied fishes (mainly the current scenario of climatic niche, could be associated with the few
Cichlidae, Anostomidae, Characidae, and Pimelodidae; see Passamani presence data within the Atlantic Forest, even we performed the models
and Camargo, 1995; Spinola and Vaughan, 1995), crustaceans, mol­ using data from the entire Neotropical realm. Recent data papers listed
lusks, and other homeotherms and heterotherms vertebrates, once their these six species across the Atlantic Forest in low abundance in few lo­
diet partially can overlaps. cations, in comparison with other mammalian species (see Bovendorp
This ecologically restrict mammalian trophic guild can be also clas­ et al., 2017; Figueiredo et al., 2017; Lima et al., 2017; Muylaert et al.,
sified as a functional group considering their ecological function of top- 2017). Further, the large-bodied species (i.e. L. longicaudis and
down control of freshwater fish (Blondel, 2003). The Neotropical realm P. brasiliensis) represents 1.99% and 1.21% (respectively) of mammalian
currently is experiencing the phenomena of empty rivers (Antunes et al., carnivorans records in a recent “tour de force” of data compilation across
2016). The decline of vertebrates across the Neotropical rivers (i.e. the Neotropics (see Nagy-Reis et al., 2020). For Brazil, this issue (i.e.,
aquatic and semi-aquatic mammals) was similar to the decline of low species records) is common. For example, 71% of Brazilian Pro­
terrestrial species, thus the emptying of Neotropics is generalized from tected Areas — the territories typically that embraces the largest sam­
both aquatic and terrestrial ecosystems (Antunes et al., 2016; Bogoni pling effort — have less than 0.01 species records per km2 (Oliveira
et al., 2020a). Our result on the distribution of the trophic guild in­ et al., 2017) and biodiversity knowledge shortfalls are induced by
dicates that this ecological function can be threatened, once more than sampling bias, such as access and proximity of economically affluent
30% of the current potential distribution could be contracted. There is regions (Pereira et al., 2021).
an intraguild variation on prey size, once these species feed on small- to Our models reveal that the virtual absence of some piscivore mam­
large-bodied fishes, but all top-down pressure will be relaxed with the mals (e.g. Noctilio spp., C. minimus and P. brasiliensis) was concentered in
contraction of species distribution and their potential sympatry. How­ the historically most degraded areas of the Atlantic Forest, such as
ever, according to Wilman et al. (2014) at least 16 other Neotropical interior plateaus and northeast portion. Thus, the habitat loss was
mammalian species, such as Panthera onca, Leopardus pardalis, Procyon intrinsically embodied in the models, once these species are highly
cancrivorus, and Nectomys spp. can feed on fish, composing at least affected by habitat change (Bogoni et al., 2020a; Püttker et al., 2020;
10–30% of their diet basis and performing the ecological function. Voigt and Kingston, 2016) decreasing their records on high-modified
Under future climate changes across the Atlantic Forest, this ecological regions. Therefore, it is paramount to understand the habitat availabil­
function probably will be agglutinated in other terrestrial vertebrates, ity across these species distribution ranges based on realized climatic
mainly birds that numerically embody the acme of this trophic guild niche, which surely is more restrict once the optimum of species dis­
worldwide, but are more restricted to coastal zones (Marquiss and Carss, tribution coincides with densely-settled areas, agricultural landscapes,
1994; Wilman et al., 2014). and/or high-defaunated mesoregions. Our prioritization approach

8
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

reveals that conservation priority areas for the trophic guild conserva­ spatiotemporal changes have occurred concerning the sampling effort
tion coincide with densely-settled regions or areas historically devoted and studies concentration conducted with small-bodied mammals across
to agriculture, but were also concentered in regions with high biodi­ the tropics (e.g. Figueiredo et al., 2017; Pereira et al., 2021), but still
versity profusion. Conservation action to piscivore mammals would insufficient or biased to answer some specific ecological questions, such
serve as a proxy to ensure the thriving of all sub-regional biotas. We as the abiotic tolerance of species, especially for the elusive marsupials
should prioritize the most suitable regions for the trophic guild but C. minimus and L. crassicaudata. An agenda that encompasses the
without entirely forgetting sizeable populations across the regions knowledge gaps (e.g. Linnean, Wallacean, and Hutchinsonian; see
where the threats, at least for now, are still manageable. Therefore, this Hortal et al., 2015) can be gradually hewed by the slow-drip of publi­
approach is important for short-term decisions, such as conservation cations (Lees et al., 2020), including alpha taxonomy (e.g. Helgen et al.,
priorities (Spiers et al., 2018) and public health actions (Amado et al., 2013), natural history (e.g. Amorin et al., 2020), community dynamics
2021). Expanding our viewpoint, we also identified that parts of the and trophic structure (e.g. Magioli et al., 2019), and data-papers on
Amazon forest, Pantanal wetlands, Cerrado, and Meso-America can be species distributions (e.g. Dornelas et al., 2018). Other results emerge
considered as hotspots of the species distribution and therefore targets from our approach, such as the broad limitation of piscivore bats to the
for conservation efforts. low-elevation gradient. Understand the main geoecological causes that
Under the methodological viewpoint, it is established that pseudo- govern the macroecology of bats across Neotropical realm scales re­
absences points could increase the predictive accuracy of the species mains hitherto a challenge. In accordance, our results reveal that there
distribution models (Phillips and Dudík, 2008). We added 10,000 are different compositions of climatic variables among the species. For
pseudo-absences points to better implement our ensemble modelling (i. instance, both bats are distributed in the hottest areas and fewer sea­
e. due to the different algorithms selected; Barbet-Massin et al., 2012). sonal environments than both marsupials and L. longicaudis.
