ArticlePDF Available

Abstract and Figures

Aim: Elucidate the potential impacts of climate changes on the distribution and conservation of the multiple habitats of the Mata Atlântica biodiversity hotspot, which are often treated as a unique entity in ecological studies. Location: The whole extension of the South American Atlantic Forest Domain plus forest intrusions into the neighbouring Cerrado and Pampa Domains, which comprises rain forest ('core' habitat) and five environmentally marginal habitats, namely high elevation/latitude forest, rock outcrop habitats, riverine forest, semideciduous forest and restinga woodlands. Time period: Current (2000) and future scenarios (2050 and 2070). Major taxa studied: Tree species. Methods: We modelled the responses of 282 diagnostic tree species, using multiple algorithms and distinct scenarios of climate change (828,234 projections). Results: Potential loss of suitable environment summed 50.4% in semideciduous forest , 58.6% in riverine forest and 66% in rock outcrop habitats. Predictions for rain forest (12.2%), restinga woodlands (7.6%) and high elevation/latitude forest (5.2%) showed that overall loss of suitable environment will be relatively less severe for these habitats. Habitats that are confined to narrow edaphic conditions, namely rock outcrop habitats and riverine forest, are less studied and will likely suffer the greatest loss of biodiversity because their species are more dispersal limited. Main conclusions: Because these habitats occupy distinct environmental conditions, lumping them in ecological analyses might lead to erroneous interpretations in studies aiming to evaluate the impacts of global change in the Mata Atlântica biodiversity hotspot. This reinforces the importance of our approach and urges for conservation strategies that account for habitat heterogeneity in the Mata Atlântica and other species-rich environments.
This content is subject to copyright. Terms and conditions apply.
1846
|
Diversity and Distributions. 2019;25:1846–1856.
wileyonlinelibrary.com/journal/ddi
Received: 22 Augus t 2018 
|
  Revised: 14 June 2019 
|
  Accepted: 2 August 2019
DOI : 10.1111/ddi .12984
BIODIVERSITY RESEARCH
Habitat‐specific impacts of climate change in the Mata
Atlântica biodiversity hotspot
Luíz Fernando Esser1| Danilo M. Neves2,3 | João André Jarenkow1
This is an op en access arti cle under the ter ms of the Creative Commons Attribution L icense, which pe rmits use, dis tribu tion and reprod uction in any med ium,
provide d the original wor k is properly cited.
© 2019 The Auth ors. Diversity and Distributions published by John Wiley & Sons Ltd.
1Federal Univer sity of Rio Gran de do Sul ,
Porto Alegre, B razil
2University of Arizona, Tucs on, AZ, USA
3Federal Univer sity of Minas Ger ais, Be lo
Horizonte, Brazil
Correspondence
Luíz Fern ando Es ser, Federal University of
Rio Gra nde do Sul, Porto Alegre, Brazil.
Email: luizesser@gmail.com
Editor: Janet Franklin
Abstract
Aim: Elucidate the potential impacts of climate changes on the distribution and con-
servation of the multiple habitats of the Mata Atlântica biodiversity hotspot, which
are often treated as a unique entity in ecological studies.
Location: The whole extension of the South American Atlantic Forest Domain plus
forest intrusions into the neighbouring Cerrado and Pampa Domains, which com-
prises rain forest (‘core’ habitat) and five environmentally marginal habitats, namely
high elevation/latitude forest, rock outcrop habitats, riverine forest, semideciduous
forest and restinga woodlands.
Time period: Current (2000) and future scenarios (2050 and 2070).
Major taxa studied: Tree species.
Methods: We modelled the responses of 282 diagnostic tree species, using multiple
algorithms and distinct scenarios of climate change (828,234 projections).
Results: Potential loss of suitable environment summed 50.4% in semideciduous for-
est, 58.6% in riverine forest and 66% in rock outcrop habitats. Predictions for rain
forest (12.2%), restinga woodlands (7.6%) and high elevation/latitude forest (5.2%)
showed that overall loss of suitable environment will be relatively less severe for
these habitats. Habitats that are confined to narrow edaphic conditions, namely rock
outcrop habitats and riverine forest, are less studied and will likely suffer the greatest
loss of biodiversity because their species are more dispersal limited.
Main conclusions: Because these habitats occupy distinct environmental conditions,
lumping them in ecological analyses might lead to erroneous interpretations in stud-
ies aiming to evaluate the impacts of global change in the Mata Atlântica biodiversity
hotspot. This reinforces the importance of our approach and urges for conserva-
tion strategies that account for habitat heterogeneity in the Mata Atlântica and other
species‐rich environments.
KEY WORDS
biodiversity conservation, communities' distribution models, habitat conservation,
macroecology, tree communities, vegetation
    
|
 1847
ESSER Et a l.
1 | INTRODUCTION
The Mata Atlântica of South America is renowned worldwide for
being one of the 36 biodiversity hotspots for conservation prioriti-
zation (Mit termeier, Turner, Larsen, Brooks, & Gascon, 2011; Myers,
Mittermeier, Mittermeier, da Fonseca, & Kent, 20 00; Williams et al.,
2011). Less known facts, however, are that (a) the hotspot status
is specifically referring to its core vegetation type, the rain forest,
and that (b) the Mata Atlântica also houses a diverse and complex
mosaic of vegetation types, with their occurrence and distribution
determined by the harshest extremes of five key environmental fac-
tors (Figure 1; Neves et al., 2017; Scarano, 2009). Thus, vegetation
types are defined here as a plant assemblage and its associated en-
vironmental conditions (hereafter ‘habitat’). Following Walter (1971),
these factors can be classified into azonal (non‐climatic) and zonal
(climatic). The distribution of azonal habitats in the Mata Atlântica
is determined by rocky substrates (rock outcrop dwarf‐forests and
savannas, hencefor th rock outcrop habitats), salinity (white‐sand
woodlands, henceforth restinga woodlands) or waterlogged soils
(tropical riverine forests, henceforth riverine forest), while the
distribution of zonal habitats is determined by frost (montane and
subtropical riverine forests, henceforth high elevation/latitude for-
est), drought stress (semideciduous forests) or high levels of rainfall
(cloud and rain forest, henceforth rain forest).
In a seminal article, Scarano (20 09) argued that environmentally
marginal habitat s in the Mata Atlântica comprise an impoverished
subset of rain forest species that can tolerate the harshest extremes
of their environmental conditions. A recent study, however, showed
that all Mata Atlântica habitats are strikingly distinct both floristi-
cally and environmentally (Neves et al., 2017), suggesting that mar-
ginal habitats are not simply a nested subset of the more diverse
Mata Atlântica rain forest. For conservation purposes, a pertinent
takeaway message in Neves et al. (2017) is that a substantial por-
tion of the plant diversity in the Mata Atlântica might be neglected
if the spatial design for new protected areas is solely based upon
studies that places these multiple habitats together (e.g., Zwiener et
al., 2017).
Currently, marginal habitats receive much less protection com-
pared with the rain forest (Neves et al., 2017), despite harbouring
3,160 tree species that are not found anywhere else in the world,
including in the rain forest of the Mata Atlântica. Yet, current levels
of fragmentation and the continuous habitat loss are high through-
out the Mata Atlântica, raising several concerns in the scientific com-
munity (Galindo‐Leal, Jacobsen, Langhammer, & Olivieri, 2003; Joly,
Metzger, & Tabarelli, 2014; Neves et al., 2017; Tabarelli, Cardoso
Da Silva, & Gascon, 2004; Tabarelli, Pinto, Silva, Hirota, & Bede,
2005). In addition to these impacts associated with land use change
in Mata Atlântica habitats, human‐induced climate change (IPCC,
2013) will have widespread effects on Mata Atlântica's ecosystems
(Ferro, Lemes, Melo, Loyolo, & Fenton, 2014; Lemes, Loyolaet, &
Flammini, 2013; Loyola, Lemes, Brum, Provete, & Duarte, 2014).
The persistence of biodiversit y through such global change will
demand biogeographic shifts at all levels of biological organization
(e.g. from populations to communities to functional groups, Bhatta,
Grytnes, & Vetaas, 2018; Frainer et al., 2017; McLachlan, Hellmann,
& Schwar tz, 2007, respectively. See also Barnosk y et al., 2017, for a
recent review).
In the last decades, ecological niche modelling became a major
tool to predict the impacts of climate changes on biodiversity, aiding
conservation planning in future, dynamic scenarios (Peterson, 2001;
Peterson, Egber t, Sánchez‐Cordero, & Price, 2000; Peterson et al.,
2002). With the development of novel learning machine algorithms
(Guisan & Thuiller, 20 05) and more accurate climate change predic-
tions (Moss et al., 2010), we are now capable to reduce analy tical
FIGURE 1 Distribution of Mata Atlântica habitat s in South America (sensu Scarano, 2009) and main environmental factors (arrows)
sorting species across these habitats (adapted from Neves et al., 2017). Ellipses indicate zonal habitats, and rectangles indicate azonal
habitats
1848 
|
   ESSER Et al.
uncertainties and provide the much‐needed information to support
conservation prioritization while accounting for global change sce-
narios (Elith et al., 2006). This is of particular relevance for biodiver-
sity hotspots, where species are likely to be more susceptible due to
its reduced population sizes caused by habitat fragmentation.
Our goal here is to elucidate the potential impacts of climate
changes in Mata Atlântica habitats' distribution and conservation.
Because Mata Atlântica habitats occupy distinct climatic and geo-
graphic space, our hypothesis is that climate changes will severely
impact all habitats, though to different degrees. In addition, because
South America will experience increasing temperatures with re-
duced water availability (IPCC, 2013), we predict that future climate
changes will have less severe impacts in restinga, rock outcrop hab-
itats and semideciduous habitats, and more severe impacts in plant
communities found at high elevation/latitude and in riverine and rain
forests.
2 | METHODS
2.1 | The dataset
We conducted environmental niche modelling for Mata Atlântica
habitat s using diagnostic species obtained from Neves et al. (2017),
with their presence points available in NeoTropTree (Oliveira‐Filho,
2017). Using diagnostic species to model the climatic distribution of
neotropical vegetation has proven a more efficient approach, given
its higher TSS and AUC values (Bueno et al., 2017) compared with
previous studies (Carnaval & Moritz, 2008; Pena, Kamino, Rodrigues,
Mariano‐Neto, & de Siqueira, 2014; Werneck, 2011; Werneck,
Nogueira, Colli, Sites, & Costa, 2012), and has been effectively used
to determine ecological indicators of community types, habitat con-
ditions and environmental changes (Carignan & Villard, 2002; De
Cáceres & Legendre, 2009; De Cáceres, Legendre, & Moretti, 2010;
De Cáceres, Legendre, Wiser, & Brotons, 2012; Dufrêne & Legendre,
1997; Niemi & McDonald, 2004). To avoid overparameterization
(SDMs in this study have three climatic variables as input data;
see Section 2.2 below), we first excluded species with <20 records
(Thuiller, Guéguen, Renaud, Karger, & Zimmermann, 2019), summing
a tot al of 282 species (see Table S1). These species were classified
in Neves et al. (2017) as diagnostic (see Tichy & Chytry, 2006) of six
Mata Atlântica habitats, with each habitat being distributed across
limited ranges of six environmental gradients: rain forest (warm and
wet climates), high elevation/latitude forest (environments associ-
ated with seasonal cold), semideciduous forest (seasonal drought),
restinga (salinity), rock outcrop habitats (seasonal fire and shallow
soils) and riverine forests (seasonal soil waterlogging). In order to
reflect these limiting environmental conditions in the analyses, we
modelled the species of each habitat using distinct geographic de-
limitations, detailed below (see Figure S2).
Spatial scope for species from high elevation/latitude and rain
forests comprised the whole extent of the Mata Atlântica and the
biogeographical Domains found in the neighbouring South American
dry diagonal, namely Caatinga, Cerrado and Chaco. Because species
from semideciduous forests are widely distributed across the dry
diagonal, their spatial scope comprised the Mata Atlântica, dry di-
agonal Domains and the neighbouring lowland Amazon (warmer cli-
mates). Restingas, riverine and rock outcrop habitats are constrained
within conditions that are primarily related to soil. Therefore, despite
species from restingas, riverine and rock outcrop habitats having cli-
matic suitability in other habitats (e.g. rain forests), these species
are restricted to specific edaphic conditions (e.g. soil waterlogging
in riverine forests). Thus, we modelled the potential distribution of
these species within their edaphically suitable areas, which we es-
tablished as the current distribution of restingas, riverine and rock
outcrop habitats, respectively. We defined the distribution of the
Mata Atlântica habitats, dry diagonal Domains and lowland Amazon
in geographic space by creating polygons from a set of points. The
6,243 NeoTropTree sites (points) were previously classified into one
of the South American biogeographic Domains and into one of the
Mata Atlântica habitats where applicable. The size of each polygon
was then estimated based on the distance between a given site and
the other sites around it (wall‐to‐wall map).
Bioclimatic variables were obtained from Wor ldClim v.1.4
(Hijmans, Cameron, Parra, Jones, & Jarvis, 2005). Climatic layers
were obtained at a 5‐arcmin grain size (~10 km). This spatial resolu-
tion is particularly appropriate for this study because species check-
lists (sites) in NeoTropTree are defined by a single habitat, following
the classification system proposed by Oliveira‐Filho (2017), con-
tained in a circular area with a 10 km diameter. NeoTropTree data
were originally compiled from an extensive survey of published and
unpublished (e.g. PhD theses) literature, particularly those on woody
plant community surveys and floristic inventories. New species oc-
currence records obtained from both major herbaria and taxonomic
monographs were then added to the checklists when they were col-
lected within the 10 km diameter of the original NeoTropTree site
and within the same habitat. The habitat delimitation was conducted
using the package ‘dismo’ (Hijmans & Elith, 2015) in R Statistical
Environment (R Development Core Team, 2011).
2.2 | Variable selection
Variable selection was very conservative in order to build under-
standable and ecologically meaningful models (Figure 2). We fol-
lowed a multiple‐step variable selection routine, consisting of the
following: (a) using variance inflation factors (VIF) to identify highly
collinear variables, which were progressively excluded through a
stepwise procedure. VIFs were computed using two methods: VIFcor
(threshold = 0.5) and VIFstep (threshold = 10; see Marquardt, 1970,
for meth od details). We th en extracte d bioclimatic values from pre s-
ence points and (b) performed a principal components analyses (PCA)
to visualize which variables were more effective in segregating the
climatic space of each habitat relative to the climatic space of all other
habitat s. We also (c) per formed PCA s for each habitat separately to
assess which climatic variables showed higher correlations with the
first three principal components (there was a negligible increase in
constrained variance by adding a four th component). Lastly, we (d)
    
|
 1849
ESSER Et a l.
used Pearson's correlation to test whether all variables selected for
a given habitat showed low correlation (cut‐off = − 0.5 < p < .5). We
also legitimated the variable selection with literature review, which
allowed us to select variables that better represented the climatic
space occupied by the species of each habitat, while taking into ac-
count their ecological relevance (see Table S3).