This could affect species prevalence, representing the ratio between We also perform models for the potential distribution of P. brasiliensis
presence and absence data used to fit the model (Barbet-Massin et al., despite this species is virtually extinct in the Atlantic Forest, but listed in
2012) or the fraction of the study area occupied by the species the Atlantic Forest mammal checklist (see Graipel et al., 2017; Paglia
(McPherson et al., 2004). Thus, it is necessary the comparison between et al., 2012). In southern Brazil, the species is extinct since the early 20th
the several models produced by the ensemble modelling approach and century (Chehébar, 1990). In this study, we obtained only four presence
the k-fold repetition process using testing and training data, in order to points for this species across the Atlantic Forest. Pteronura brasiliensis is
reduce the problem imposed by sampling bias (Crisci et al., 2012; Hij­ the only mammal species in the study classified as “Endangered” by
mans, 2012). However, in our modelling approach, the rarest species (i. IUCN (IUCN, 2021), and perhaps this status should be updated due to
e. L. crassicaudata and C. minimus) were backed by elevated Specificity the constant expansion of human-induced environmental degradation.
(or the capacity to correct predict species absence) reinforcing the good Further, here we were interested in an overview of the realized climatic
predictive capacity of our modelling approach. Under the mild climate niche distribution, but future studies can encompass other environ­
change scenario (i.e., SSP-126) except of N. leporinus all piscivores mental predictors that surely will demonstrate a more limited distribu­
species can be decreased in their current restricted distribution from tion of this restricted trophic guild across the Atlantic Forest. Our results
− 3.9% to − 11.9%. The least affected species under this condition were reveals that no more than 3% of the Atlantic Forest are able to harbor the
both bats (Noctilio spp) and C. minimus. Under the worst-case climate P. brasiliensis.
change scenario (i.e. SSP-585) the restricted distribution of C. minimus Our prior hypotheses were partially confirmed, once the distribution
could decrease by − 8.3%. Among the piscivore species, N. leporinus and of species was typically lesser than 10% of the Atlantic Forest original
N. albiventris are species well adapted to fish-eating (Brooke, 1994; Hood territory but still congruent with the wide limits of IUCN distribution
and Jones, 1984; Hood and Pitocchelli, 1983). Both species inhabit range polygons. Future climate change will induce on average a
caves and hollow trees near lakes, rivers, or mangroves (in case of decrease in the bioclimatic niche of the studied group by ca. 9% (under
N. leporinus) or marshes (in case of N. albiventris) (Hood and Jones, 1984; Limited Dispersal scenario) comparing with the current distribution of
Hood and Pitocchelli, 1983). Thus, all these mammals naturally have a piscivore mammals and their trophic guild throughout the Atlantic
distribution restricted not in terms of extensions but in terms of Forest. The trophic guild currently only reaches 6.4% of the Atlantic
restricting habitats and naturally not-plastic habits, explaining the low Forest territory and will decrease on average ca. 30% under climate
records — and consequently current and future distribution — across change scenarios under Limited Dispersal scenario. Moreover, the pri­
the Atlantic Forest. oritization approach indicate that preeminent areas to conservation
Model evaluation indicated that TSS ≥ 0.4 is satisfactory (Zhang effort overlap both our approach of “trophic guild” modelling and
et al., 2015). As the majority of our ensemble modelling had values highly-degraded (but highly-biodiverse) regions across the Atlantic
where TSS ≥ 0.592 we might confirm that we were backed by acceptable Forest. From a conservation ecology viewpoint, our study allows a
projections. Furthermore, we selected a very restrictive mammal group, general depiction of piscivores potential distribution across the Atlantic
with a restrictive diet and specific distribution range, characteristic of Forest of South America and understanding that the low probabilities of
rare species. Rare species represent the vast majority of species and are species presence can be related to Wallacean (i.e. distribution), Pres­
identified in small samples (van Proosdij et al., 2016), as we presented tonian (i.e. abundance data), Hutchinsonian (i.e. abiotic tolerance) and
here. Yet, our results showed that the species probabilities of occurrence Eltonian (i.e. ecological interactions) shortfalls (Hortal et al., 2015).