2.3 | Environmental niche modelling
Models were calculated in three independent cross‐validation runs
with 30% of data kept to evaluate the model and two evaluation
met ho ds (true sk ill st atistic, TSS, and are a un de r the recei ver operat-
ing charac teristic, AUC) for every algorithm available in biom od2r
Package (Thuiller, Georges, Engler, Georges, & Thuiller, 2014; gen-
eralized linear models, generalized additive models, boosted regres-
sion trees, classification tree analysis, ar tificial neural net works,
Bioclim, flexible discriminant analysis, multiple adaptive regression
splines, random forest and MaxEnt). We only kept ensemble mod-
els with TSS higher than 0.7. We generated 1,000 pseudoabsences
through different background areas for each habitat, since they have
distinc t spatial scopes in our analyses (see section 2.1 and Figure
S2 for more details). A caveat to this approach is the recommenda-
tions of Barbet‐Massin, Jiguet, Albert, and Thuiller (2012) regarding
the use of lower pseudoabsences in some algorithms. Nonetheless,
here we followed Thuiller (2014), which points out that the main ad-
vantage of biomod2 lies in the capability to compare and combine
mul tiple algor it hms using the sam e se t of init ial data and paramete ri-
zation. We controlled for spatial autocorrelation in our models using
a generalized least squares framework (Zuur, Ieno, Walker, Saveliev,
& Smith, 2009), wh ic h consists in modelling alpha diver sity as a func-
tion of predicting variables using different spatial correlation struc-
tures (exponential, gaussian, spherical, linear and rational quadratics)
and then selecting the best model (highest delta AIC relative to the
null model; i.e. no spatial autocorrelation). We then built a raster with
cell sizes as values weighted by presence probabilities to provide a
more conser vative measure of the potential area occupied by each
habitat (Figure 2). Models were projected to CMIP5 data (Coupled
Model Intercomparison Project Phase 5; downscaled at 5‐arc‐min-
ute spatial resolution) using all General Circulation Models available
in WorldCli m v.1.4 (Hijmans et al., 2005) to the four Representative
Concentration Pathways (RCP2.6, 4.5, 6.0 and 8.5) to the years of
2050 and 2070 (BCC‐CSM1‐1, CCSM4, GISS‐E2‐R, HadGEM2‐AO,
HadGEM2‐ES, IPSL‐CM5A‐LR, MIROC‐ESM‐CHEM, MIROC‐ESM,
MIROC5, MRI‐CGCM3 and NorESM1‐M), summing 88 scenarios and
a total of 828,234 projections. Species projections were summed
into an alpha diversity raster for each habitat and weighted by the
FIGURE 2 Methods summary.
Environmental niche models (ENM) were
projected to 11 Atmosphere Ocean
General Circulation Models (AOGCM)
and four represent ative concentration
pathways (RCP) to 2050 and 2070.
To calculate potential occupied area,
the presence probability rasters were
multiplied by cell area rasters, generating
a weighted area raster, following two
approaches: (i) considering a presence–
absence map with a threshold = 0.5, that
is, each cell with presence probabilit y
>0.5 sum 100 km2 (grid cell size) of the
total potential area. (ii) Considering that
cells could be partially occupied, that
is, occupancy models that are either
gradually fading (a) or abruptly changing
(b) ecotones
1850 
|
   ESSER Et al.
maximum number of species before generating habitat suitability
maps (Figure 2). Ensemble models were generated for each habitat
by first summing their diagnostic species distribution maps and then
dividing the resulting map by the number of diagnostic species in a
given habitat. This generated a final suitability map (ranging from
zero to one) for each habitat.
Finally, to assess the potential conservation status of Mata
Atlântica habitats, we overlaid the current and future distributions
of each habitat on to the coverage of protected areas in the World
Database on Protected Areas (IUCN & UNEP‐WCMC , 2015).
3 | RESULTS
3.1 | Potential area and conservation status
Our models showed that in current climatic conditions, the existing
network of protected areas is more effective in protecting the po-
tential distribution of azonal habitat s (17.4% compared to only 9.0%
in zonal habitats, Figure 3). Semideciduous forest is the least pro-
tected habitat, with only 7.1% of its potential area (537,640.29 km2)
occurring within protected areas (39,320.39 km2). Amongst azonal
habitat s, riverine forest was the least protected, with only 8.5% of
its potential area (91,492.64 km2) occurring within protec ted areas
(7,816 km2). On average, 13.8% of the potential distributions of mar-
ginal habitats are found within protec ted areas, which is higher than
the potential distribution of rain forest occurring within protec ted
areas (10.2%; 41,203.47 km2).
From current conditions to the worst climate change scenario,
the high elevation/latitude forest was the least affected, with 5.2%
of potential area shrinkage, followed by restinga (7.6%), and rain for-
est (12.2%). In contrast, future scenarios for semideciduous, riverine
and rock outcrop habitats were worrisome. Potential area shrinkage
in future climatic scenarios can be as high as 50.4% in semideciduous
forest, 58.6% in riverine forest and 66% in rock outcrop habitats.
This loss of climatically suit able areas across all habit ats is also re-
flected in their levels of protection. From current to worst scenario,
restinga woodlands are predicted to lose climatic suitability in 6.6%
of its currently protected area, followed by high elevation/latitude
(8.0%), rain (13.6%) and riverine (55.3%) forests. The current net-
work of protected areas in rock outcrop habitats is predicted to
undergo the most severe impacts of climate change, with 60.1% of
shrinkage in areas of climatic suitability for species of rock outcrop
habitat s in these protected areas. Conversely, shrinkage in areas of
climatic suitability for species of semideciduous forest (50.4%) will
mainly occur outside protec ted areas (19.0% of protected area loss).
3.2 | Distribution of azonal habitats
Riverine forests, which are mainly found in Central Brazil, and rock
outcrop habitats, which are mainly found in the transition between
Mata Atlântica and Cerr ado, are predicted to lose higher levels of cli-
matic suitability in lower latitudes (see Figures S4 and S5). Restinga is
predicted to lose lower amounts of suitable climatic space relative to
the other Mata Atlântica habitats, suggesting higher climatic stability
across coastal white‐sand environments in eastern South America
(see Figures S6 and S7).
3.3 | Distribution of zonal habitats
Our results showed a substantial degree of overlap in the climatic
spaces occupied by species from high elevation/latitude, semidecid-
uous and rain forests (for decoupled maps check Figures S8, S9 and
S10). This suggests that the abrupt contours that are currently used
for delimiting the distribution of these three habitats might be too
simplistic (Figure 4). Under current climatic conditions, for instance,
our models showed that for 3.7% of the geographic space covered
FIGURE 3 Potential area (in km2) of
Mata Atlântica habitats (total area in a
and b; protec ted area in c and d) through
scenarios of increase in CO2 concentration
(a) (b)
(c) (d)
    
|
 1851
ESSER Et a l.
by zonal habitats, there is an equivalent probability that a given area
(~100 km2) is suitable for species from high elevation/latitude, sem-
ideciduous and rain forests. This intercept increases through scenar-
ios, varying from 6.2% of overlap in RCP2.6/2070 and RCP6.0/2070
to 7.1% in RCP8.5/2070.