were concentered mainly from 2.4% (P. brasiliensis) to 18.8% Moreover, according to current species distribution, the Amazon forest,
(L. longicaudis) in the Atlantic Forest, under short-term climate change in Cerrado, Pantanal and Meso-America may be a possible refuge area both
2030. Furthermore, the trophic guild encompasses no more than 5% in in the present (Bogoni et al., 2020a) and in the future for the piscivore
the more optimistic scenario (SSP 126) and might be even more terrestrial mammals of South America.
restricted by 2030 if we consider the worst-case scenario (SSP 585), We can conclude that combinations of factors (i.e. defaunation,
where only 4.38% of the entire Atlantic Forest extent could be suitable species ecological traits, loss and fragmentation of tropical forest) in­
for the piscivore mammal trophic guild. This means that even if the ductors of biodiversity loss will be empowered by climate change,
probability of presence is low (and spatial records data confirm this), the leading to an even more restricted species modern distribution across
Atlantic Forest can still count the species (in low probability) in almost the Atlantic Forest. Thus, our main insights highlight those conservation
all its original distribution territory and being congruent with the IUCN strategies — both for mammals and for tropical forest conservation —
range polygons of each species (see Fig. 1B). must focus on all these aforementioned combinations of factors. Public
Our results also evidence the urgent demand for new faunal in­ policies may target future refuge areas in order to enhance mammalian
ventories, especially small-bodied mammals. Over the past few decades, piscivores' conservation efforts, such as the creation of ecological

9
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

corridors between refuge areas and existing Protected Areas. Lastly, the E., 2016. Landscape features lead to shifts in communities of medium- to large-
bodied mammals in subtropical Atlantic Forest. J. Mammal. 97, 713–725.
species distribution model is an important tool to understand the species
Bogoni, J.A., Graipel, M.E., Oliveira-Santos, L.G.R., Cherem, J.J., Giehl, E.L.H.,
distribution to prioritize conservation efforts from local to pan- Peroni, N., 2017. What would be the diversity patterns of medium- to large-bodied
continental scales, being paramount to imposing a break on climate mammals if the fragmented Atlantic Forest was a large metacommunity? Biol.
change and other human-induced changes. Conserv. 211, 85–94.
Bogoni, J.A., Pires, J.S.R., Graipel, M.E., Peroni, N., Peres, C.A., 2018. Wish you were
here: how defaunated is the Atlantic Forest biome of its medium- to large bodied
Authors' contributions mammal fauna? PLoS One 13 (9), e0204515. https://doi.org/10.1371/journal.
pone.0204515.
Bogoni, J.A., Peres, C.A., Ferraz, K.M.P.M.B., 2020a. Extent, intensity and drivers of
Both authors contributed equally to this manuscript. mammal defaunation: a continental-scale analysis across the Neotropics. Sci. Rep. 10
https://doi.org/10.1038/s41598-020-72010-w, 10:14750.
Bogoni, J.A., Batista, G.O., Graipel, M.E., Peroni, N., 2020b. Good times, bad times:
Funding
resource pulses influence mammal diversity in meridional Brazilian highlands. Sci.
Total Environ. 734, 139473.
JAB and MMT were supported by CAPES-DS Ph.D. scholarships Bordignon, M.O., 2006. Diet of fishing bat Noctilio leporinus (Linnaeus) (Mammalia,
Chiropera) in a mangrove area of southern Brazil. Rev. Bras. Zool. 23, 256–260.
(201401069 and 201604174, respectively). JAB is supported by the São
Bovendorp, R.S., Villar, N., Abreu-Junior, E.F., Bello, C., Regolin, A.L., Percequillo, A.R.,
Paulo Research Foundation (FAPESP) postdoctoral fellowship grants Galetti, M., 2017. Atlantic small-mammal: a dataset of communities of rodents and
2018-05970-1 and 2019-11901-5. marsupials of the Atlantic Forests of South America. Ecology. 98 (8), 2226.
Bovo, A.A.A., Ferraz, K.M.P.M.B., Ribeiro, F., Lins, L.V., et al., 2021. Remaining suitable
areas for the critically endangered BrazilianMerganser (Mergus octosetaceus; Aves,
Declaration of Competing Interest Anseriformes) are threatenedby hydroelectric power plants. Perspect. Ecol. Conserv.
https://doi.org/10.1016/j.pecon.2021.04.002.
Brooke, A.P., 1994. Diet of fishing bat, Noctilio leporinus (Chiroptera: Noctilionidae).
The authors declare that they have no conflict of interest. J. Mammal. 75, 212–218.
Cabral, M.M.M., Zuanon, J., Mattos, G.E., Rosas, F.C.W., 2010. Feeding habits of giant
otters Pteronura brasiliensis (Carnivora: Mustelidae) in the Balbina hydroelectric
Acknowledgments reservoir, Central Brazilian Amazon. Zoologia. 27, 47–53.
Cáceres, N.C., Ghizoni-Jr, I.R., Graipel, M.E., 2002. Diet of two marsupials, Lutreolina
We acknowledge all the researchers who gathered and organized the crassicaudata and Micoureus demerarae, in a coastal Atlantic Forest island of Brazil.