Climatic overlap between two habitats is even higher. Species
from high elevation/latitude and rain forest showed the highest de-
gre e of overlap in climati c su it abil it y, ranging from 14% in cur re nt cli-
matic conditions to 24.3% in RCP8.5/2070. In contr ast, species fro m
se mid e cidu ous an d ra in fo res ts sh owe d a much lo w er deg ree of ove r-
lap in climatic suitability (6.7% in current climate), which decreases
over time (1.2% in RCP8.5/2070). Unique climatic space (i.e., suitable
for species of a single habitat) is highly variable across high eleva-
tion/latitude, semideciduous and rain forests, and unstable through
time. Potential climatic uniqueness for rain forest species ranges
from 1.1% in current climate conditions to 0.5% in RCP8.5/2050,
reaching a minimum of 0.2% in RCP6.0/2050. Semideciduous forest
showed both the highest degree of climatic uniqueness and future
instability, ranging from 25.1% in current climatic conditions to only
8% in RCP8.5/2070. Species from high elevation/latitude forest
showed 18.9% of potential climatic uniqueness, which decreases to
18.1% in RCP8.5/2070 and 13.8% in RCP4.5/2050 (Figure 5).
3.4 | Climatically stable areas
Areas in southeastern Brazil showed a high probability of climatic
stability for species of all three zonal habitats (Figure 6). These po-
tential refugia occur mainly in Rio de Janeiro and São Paulo states.
Potential refugia for species of high elevation/latitude forest are also
found in southern Brazil and Uruguay. Potential refugia for species
of semideciduous forests are scattered across central and south-
eastern Brazil, with larger areas in Minas Gerais state. The distribu-
tion of protected areas shows a low level of coincidence with these
postulated refugia (Figure 6), ranging from 13.4% in high elevation/
latitude and semideciduous forest to 32.8% in rain forests.
Areas in eastern and central‐western Brazil showed a high prob-
ability of climatic stability for species of azonal habitat s. Existing
protected areas in the Federal District and across Minas Gerais state
(e.g., Canastra National Park) are potential refugia for species of riv-
erine forest. Potential refugia for species of rock outcrop habitats
FIGURE 4 Distribution of zonal
habitats through climate change
scenarios yielded by environmental niche
modelling of their diagnostic species (see
Table S1 and Section 2). Habitats were
plotted using a red‐green‐blue colour
scheme. The brightest shades of red,
green and blue represent the highest
probability of occurrence of species from
semideciduous, rain and high elevation/
latitude forests, respectively. If a grid cell
is potentially occupied by two habitats,
its colour will represent an intermediate
palette between the colours for these
two habit ats. White indicates presence
of all habitats, while black indicates full
absence. Pearson's correlations between
current and each of the future distribution
maps are given for semideciduous, rain
and high elevation/latitude forests (S, R
and H, respectively)
1852 
|
   ESSER Et al.
are scattered in Minas Gerais state (Gandarela and Caparaó National
Parks, and Brigadeiro State Park). Large areas of climatic stability
areas for species of restinga woodlands are found in northeastern
Brazil, across the coastline of Bahia, Alagoas and Pernambuco states.
However, these climatically stable restinga woodlands are mostly
found out side existing protected areas (only 19.9% within protected
areas; Figure 6).
4 | DISCUSSION
Here, we showed that both core and marginal habitats of the Mata
Atlântica will be severely impacted by human‐induced climate
change, though to different, uneven degrees. For instance, consider-
ing variation from current conditions to the most pessimistic sce-
nario of climate change in our models (RCP8.5/2070), rain forest is
likely to be more climatically stable relative to semideciduous, river-
ine and rock outcrop habitats, but more impacted than high eleva-
tion/latitude forest and restinga woodlands. These findings are of
relevance for conser vation planning predicated on the protection of
biodiversity under climate change scenarios. Because there is a con-
siderable level of plant endemism in both core and marginal habitats
(Neves et al., 2017), a portion of such species could be neglec ted if
future conservation strategies prioritise regions of highest climatic
stability regardless of habitat heterogeneity (e.g., Lemes et al., 2013;
Ferro et al., 2014; Loyola et al., 2014; Zwiener et al., 2017, Sobral‐
Souza, Vancine, Ribeiro, & Lima‐Ribeiro, 2018), but core and mar-
ginal habitats are unevenly distributed across these stable regions.
4.1 | Potential area and conservation status
Through the scenarios, protected areas in riverine forest will have
more stable climates across the southeastern portion of its current
distribution, highlighting the importance of these areas for protect-
in g via b le pop u lat ion si ze s of riv erin e spec i es. Co ngr uen t wit h the re -
sults for riverine forest, our future scenario models showed that rock
outcrop habitats will lose more climatically suitable areas in their
lower latitudes, suggesting that southernmost sites may function
as climatic refugia for this hyperdiverse habitat (Neves et al., 2018).
However, given the scattered spatial configuration of these rock
outcrop sites, dispersal is likely to be very limited, which suggests
that conservation strategies might need to consider new protected
areas that connect these outcrop islands through the lowlands. In
fact, previous studies (Mews, Pinto, Eisenlohr, & Lenza, 2014; Neves
et al., 2018) provided evidence that rock outcrop habitats and their
surrounding lowland savannas are likely to form a continuous meta-
communit y with spatial variation in woody plant population sizes
being mainly driven by source–sink dynamics (Pulliam & Danielson,
FIGURE 5 Number of grid cells per
zonal habitat and their intercepts through
climate change scenarios. Bottom groups
represent the total number of grid cells
of semideciduous (S.Set), rain (R. Set)
and high elevation/latitude forests
(H.Set). Upper‐left groups represent
grid cells that in our models are uniquely
covered by semideciduous (S), rain (R)
and (H) high elevation/latitude forest s.
Upper‐right groups represent grid cells
where three (high elevation/latitude‐
rain‐semideciduous, HRS) or two habitats
overlap (semideciduous‐rain, SR; high
elevation/latitude‐semideciduous, HS;
high elevation/latitude‐rain, HR). Habitat s
with climatic suitability ≥0.33 in a grid cell
were considered present. Chord diagrams
were made using circlize R package (Gu,
Gu, Eils, Schlesner, & Brors, 2014)
    
|
 1853
ESSER Et a l.
1991). Therefore, here we stress that protected areas aiming to se-
cure biodiversity of rock outcrop habit ats should not be limited to
rock outcrop areas. Rather, effective protected areas should func-
tion as ecological corridors connecting multiple rock outcrop sites
through lowland environments.
Our models showed that while climate in restinga woodlands are
expected to be more stable over time when compared to other hab-
itats, this level of stability is highly variable within its distribution,
with central and southern restingas being relatively more stable.
In addition to this uneven impact of climate change across restinga
woodlands, coastal environments are also expected to be affec ted
by erosion and sea level raise (EUROSION, 2004; IPCC , 2013). This
suggests that conservation planning for restinga woodlands will
require a high degree of complexity, with its effectivity depending
on strategies that account for geomorphological variation changes
associated with both climate and land use change. Restinga has
suffered massive fragmentation due to high human occupation in
coastal areas and a rapidly developing, unplanned tourism industry.
Amongst zonal habitats, semideciduous forest is predicted
to be the most impacted, losing 64% of its current potential dis-
tribution under the most pessimistic scenario (RCP8.5/2070).
Mo re ove r, whi le ou r mode ls pre di c t cli mat ic st ability fo r spec ies of
semideciduous forest in southeastern Brazil, there is a high degree
of potential area shrinkage for species of semideciduous forests
in northeastern Brazil (see Figure S10). These results therefore
suggest that conservation strategies aiming to protect suitable
climatic space for these northern species would have to consider
corridors that could potentially link their current and future suit-
able climates. Conversely, high elevation/latitude and rain forests
are relatively stable over time, indicating the need for tailor‐made
conservation strategies for each habitat of the Mata Atlântica.