Mammalia 66, 331–340.
datasets used in this study. We sincerely thank to two anonymous re­
Cardillo, M., Mace, G.M., Jones, K.E., Bielby, J., Bininda-Emonds, O.R.P., Sechrest, W.,
viewers for their important contribution to this manuscript. Orme, C.D.L., Purvis, A., 2005. Multiple causes of high extinction risk in large
mammal species. Science 309, 1239–1241.
Cardillo, M., Mace, G.M., Gittleman, J.L., Jones, K.E., Bielby, J., Purvis, A., 2008. The
Appendix A. Supplementary data predictability of extinction - biological and external correlates of decline in
mammals. Proc. R. Soc. B 275, 1441–1448.
Supplementary data to this article can be found online at https://doi. Carrillo, J.D., Forasiepi, A., Jaramillo, C., Sánchez-Villagra, M.R., 2014. Neotropical
mammal diversity and the Great American Biotic Interchange: spatial and temporal
org/10.1016/j.ecoinf.2021.101357.
variation in South America’s fossil record. Front. Genet. 5, 1–11. https://doi.org/
10.3389/fgene.2014.00451.
References Chehébar, C.E., 1990. Action Plan from Latin American Otters. In: Foster-Turley, P.,
MacDonald, S., Mason, C. (Eds.), Otters: An Action Plan for their Conservation. IUCN
Otter Spec Group Bull, pp. 64–73.
Abreu-Jr, E.F., Casali, D.M., Garbino, G.S.T., Libardi, G.S., Loretto, D., Loss, A.C.,
Colwell, R.K., Rangel, T.F., 2009. Hutchinson’s duality: the once and future niche. Proc.
Marmontel, M., Nascimento, M.C., Oliveira, M.L., Pavan, S.E., Tirelli, F.P., 2021.
Natl. Acad. Sci. 106, 19561–19568.
Lista de Mamíferos do Brasil, versão 2021-1 (Abril). Comitê de Taxonomia da
Crisci, C., Ghattas, B., Perera, G., 2012. A review of supervised machine learning
Sociedade Brasileira de Mastozoologia (CT-SBMz). Available at: https://www.sbmz.
algorithms and their applications to ecological data. Ecol. Model. 240, 113–122.
org/mamiferos-do-brasil/.
Davidson, A.D., Hamilton, M.J., Boyer, A.G., Brown, J.H., Ceballos, G., 2009. Multiple
Aguiar, L.M.S., Pereira, M.J.R., Zortéa, M., Machado, R.B., 2019. Where are the bats? An
ecological pathways to extinction in mammals. Proc. Natl. Acad. Sci. 106,
environmental complementarity analysis in a megadiverse country. Divers. Distrib.
10702–10705.
26, 1510–1522.
Dornelas, M., Antão, L.H., Moyes, F., Bates, A.E., Magurran, A.E., 2018. BioTIME: a
Allouche, O., Tsoar, A., Kadmon, R., 2006. Assessing the accuracy of species distribution
database of biodiversity time series for the Anthropocene. Glob. Ecol. Biogeogr. 27
models: prevalence, kappa and the true skill statistic (TSS). J. Appl. Ecol. 43,
(7), 760–786.
1223–1232. https://doi.org/10.1111/j.1365-2664.2006.01214.x.
Dunlop, J.A., Rayner, K., Doherty, T.S., 2017. Dietary flexibility in small carnivores: a
Amado, T.F., Moura, T.A., Riul, P., Lira, A.F.A., Badillo-Montaño, R., Martinez, P.A.,
case study on the endangered northern quoll, Dasyurus hallucatus. J. Mammal. 98,
2021. Vulnerable areas to accidents with scorpions in Brazil. Tropical Med. Int.
858–866.
Health. https://doi.org/10.1111/tmi.13561.
Dyer, L.A., Letorneau, D.K., 2013. Can climate change trigger massive diversity cascades
Amorin, F.W., Ballarini, C.S., Mariano, G., Lacerda-Bardosa, P.A., et al., 2020. Good
in terrestrial ecosystems? Diversity 5, 479–504.
heavens what animal can pollinate it? A fungus-like holoparasitic plant potentially
Eisenberg, J.F., 1981. The Mammalian Radiations: An Analysis of Trends in Evolution,
pollinated by opossums. Ecology 101 (5), e03001.
Adaptation, and Behaviour. The Athlone Press Ltd., London, UK.
Antunes, A.P., Fewster, R.M., Venticinque, E.M., Peres, C.A., Levi, F.R., Shepard, G.H.,
Elith, J., Graham, C.H., 2009. Do they? How do they? Why do they differ? On finding
2016. Empty forest or empty rivers? A century of commercial hunting in Amazonia.
reasons for differing performances of species distribution models. Ecography. 32,
Sci. Adv. 2, 1–15.
66–77.
Araújo, M.B., New, M., 2007. Ensemble forecasting of species distributions. Trends Ecol.
Elith, J., Leathwick, J.R., 2009. Species distribution models: ecological explanation and
Evol. 22 (1), 42–47. https://doi.org/10.1016/j.tree.2006.09.010.
prediction across space and time. Annu. Rev. Ecol. Evol. Syst. 40, 677–697.
Araújo, M.B., Anderson, R.P., Barbosa, A.M., Beale, C.M., Dormann, C.F., Early, R.,
Elith, J., Phillips, S.J., Hastie, T., Dudík, M., Chee, Y.E., Yates, C.J., 2011. A statistical
Garcia, R.A., Guisan, A., et al., 2019. Standards for distribution models in
explanation of MaxEnt for ecologist. Divers. Distrib. 17, 43–57.
biodiversity assessments. Sci. Adv. 5 (1), eaat4858. https://doi.org/10.1126/sciadv.
Facure, K.G., Ramos, V.N., 2011. Food habits of the thick-tailed opossum Lutreolina
aat4858.
crassicaudata (Didelphimorphia, Didelphidae) in two urban areas of southeastern
Barbet-Massin, M., Jiguet, F., Albert, C.H., Thuiller, W., 2012. Selecting pseudo-absences
Brazil. Mamm. Biol. 76, 234–236.
for species distribution models: how, where and how many? Methods Ecol. Evol. 3,
Ficetola, G.F., Thuiller, W., Miaud, C., 2007. Prediction and validation of the potential
327–338.
global distribution of a problematic alien invasive species: the American bullfrog.
Barletta, M., Cysneiros, F.J.A., Lima, A.R.A., 2016. Effects of dredging operations on the
Divers. Distrib. 13, 476–485.
demersal fish fauna of a South American tropical–subtropical transition estuary.
Fick, S.E., Hijmans, R.J., 2017. WorldClim 2: new 1-km spatial resolution climate
J. Fish Biol. 89, 890–920.
surfaces for global land areas. Int. J. Climatol. 37, 4302–4315. https://doi.org/
Bellard, C., Leclerc, C., Leroy, B., Bakkenes, M., Veloz, S., Thuiller, W., Courchamp, F.,
10.1002/joc.5086.
2014. Vulnerability of biodiversity hotspots to global change. Glob. Ecol. Biogeogr.
Figueiredo, M.S.L., Barros, C.S., Delciellos, A.C., Guerra, E.B., Cordeiro-Estrela, P.,
23, 1376–1386.
Kajin, M., Alvarez, M.R., Asfora, P.H., et al., 2017. Abundance of small mammals in
Bivand, R., Keitt, T., Rowlingson, B., 2020. rgdal: Bindings for the 'Geospatial' Data
the Atlantic Forest (ASMAF): a data set for analyzing tropical community patterns.
Abstraction Library. R package version 1.5-16. https://CRAN.R-project.org/p
Ecology. 98, 2981.
ackage=rgdal.
Franklin, J., 2009. Mapping Species Distributions: Spatial Inference and Prediction.
Blondel, J., 2003. Guilds or functional groups: does it matter? Oikos 100, 223–231.
Cambridge University Press, Cambridge.
Bogoni, J.A., Cherem, J.J., Giehl, E.L.H., Oliveira-Santos, L.G.R., Castilho, P.V.,
Picinatto-Filho, V., Fanticini, F.M., Tortato, M.A., Luiz, M.R., Rizzaro, R., Graipel, M.

10
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

Galetti, M., Guevara, R., Côrtes, M.C., Fadini, R., et al., 2013. Functional extinction of Oliveira, U., Soares-Filho, B.S., Paglia, A.P., Brescovit, A.D., et al., 2017. Biodiversity
birds drives rapid evolutionary changes in seed size. Science. 340, 1086–1090. conservation gaps in the Brazilian protected areas. Sci. Rep. 7, 9141. https://doi.
Garmendia, A., Arroyo-Rodriguez, V., Estrada, A., Naranjo, E.J., Stoner, K.E., 2013. org/10.1038/s41598-017-08707-2.
Landscape and patch attributes impacting medium- and large-sized terrestrial Pacifici, M., Foden, W.B., Visconti, P., Watson, J.E.M., et al., 2015. Assessing species
mammals in a fragmented rain forest. J. Trop. Ecol. 29, 331–344. vulnerability to climate change. Nat. Clim. Chang. 5, 215–225.
Garnier, S., 2018. viridis: Default Color Maps from 'matplotlib'. R package version 0.5.1. Pacifici, M., Visconti, P., Butchart, S.H.M., Watson, J.E.M., Cassola, F.M., Rondinini, C.,
https://CRAN.R-project.org/package=viridis. 2017. Species’ traits influenced their response to recent climate change. Nat. Clim.
Graipel, M.E., Cherem, J.J., Ximenez, A., 2001. Mamíferos terrestres não voadores da Chang. 7, 205–209.