Nonetheless, biodiversity in these forests is poorly and unequally
captured by the current net work of protected areas, especially in
southern Brazil (Saraiva, dos Santos, Overbeck, Giehl & Jarenkow,
2018). Here, we suggest that accounting for climate change sce-
narios, in addition to multi‐dimensional biodiversity assessments
FIGURE 6 Climatic st abilit y in Mata
Atlântica habit ats yielded by ecological
niche models of 269 diagnostic tree
species. Coloured grid cells (stable sites)
represent areas where all diagnostic
species of a given habitat are predicted
to occur in all 89 scenarios of current and
future climates (four concentrations of
atmospheric carbon for the years 2050
and 2070, 11 AOGCMs). Red contours
indicate the current network of protec ted
areas in South America. Black contours
represent the national borders and state
limits in Brazil. Values in parentheses
indicate the amount of climatic ally st able
areas in square kilometres for each
habitat. Acronyms represent Brazilian
states mentioned in the Results: Alagoas
(AL), Bahia (BA), Federal District (FD),
Minas Gerais (MG), Pernambuco (PE), Rio
de Janeiro (RJ) and São Paulo (SP)
1854 
|
   ESSER Et al.
as in Saraiva et al. (2018), might improve current and future con-
servation strategies for these neglected high elevation/latitude
and rain forests.
4.2 | Climate change and compositional complexity
Previous studies (Neves et al., 2017; Oliveira‐Filho, Budke, Jarenkow,
Eisenlohr, & Neves, 2015; Oliveira‐Filho & Fontes, 2000) that ad-
dressed climatic differentiation amongst Mata Atlântica habitats
showed that while these habitats are floristically distinct, such com-
positional dif ferentiation is only partially explained by variation in
current climatic conditions. Our models not only supported the idea
that delimiting the distribution of Mata Atlântica habitats is no easy
task, but also showed that such complexity will likely increase under
climate change, that is, because we currently lack a complete un-
derstanding of the factors that control the distribution of species
through space and climatic gradients, predicting climate‐driven bio-
geographical shifts is inherently uncer tain.
There are many potentially important factors in determining the
distribution of species that we have not accounted for adequately and
that should be considered/addressed in future studies (see Neves et
al. , 2018; Ti teux, Duf re ne, Jac ob, Paq uay, & Defo ur ny, 20 04). Amongst
these factors, the importance of biotic processes (e.g. competition,
natural enemies) to species distributions and community composition
is the most neglec ted in the literature, especially in studies address-
ing compositional turnover under climate change scenarios. Here, we
highlight that accounting for biotic processes and assessing how they
may potentially var y through time is not trivial for studies aiming to
accurately predict the impacts of global change on biodiversity.
4.3 | Climatic stability and protected areas
Biodiversity loss from climate change arises because species move to
track suitable climate, and agricultural lands, urban development or
transportation corridors may stop their movement (Hannah, Midgley,
Hughes, & Bomhard, 2005; Heller & Zavaleta, 2009). Protected areas
and biodiversity‐friendly land uses lessen these barriers to movement
(U r b an, 20 15) , but th e d a t a nee ded to inf o r m la nd us e man age r s req uir e
ins ight s from ecol og is ts in which the movements of var ious species ar e
modelled under multiple climate scenarios. In our models, climatically
stable areas are mostly outside the existing protected areas (83.8%).
We, therefore, suggest that the areas identified as climatically stable
in our analyses should be incorporated into systematic conservation
planning and restoration project s to preser ve Mata Atlântica habitats.
Altogeth er, these areas function as probable refugial are as and climati-
cally stable corridors connecting unstable protected areas to currently
protected refugial areas.
5 | CONCLUSIONS
Our study showed that ‘lumping’ the natural heterogeneity of
the Mata Atlântica can bring great havoc for future conser vation
strategies, and highlighted three additional factors to be consid-
ered in conservation planning for this biodiversity hotspot: (a)
we still have little understanding of how climate controls species
distribution across the Mata Atlântica, and therefore, the future
distri bution of spec ies fro m zonal habitats, namel y high elevation/
latitude, semideciduous and rain forests, is highly uncertain. New
conservation strategies will need to account for such uncertainty
when estimating which areas in geographic space are more likely
to protect species from a given habitat and which areas are likely
to represent climatic overlaps that are suitable for species from
two o r more habit ats. (b) The maintenance of ha bit at area through
tim e wi ll like ly depend on major bio geographi ca l sh if ts (see result s
for semideciduous forests). Thus, new conservation strategies
will need to account for the climatic space that will likely facili-
tate gradual migration under a changing environment. (c) Under
climate change scenarios, spatial rearrangements for species of
azonal habitats can only occur within the range that comprises
their edaphic requirements, namely rock outcrops (rock outcrop
habitats), seasonally waterlogged soils (riverine forest) and white‐
sand saline soils (restinga woodlands). This leads to a more lim-
ited array of conser vation strategies for these habitats. Thus, for
azonal habitats, considering conser vation strategies that prevent
the cu rrent ly high level s of fr ag menta ti on asso ci at ed with land use
change is a must.
Further studies assessing climate changes impacts in habitats
may trace how areas might change in (diagnostic) species compo-
sition and richness over time, culminating in the emergence of new
habitat s. In terms of azonal habitats, plant–soil relationship should
be addressed carefully, considering influences of climate on sub-
strates, as well as the suitability for plants under new combination of
climate and edaphic conditions.
ACKNOWLEDGEMENTS
We thank Marinez Siqueira, Gerhard Overbeck and Demétrio
Guadagnin for their insightful considerations on an earlier version of
th i s manu scr i pt. We al so tha nk tw o an ony mou s refer ees fo r thei r val -
uable contributions to this manuscript. D.M.N. was funded by a col-
laborative research grant from the US National Science Foundation
(DEB‐1556651) during the time this research was completed.
DATA AVA ILAB ILITY STATE MEN T
All environmental layers are available in the WorldCl im data-
base (http://www.world clim.org). Species data are available in the
NeoTropTree website (http://www.neotr optree.info).
ORCID
Luíz Fernando Esser https://orcid.org/0000‐0003‐29827223
Danilo M. Neves https://orcid.org/0000‐0002‐0855‐4169
João André Jarenkow https://orcid.org/0000‐0003‐2747‐3468
    
|
 1855
ESSER Et a l.
REFERENCES
Barbet‐Massin, M., Jiguet, F., Albert , C. H., & Thuiller, W. (2012). Selecting
pseudo‐absences for species distribution models: How, where and
how many? Methods in Ecology and Evolution, 3(2), 327–338. https ://
doi .org/10.1111/j. 2041‐210X .2011 .0 0172.x
Bar nosky, A. D., Had ly, E. A., Gonz al ez, P., Hea d, J., Polly, P. D., La wi ng ,
A. M., … Zhang, Z. (2017). Merging paleobiology with conservation
biology to guide the future of terrestrial ecosystems. Science, 355,
eaah4787. https ://doi.org/10.1126/scien ce.aah4787
Bhat ta , K . P., Gr ytne s, J. A ., & Veta as , O. R. (2 018 ). Dow nh ill shif t of alpine
plant assemblages under contemporary climate and land‐use changes.
Ecosphere, 9(1), e02084. https ://doi.org/10.1002/ecs2.2084
Bueno, M. L ., Pennington, R . T., Dex ter, K. G., Kamino, L. H. Y., Pontara,
V., Neves, D. M., … de Oliveira‐Filho, A . T. (2017). Effects of quater-
nary climatic fluctuations on the distribution of Neotropical savanna
tree species. Ecography, 40(3), 4 03–414. ht tp s ://doi. org/10 .1111/
ecog.01860
Carignan, V., & Villard, M.‐A. (2002). Selecting indicator species to mon-
itor ecological integrit y: A review. Environmental Monitoring and
Assessment, 78, 45–61.