Ilha de Santa Catarina, sul do Brasil. Biotemas 14, 109–140. Paglia, A.P., Fonseca, G.A.B., Rylands, A.B., Herrman, G., Aguiar, L.M.S., Chiarello, A.G.,
Graipel, M.E., Cherem, J.J., Monteiro-Filho, E.L.A., Carmignotto, A.P., 2017. Mamíferos Leite, Y.L.R., Costa, L.P., Siciliano, S., Kierulff, M.C.M., Mendes, S.L., Tavares, C.V.,
da Mata Atlântica. In: Monteiro Filho, E.L.D.A., Conte, C.E. (Eds.), Revisões em Mittermeier, R.A., Patton, J.L., 2012. Annotated checklist of Brazilian mammals.
Zoologia: Mata Atlântica. Editora UFPR, pp. 391–482. Occas. Pap. Conserv. Biol. 6, 1–76.
Groenendijk, J., Duplaix, N., Marmontel, M., Van Damme, P., Schenck, C., 2015. Parmesan, C., 2006. Ecological and evolutionary responses to recent climate change.
Pteronura brasiliensis. The IUCN Red List of Threatened Species 2015. e. Annu. Rev. Ecol. Evol. Syst. 37, 637–669.
T18711A21938411. https://doi.org/10.2305/IUCN.UK.2015-2.RLTS. Passamani, M., Camargo, L., 1995. Diet of the river otter Lutra longicaudis in Furnas
T18711A21938411.en. Downloaded on 18 May 2021. reservoir, South-Eastern Brazil. Int. Union Conserv. Otter Specialist Group Bull. 12,
Hallmann, C.A., Sorg, M., Jongejans, E., Siepel, H., Hofland, N., Schwan, H., 32–33.
Stenmans, W., Müller, A., Sumser, H., Hörren, T., Goulson, D., Kroon, H., 2017. More Pebesma, E.J., Bivand, R.S., 2005. Classes and methods for spatial data in R. R News 5
than 75 percent decline over 27 years in total flying insect biomass in protected (2). https://cran.r-project.org/doc/Rnews/.
areas. PLoS One 10, 1–21. Pereira, A.D., Bogoni, J.A., Siqueira, M.H., Bovendorp, R.S., Vidotto-Magnoni, A.P.,
Harris, R.M.B., Grose, M.R., Lee, G., Bindoff, N.L., Porfirio, L.L., Fox-Hughes, P., 2014. Orsi, M.L., 2021. Sampling biases of small non-volant mammals (Mammalia:
Climate projections for ecologists. Climate Change 5, 621–637. Rodentia and Didelphimorphia) surveys in Paraná state, Brazil. Stud. Neotropical
Helgen, K.M., Pinto, M., Kays, R., Helgen, L.E., Tsuchiya, M.T.N., Quinn, A., Wilson, D.E., Fauna Environ. https://doi.org/10.1080/01650521.2021.1891818.
Maldonado, J., 2013. Taxonomic revision of the olingos (Bassaricyon), with Peres, C.A., 1990. Effects of hunting on Western Amazonian primate communities. Biol.
description of a new species, the Olinguito. ZooKeys 324, 1–83. Conserv. 54, 47–59.
Hijmans, R.J., 2012. Cross-validation of species distribution models: removing spatial Peres, C.A., Emilio, T., Schietti, J., Desmoulière, S.J.M., Levi, T., 2016. Dispersal
sorting bias and calibration with a null model. Ecology 93, 679–688. limitation induces long-term biomass collapse in overhunted Amazonian forests.
Hijmans, R.J., 2020. raster: Geographic Data Analysis and Modeling. R package version Proc. Natl. Acad. Sci. 113 (4), 892–897.
3.3-13. https://CRAN.R-project.org/package=raster. Phillips, S.J., Dudík, M., 2008. Modeling of species distribution with Maxent: new
Hood, C.S., Jones, J.K., 1984. Noctilio leporinus. Mamm. Species 216, 1–7. extensions and a comprehensive evalutation. Ecograpy 31, 161–175. https://doi.
Hood, C.S., Pitocchelli, J., 1983. Noctilio albiventris. Mamm. Species 197, 1–5. org/10.1111/j.2007.0906-7590.05203.x.
Hortal, J., Bello, F., Diniz-Filho, J.A.F., Lewinsohn, T.M., Lobo, J.M., Ladle, R.J., 2015. Pollock, L.J., Tingley, R., Morris, W.K., Golding, N., O’Hara, R.B., Parris, K.M., Vesk, P.
Seven shortfalls that beset large-scale knowledge of biodiversity. Annu. Rev. Ecol. A., McCarthy, M.A., 2014. Understanding co-occurrence by modelling species
Evol. Syst. 46, 523–549. simultaneously with a Joint Species Distribution Model (JSDM). Methods Ecol. Evol.
Hutchinson, G.E., 1957. Concluding remarks. Cold Spring Harb. Symp. Quant. Biol. 22, 5, 397–406.
415–427. Porto, T.J., Carnaval, A.C., Rocha, P.L.B., 2012. Evaluating forest refugial models using
IUCN: International Union for Conservation of Nature, 2021. The IUCN Red List of species distribution models, model filling and inclusion: a case study with 14
Threatened Species 2021-1. http://www.iucnredlist.org/search. Brazilian species. Divers. Distrib. 19, 330–340.
Larivière, S., 1999. Lontra longicaudis. Mamm. Species 609, 1–5. Püttker, T., Crouzeilles, R., Almeida-Gomes, M., Schmoeller, M., et al., 2020. Indirect
Lawson, C.R., Hodgson, J.A., Wilson, R.J., Richards, S.A., 2014. Prevalence, thresholds effects of habitat loss via habitat fragmentation: a cross-taxa analysis of forest-
and the performance of presence-absence models. Methods Ecol. Evol. 5, 54–64. dependent species. Biol. Conserv. 241, 108368.
Lees, A.C., Rosenberg, K.V., Ruiz-Gutierrez, V., Marsden, S., Schulenberg, T.S., Quadros, J., Monteiro-Filho, E.L.A., 2001. Diet of the neotropical otter, Lontra
Rodewald, A.D., 2020. A roadmap to identifying and filling shortfalls in Neotropical longicaudis, in an Atlantic forest area, Santa Catarina State, Southern Brazil. Stud.
ornithology. Auk Ornithol. Adv. 137, 1–17. Neotropical Fauna Environ. 36, 15–21.
Levinsky, I., Skov, F., Svenning, J.-C., Rahbek, C., 2007. Potential impacts of climate Quintela, F., da Rosa, C.A., Feijó, A., 2020. Updated and annotated checklist of recent
change on the distributions and diversity patterns of European mammals. Biodivers. mammals from Brazil. An. Acad. Bras. Cienc. 92 (2) e20191004.
Conserv. 16, 3803–3816. R Core Team, 2020. R: A Language and Environment for Statistical Computing. R
Lima, F., Beca, G., Muylaert, R.L., Jekinns, C.N., et al., 2017. ATLANTIC-CAMTRAPS: a Foundation for Statistical Computing.
dataset of medium and large terrestrial mammal communities in the Atlantic Forest Reis, R.E., Kullander, S.O., Ferraris, C.J., 2003. Check List of the Freshwater Fishes of
of South America. Ecology 98 (11), 2979. South and Central America. Edipucrs, Porto Alegre.
Liu, C., White, M., Newell, G., 2011. Measuring and comparing the accuracy of species Reis, R.E., Albert, J.S., Dario, F., Mincarone, M.M., Petry, P., Rocha, L.A., 2016. Fish
distribution models with presence-absence data. Ecography. 34, 232–243. biodiversity and conservation in South America. J. Fish Biol. 89, 12–47.
Lucifora, L.O., Balboni, L., Scarabotti, P.A., Alonso, F.A., Sabadin, D.E., Solari, A., Riahi, K., Rao, S., Krey, V., Cho, C., Chirkov, V., Fischer, G., Kindermann, G.,
Vargas, F., Barbini, S.A., Mabragaña, E., Díaz, J.M.A., 2017. Decline or stability of Nakicenovic, N., Rafaj, P., 2017. RCP 8.5-A scenario of comparatively high
obligate freshwater elasmobranchs following high fishing pressure. Biol. Conserv. greenhouse gas emissions. Climate Change 109, 33–57.
210, 293–298. Ribeiro, M.C., Metzger, J.P., Martensen, A.C., Ponzoni, F.J., Hirota, M.M., 2009. The
MacArthur, R.H., Pianka, E.R., 1966. On optimal use of a patchy environment. Am. Nat. Brazilian Atlantic Forest: how much is left, and how is the remaining forest
100, 603–609. distributed? Implications for conservation. Biol. Conserv. 142, 1141–1153.
Magioli, M.M., Moreira, M.Z., Fonseca, R.C.B., et al., 2019. Human-modified landscapes Schmitz, O.J., 2003. Top predator control of plant biodiversity and productivity in an
alter mammal resource and habitat use and trophic structure. Proc. Natl. Acad. Sci. old-field ecosystem. Ecol. Lett. 6, 156–163.
116 (37), 18466–18472. Sobral, M., Silvius, K.M., Overman, H., Oliveira, L.F.B., Rabb, T.K., Fragoso, J.M.V.,
Marquiss, M., Carss, D.N., 1994. Avian Piscivores: Basis for Policy. National Rivers 2017. Mammal diversity influences the carbon cycle through trophic interactions in
Authority, Institute of Terrestrial Ecology, Bristol. the Amazon. Nat. Ecol. Evol. 1, 1670–1676.
Marshall, L.G., 1978a. Lutreolina crassicaudata. Mamm. Species 91, 1–4. Sousa, K.S., Bastazini, V.A.G., Colares, E.P., 2012. Feeding ecology of the Neotropical
Marshall, L.G., 1978b. Chironectes minimus. Mamm. Species 109, 1–6. otter Lontra longicaudis in the lower Arroio Grande River, southern Brazil. An. Acad.