Carnaval, A. C., & Moritz, C. (2008). Historical climate modelling pre-
dicts pat terns of current biodiversit y in the Brazilian Atlantic
forest . Journal of Biogeography, 35, 1187–1201. https ://doi.
org /10.1111/j .136 5‐269 9.20 07.01870 .x
De Cáceres, M., & Legendre, P. (20 09). Associations between species and
groups of sites: Indices and statis tical inference. Ecology, 90, 3566–
3574. https ://doi.org/10.1890/08‐1823.1
De Cáceres, M., Legendre, P., & Moretti, M. (2010). Improving indicator
species analysis by combining groups of sites. Oikos, 119, 1674–1684.
https ://doi.o rg /10.1111 /j.160 0‐070 6. 20 10.1833 4. x
De C ácer es , M., Leg endr e, P., Wis er, S. K. , & Bro ton s, L . (2 012 ). Us i n g sp ec ies
combinations in indicator value analyses. Methods in Ecol ogy a nd Evoluti on,
3, 973–982. https ://doi.org/10.1111/j.2041‐210X.2012.00246.x
Dufrêne, M., & Legendre, P. (1997). Species assemblages and indicator
species: The need for a flexible asymmetrical approach. Ecological
Monographs, 67, 345–366. https ://doi.org/10.2307/2963459
Elith, J., H. Graham, C., P. Anderson , R., Dudík, M., Ferrier, S., Guisan, A .,
… E. Zimmermann, N. (20 06). Novel methods improve prediction of
species’ distributions from occurrence data. Ecography, 29(2), 129–
151. https ://doi.org/10.1111/j.2006.0906‐7590.04596.x
EUROSION (2004). Living with coastal erosion in Europe – Sediment and
space for sustainability. 40 p, Luxemburg: Europ ean Commission.
Ferro V. G., Lemes P., Melo A. S., Loyola R., Fenton B. (2014). The Reduced
Effectiveness of Protected Areas under Climate Change Threatens
Atlantic Forest Tiger Moths. PLoS ONE, 9(9), e107792.
Frainer, A., Primicerio, R ., Kor tsch, S., Aune, M., Dolgov, A. V., Fossheim ,
M., & Aschan, M. M. (2017). Climate‐driven changes in functional
biogeogr aphy of Arctic marine fish communities. Proceedings of the
National Academy of Sciences of the United States of America, 114 (46),
12202–12207. https ://doi.org/10.1073/pnas.17060 80114
Galindo‐Leal, C., Jacobsen, T. R., Langhammer, P. F., & Olivier i, S.
(2003). State of t he hotspots: The dynamics of biodiversity loss.
In C. Galindo‐Leal, & I. G. de mara (Eds.), The Atlantic Forest of
South America: Biodiversity status, threats, and outlook (pp. 12–23).
Washington, DC: Center for Applied Biodiversity Science and
Island Press.
Gu, Z., Gu, L., Eils, R., Schlesner, M., & Brors, B. (2014). circlize imple-
ments and enhances circular visualization in R. Bioinformatics, 30,
2811–2812 .
Guisan A., Thuiller W. (20 05). Predicting species distribution: offering
more than simple habitat models. Ecology Letters, 8(9), 993–100 9.
Hannah, L., Midgley, G., Hughes, G., & Bomhard, B. (20 05). The view
from the Cape: extinction risk , protected areas and climate change.
BioScience, 55(3), 231–242.
Heller, N. E., & Zavaleta, E. S. (20 09). Biodiversit y management in the
face of climate change: a review of 22 years of recommendations.
Biological conservation, 142(1), 14–32.
Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G., & Jarvis, A.
(2005). Very high resolution interpolated climate surfaces for global la nd
areas. International Journal of Climatology, 25, 1965‐1978.
Hijmans, R. J., & Elith, J. (2015). Species distribution modeling with R . R
CRAN Project, 79 pp. Retrieved from http://www.idg.pl/mirro rs/
CRAN/web/packa ges/dismo/ vigne ttes/sdm.pdf%5Cnpa pers://
e 2 1 e 3 1 4 0 ‐ c c c 7 ‐ 4 1 4 2 ‐ 9 1 1 c ‐ e 7 4 f c 5 c e e c f 7 / P a p e r / p 1 0 1 6 2
IPCC (2013). IPCC Fifth Assessment Report (AR5). IPCC , s. 10–12.
Cambridge: Cambridge University Pre ss.
IUCN and UNEP‐WCMC (2015). The World Database on Protected Areas
(WDPA). Retrieved from ww w.prote ctedp lanet.net
Joly, C. A., Metzger, J. P., & Tabarelli, M. (2014). Experiences from the
Brazilian Atlantic Forest: Ecological findings and conservation ini-
tiatives. New Phytologist, 204(3), 459–473. https ://doi.o rg /10.1111 /
nph.12989
Lemes, P., Loyola, R. D., & Flammini, A. (2013). Accommodating Species
Climate‐Forced Dispersal and Uncertainties in Spatial Conservation
Planning. PLoS ONE, 8(1), e54323.
Loyola, R. D., Lemes, P., Brum, F. T., Provete, D. B., & Duarte, L. D. S.
(2014). Clade‐specific consequences of climate change to amphibian-
sin Atlantic Forest protected areas. Ecography, 37(1), 65–72.
Marquardt, D. W. (1970). Generalized inverses, ridge regression, biased
linear estimation, and nonlinear estimation. Technometrics, 12(3),
591–612 .
M c La ch la n , J. S. , He ll m an n, J . J. , & S c hw ar t z , M . W. (2 0 0 7 ). A fr a m ew or k fo r de -
bate of assisted migration in an era of climate change. Conservation Biology,
21(2), 297–302. https ://doi.org/10.1111/j.1523‐1739.20 07.00676.x
Mews, H. A., Pinto, J. R. R., Eisenlohr, P. V., & Lenza, E. (2014). Does size
matter? Conservation implications of differing woody population
sizes with equivalent occurrence and diversit y of species for threat-
ened savanna habitats. Biodiversity and Conservation, 23(5), 1119–
1131. ht tps ://doi.org/10.10 07/s10531‐014‐0 651‐4
Mittermeier, R. A., Turner, W. R., Larsen, F. W., Brooks, T. M., & Gascon, C.
(2011). Global biodiversity conser vation: The critical role of hotspots.
In F., Zachos & J., Habel (Eds), Biodiversity hotspots (p p. 3–2 2). Sp r i nger,
Berlin, Heidelberg. https ://doi.org/10.1007/978‐3‐642‐20992‐5_1
Moss, R. H., Edmonds, J. A., Hibbard, K. A., Ma nning, M. R., Rose, S. K., Van
Vuu ren , D. P., … & Meeh l, G. A. (2010 ). The nex t g en era ti on of sc en ari os
for climate change research and assessment. Nature, 463(7282), 747.
Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B.,
& Kent, J. (2000). Biodiversity hotspots for conservation priorities.
Nature, 403(6772), 853–858. ht tps ://doi.org/10.1038/350 02501
Neves, D. M., Dexter, K. G., Pennington, R. T., Bueno, M. L., de Miranda, P.
L. S., & Oliveira‐Filho, A. T. (2018). Lack of floristic identity in campos
rupestres—A hyperdiverse mosaic of rocky montane savannas in South
America. Flora, 238, 24–31. https ://doi.org/10.1016/j.flora.2017.03.011
Neves, D. M., Dex ter, K. G., Pennington, R . T., Valente, A. S. M., Bueno,
M. L., Eisenlohr, P. V., … Oliveira‐Filho, A. T. (2017). Dissecting a bio-
diversity hotspot: The importance of environmentally marginal hab-
itats in the Atlantic Forest Domain of S outh America. Diversity and
Distributions, 23(8), 1–12. https ://doi.org/10.1111/ddi.12581
Niemi, G. J., & McDonald, M. E. (2004). Application of ecological indica-
tors. Annual Review of Ecology, Evolution, and Systematic s, 35, 89–111.
https ://doi.org/10.1146/annur ev.ecols ys.35.112202.130132
Oliveira‐filho, A. T. (2017). NeoTropTree, Flora arbórea da Região
Neotropical: Um banco de dados envolvendo biogeografia, diversidade e
conservação. Retrieved from http://www.neotr optree.info
Oliveira‐Filho, A. T., Budke, J. C., Jarenkow, J. A., Eisenlohr, P. V., & Neves,
D. R . M. (2015). Delving into the variations in tree species compo-
sition and richness across South American subtropical Atlantic and
Pampean forests. Journal of Plant Ecology, 8(3), 242–260. https ://doi.
org /10.1093/jp e/rtt058
1856 
|
   ESSER Et al.
Oliveira‐Filho, A. T., & Fontes, M . A. L. (2000). Patterns of floris tic dif-
ferentiation among Atlantic Forests in Southeas tern Br azil and
the influence of climate1. Biotropica, 32(4b), 793–810. https ://doi.
org /10.1111/j .1744 ‐7429.200 0. tb 006 19.x
Pena, J. C . D. C., Kamino, L . H . Y., Rodrigues, M., Mariano‐Neto, E., & de
Siqueir a , M. F. (2 0 14) . As s e ssing the conser v a t ion st a t u s of sp e c i e s wi t h
limited available data and disjunct distribution. Biological Conservation,
170, 130–136. https ://doi.org/10.1016/j.biocon.2013.12.015
Peterson, A . T. (20 01). Predicting species ’ geographic distributions
based on ecologic al niche m odeling. The Condor, 103, 599–605. https
://doi.org/10.1650/0010‐5422(2001)103[0599:PSGDB O]2.0.CO;2
Peterson, A. T., Egbert, S. L., Sánchez‐Cordero, V., & Price, K. P. (2000).
Geographic analysis of conservation priority: Endemic birds and
mammals in Veracruz, Mexico. Biological Conservation, 93 (1), 85–94.
https ://doi.org/10.1016/S0006‐3207(99)00074‐9
Peterson, A. T., Ortega‐Huerta, M. A., Bartley, J., Sanchez‐Cordero, V.,
Soberon, J., Buddemeier, R. H., & Stockwell, D. R . B. (2002). Future
projections for Mexican faunas under global climate change scenar-
ios. Nature, 416 (6881), 626–629. https ://doi.org/10.1038/416626a
Pull iam , H. R ., & Da nie lso n, B. J. (1991). Sour ces , sinks, and habit at selec-
tion: A landscape perspective on population dynamics. The American
Naturalist, 137, S50–S66. https ://doi.org/10.1086/285139
R Development Core Team (2011). R: A language and environment for sta-
tistical computing. R Foundation for Statistical Computing.
Saraiva, D. D., Santos, A. S. D., Overbeck , G. E., Giehl, E. L. H., &
Jarenkow, J. A. (2018). How effective are protected areas in con-
serving tree taxonomic and phylogenetic diversity in subtropical
Brazilian Atlantic Forests? Journal for Nature Conservation, 42, 28–35.
Scarano, F. R. (2009). Plant communities at the periphery of the Atlantic
rain fores t: Rare‐species bias and its risk s for conser vation . Biological
Conservation, 142(6), 1201–1208. https ://doi.org/10.1016/j.biocon.
2009.02.027
Sobral‐Souza, T., Vancine, M. H., Ribeiro, M. C., & Lima‐Ribeiro, M. S.
(2018). Ef ficiency of protected areas in Amazon and Atlantic Forest
conser vation: A spatio‐temporal view. Acta Oecologica, 87, 1–7.
Tabarelli, M., Cardoso Da Silva, J. M., & Gascon, C. (2004). Forest frag-
mentation, synergisms and the impoverishment of neotropical for-
ests. Biodiversity and Conservation, 13 (7), 1419–1425. https ://doi.
org/10.1023/B:BIOC.00000 19398.36045.1b
Tabarelli, M., Pinto, L. P., Silva, J. M. C., Hirota, M., & Bede, L. (2005).
Challenges and opportunities for biodiversity conservation in the
Brazilian Atlantic Forest. Conservation Biology, 19(3), 695–700. https
://doi.org/10.1111/j.1523‐1739.2005.00694.x
Thuiller, A. W., Georges, D., Engler, R., Georges, M. D., & Thuiller, C. W.
(2014). Package ‘biomod2’ . Species distribution modeling within an en-
semble forecasting framework. https://CRAN. R-project. org/package=
biomod2.
Th ui ll er, W. (2014) . Ed it or ial comm en tar y on ‘B IO MOD – Opt im iz ing pre -
dictions of species distributions and projecting potential future shifts
under global change’. Global Change Biology, 20, 3591–3592. https ://
doi .org/10.1111/gcb.1272 8
Thuiller, W., Guéguen, M ., Renaud, J., Karger, D. N., & Zimmermann , N.
E. (2019). Uncertainty in ensembles of global biodiversity scenar-
ios. Nature Communications, 10(1), 1446. https ://d oi.or g/10 .103 8/
s41467‐019‐0 9519‐w
Ti ch y, L., & Ch ytr y, M. (2 006 ). Stat is tical det erm ina ti on of dia gno sti c sp e-
cies for site groups of unequal size. Journal of Vegetation Science, 17,
809–818 . ht tps ://doi.o rg /10.1111 /j.1654‐1103 .2 006.tb 02 5 04.x
Titeux, N., Dufrene, M., Jacob, J., Paquay, M., & D efourny, P. (2004).
Multivariate analysis of a fine‐scale breeding bird atlas using a geo-
graphical information system and partial canonical correspondence
analysis: environmental and spatial effects. Journal of Biogeography,
31(11), 1841–1856 .
Urban, M. C. (2015). Accelerating extinction risk from climate change.
Science, 348(6234), 571–573.
Walter, H. (1971). Ecology of tro pical and subtro pical vegetation. Edinburgh,
UK: Oliver & Boyd.
Werneck, F. P. (2011). The diversification of eastern South American
open vegetation biomes: Historical biogeogr aphy and perspec-
tives. Quaternary Science Reviews, 30, 16 30–1648. https ://doi.
org/10.1016/j.quasc irev.2011.03.009
Werneck, F. P., Nogueira, C., Colli, G. R., Sites, J. W., & Costa, G. C. (2012).
Climatic stability in the Brazilian Cerrado: Implications for biogeo-
graphical connections of South American savannas, species richness
and conservation in a biodiversity hotspot. Journal of Biogeography,
39, 1695–1706. https ://doi.org /10.1111/j.1365‐2699.2012.02715.x
Williams, K. J., Ford, A., Rosauer, D. F., De Silva, N., Mittermeier, R.,
Bruce, C., … Margules, C. (2011). Forests of East Australia: The 35th
biodiversity hotspot. In Biodiversity hotspots (pp. 295–310). Springer,
Berlin, Heidelberg. https ://doi.org/10.1007/978‐3‐642‐20992‐5_16
Zuur, A. F., Ieno, E. N., Walker, N., Saveliev, A. A., & Smith, G. M.
(2009). Mixed effects models and extensions in ecology with R.