McPherson, J.M., Jetez, W., Rogers, D.J., 2004. The effects of species’ range sizes on the Bras. Cienc. 85, 285–294.
accuracy of distribution models: ecological phenomenon or statistical artefact? Spencer, E.E., Newsome, T.M., Dickman, C.R., 2017. Prey selection and dietary flexibility
J. Appl. Ecol. 41, 811–823. of three species of mammalian predator during an irruption of non-cyclic prey.
Moilanen, A., Pouzols, F.M., Meller, L., Veach, V., Arponen, A., Lappänen, J., Kujala, H., R. Soc. Open Sci. 4, 1–11.
2014. Zonation - Spatial conservation planning and software. Version 4. User Spiers, J.A., Oatham, M.P., Rostant, L.V., Farrell, A.D., 2018. Applying species
Manual. C-BIG Conservation Biology, Informatics Group, Department of Biosciences, distribution modelling to improving conservation based decisions: a gap analysis of
University of Helsinki, Finland. Trinidad and Tobago’s endemic vascular plants. Biodivers. Conserv. 27, 2931–2949.
Muylaert, R.L., Stevens, R.D., Esbérard, C.E.L., Mello, M.A.R., et al., 2017. Atlantic bats: Spinola, R.M., Vaughan, C., 1995. Abundancia relativa y actividad de marcaje de la
a dataset of bat communities from the Atlantic Forests of South America. Ecology. 98 nutria neotropical (Lutra longicaudis) en Costa Rica. Vida Silvestre Neotropical. 4,
(12), 3227. 38–45.
Myers, N., Mittermeier, R.A., Mittermeier, C.G., Fonseca, G.A.B., Kent, J., 2000. Tabarelli, M., Aguair, A.V., Ribeiro, M.C., Metzger, J.P., Peres, C.A., 2010. Prospects for
Biodiversity hotspots for conservation priorities. Nature 403, 853–858. biodiversity conservation in the Atlantic Forest: lessons from aging human-modified
Nagy-Reis, M., Oshima, J.E.F., Kanda, C.Z., Lopes, F.B., et al., 2020. NEOTROPICAL landscapes. Biol. Conserv. 143, 2328–2340.
CARNIVORES: a data set on carnivore distribution in the Neotropics. Ecology 101 Tagliari M.M., Danthu P., Tsy J-M.L.P., Cornu C., Lenoir J., Carvalho-Rocha V.,
(11), e03128. https://doi.org/10.1002/ecy.3128. Vielledent G. Not all species will migrate poleward as the climate warms: the case of
Navarro, A.B., Magioli, M., Bogoni, J.A., Moreira, M.Z., Silveira, L.F., Alexandrino, E.R., the seven baobab species in Madagascar. In review. Global Change Biology.
Luz, D.T.A., Pizo, M.A., Silva, W.R., Oliveira, V.C., Donatelli, R.J., Christianini, A.V., Thuiller, W., Lafourcade, B., Engler, R., Araújo, M.B., 2009. BIOMOD: a platform for
Piratelli, A.J., Ferraz, K.M.P.M.B., 2021. Human-modified landscapes narrow the ensemble forecasting of species distributions. Ecography 32 (3), 369–373.
isotopic niche of neotropical birds. Oecologia. https://doi.org/10.1007/s00442-021- van Proosdij, A.S.J., Sosef, M.S.M., Wieringa, J.J., Raes, N., 2016. Minimum required
04908-9. number of specimen records to develop accurate species distribution models.
Ecography 39, 542–552.

11
J.A. Bogoni and M.M. Tagliari Ecological Informatics 64 (2021) 101357

Vieilledent, G., Cornu, C., Cuní-Sanchez, A., Pock-Tsy, J.M.L., Danthu, P., 2013. Wilman, H., Belmaker, J., Simpson, J., Rosa, C., Rivadeneira, M.M., Jetz, W., 2014. Elton
Vulnerability of baobab species to climate change and effectiveness of the protected Traits 1.0: species-level foraging attributes of the world’s birds and mammals.
area network in Madagascar: towards new conservation priorities. Biol. Conserv. Ecology 95 (7), 2027.
166, 11–22. Zhang, L., Liu, S., Sun, P., Wang, T., Wang, G., Zhang, X., Wang, L., 2015. Consensus
Voigt, C.C., Kingston, T., 2016. Bats in the Anthropocene: Conservation of Bats in a Forecasting of Species Distributions: The Effects of Niche Model Performance and
Changing World. Springer. Niche Properties. PLoS One 10, e0120056. https://doi.org/10.1371/journal.
Walker, E.P., 1968. Mammals of the World, 2nd ed. Johns Hopkins Press, Baltimore. pone.0120056.
Wiens, J.A., Stralberg, D., Jongsomjit, D., Howell, C.A., Snyder, M.A., 2009. Niches,
models, and climate change: assessing the assumptions and uncertainties. Proc. Natl.
Acad. Sci. 106 (2), 19729–19736.

12