New York: Springer Science & Busine ss Media. https ://doi.
org/10.10 07/978‐0‐387‐87458‐6
Zwiener, V. P., Padial, A. A., Marques, M. C . M., Faleiro, F. V., Loyola, R .,
& Peterson , A. T. (2017). Planning for conser vation and restor ation
under climate and land use change in the Brazilian Atlantic Forest.
Diversity and Distributions, 23(8), 955–966. https ://doi.org/10.1111/
ddi.1258 8
BIOSKETCHES
Luíz Fernando Esser is a PhD student in Botany at the Federal
University of Rio Grande do Sul, Brazil. He is interested in biodi-
versity evolution and environmental niche evolution, focusing on
what influences the establishment of plant communities in space
and time. Danilo M. Neves is a professor of macroecolog y at the
Federal Universit y of Minas Gerais, Brazil. He is interested in the
evolutionary dimension of community ecology, with an empha-
sis on historical biogeography of terrestrial biomes. João And
Jarenkow is a research professor at the Federal University of Rio
Grande do Sul. His main research interests include phytogeogra-
phy and plant community ecology of southern Brazilian Atlantic
rain forests.
Author contributions: L .F.E. conceived the ideas, applied the
methodology and led the writing; D.M.N. contributed for the im-
provement of writing and the methods; J.A .J. improved the writ-
ing and supervised the first author.
SUPPORTING INFORMATION
Additional supporting information may be found online in the
Suppor ting Information section at the end of the article.
How to cite this article: Esser LF, Neves D, Jarenkow JA.
Habitat‐specific impacts of climate change in the Mata
Atlântica biodiversity hotspot. Divers Distrib. 2019;25:1846–
1856. htt ps ://doi.or g/10.1111/ddi .12984
... While many assessments still generate single projections from a single SDM with a single emission scenario, there is growing recognition of the need to use ensembles of projections from multiple SDMs, global circulation models (GCMs), and climate change scenarios (representative concentration pathways, RCPs) to explicitly report the uncertainties in forecasts (Araújo & New, 2007;Diniz-Filho et al., 2009;Thuiller et al., 2019). With the development of new SDM algorithms and a new generation of GCMs and RCPs within an ensemble modeling framework, we are now able to address analytical uncertainties and provide more accurate predictions for conservation planning (Pereira et al., 2010;Dawson et al., 2011;Esser et al., 2019). ...
... Nevertheless, most such assessments have been conducted in European mountain systems. In Brazil, few assessments have examined the impact of climate change on highland vegetation types (Bergamin et al., 2019;Esser et al., 2019) even though South America is likely to face the highest species extinction rates due to warming (Urban, 2015). ...
... Ecological indicators of habitat types or environmental conditions (i.e., indicator or diagnostic species) have been used successfully for modeling the distribution of vegetation types (Bueno et al., 2016;Esser et al., 2019). Here, we used a non-exhaustive set of 18 indicator tree species as a proxy to model the current and potential distribution of Araucaria moist forests (Appendix S1). ...
Article
Full-text available
Aim: Araucaria moist forests are particularly vulnerable to climate change due to their strict climatic requirements and patchy distribution. Therefore, identifying areas where these forests are expected to lose or retain climatically suitable space (i.e., climate change refugia) is urgently required. Here, we modeled the current and future climatic suitability for the Araucaria moist forests aiming (a) to identify areas of suitable climate (i.e., in situ and ex situ refugia), (b) to identify areas of climate retraction, and (c) to assess the effectiveness of protected areas to capture climatically suitable space. Location: Araucaria moist forests ecoregion, southern Brazil, and northeast Argentina. Methods: We mapped the potential distribution of the Araucaria moist forests using an ensemble forecasting approach with 18 indicator tree species (all wet- and cold-adapted taxa), six algorithms, eight global circulation models, three representative concentration pathways (RCPs 4.5, 6.0, and 8.5), and three periods (current, 2050, and 2070). Results: We predicted substantial losses of future climatic suitability across almost the entire range where these forests occur, ranging from 43 to 64% under optimistic (RCP4.5/2050) and high emissions (RCP8.5/2070) scenarios. We found that the protected areas network captured only 3% of the climatically suitable space under current conditions. We found that only 4% (top 1% of cells) and 12% (top 5% of cells) of the potential refugia would be protected in the future, with less than half of their areas corresponding to in situ refugia. Conclusions: Projected losses of potential distribution and the low efficacy of protected areas to buffer climate-change impacts point to a high-risk scenario for the Araucaria moist forests in the near future. Cold-adapted tree species likely will face increased extinction risk, especially as climate change will interact with other anthropogenic drivers.
... In addition, more than 70% of protected areas in the Atlantic Forest belong to the sustainable use category, with few restrictions on human activities (Pacheco et al. 2018). This has resulted in reduced efficiency for maintaining local biodiversity and an elevated and ongoing loss of biomass, suitable habitats and taxonomic and functional diversity, processes that are expected to intensify in conjunction with climate change (Esser & Neves 2019, Lima et al. 2020. However, protected areas in Brazil and their surrounding areas are four times less vulnerable to vegetation conversion than unprotected ones (Gonçalves-Souza et al. 2021). ...
... The impact of climate change on the loss of suitable forest habitats has been estimated to be less pronounced in rainforests and high-elevation/high-latitude forests (Esser & Neves 2019), where conservation status is comparatively better. With regards to Brazilian coastal vegetation, biotic homogenization may be a pervasive process caused by changes in the distribution of taxonomic and functional diversity (Inague et al. 2021). ...
Article
Many biomes still lack an overall view of their macro-functional structure (i.e., natural biogeographical regions and zones), including the Atlantic Forest biodiversity hotspot. The effective design of protected areas depends on the spatial identification of units with ecologically distinct content, whether it be floristic, phylogenetic or functional. This study used a regionalization approach to identify the potential functional regions of the Atlantic Forest by interpolating functional data from forest remnants into the entire original occurrence area of the biome, including deforested lands. Conservation status was then estimated. Analysis of seven traits of leaf, wood, seed and plant size revealed that the biome is structured over 14 functional regions and three zones (clusters of regions). Functional regions represented specific combinations of traits rather than being characterized by extremely high or low values of a single trait. They retained an average of 29.5% of forest remnants (range: 7.63–54.66%) and 10.82% of protected areas (range: 0.35–35.78%). By analysing the functional space occupied by all regions, captured by two principal component analysis axes using the pixel-level information contained in interpolated trait maps, we showed that large parts of this space were not covered by forest remnants or protected areas and that the most represented regions had serious deficits in protected areas. Although the Serra do Mar mountain range in the south and south-east Atlantic Forest is relevant as a centre of species endemism and richness and has received considerable attention for carrying out ecological studies and creating protection areas, this range does not fully encompass the functional biodiversity of such a rich biome. Our results demonstrate the potential for combining regionalization and conservation approaches to unravel the macro-structures of biomes.
... Current geographic distributions certainly play a role in a species response to climate change and an understanding of the limitations of species will help predict how easily that species may disperse into new territory (Peterman & Semlitsch, 2014). Furthermore, projecting the environmental envelope of a species onto future climate models has been used to assess the available habitat and dispersal corridors under climate change for conservation purposes (Esser et al., 2019;Zellmer et al., 2020;Zhang et al., 2019). If a species suitable habitat shrinks with changing conditions, as we expect to be the case in salamanders, they may be at higher risk of extinction (Thomas et al., 2004). ...
... Despite these limitations, such modeling efforts are beneficial in order to determine the validity of various predictions in the future (Esser et al., 2019;Zhang et al., 2019). Finally, in order to computationally assess a species' niche with fidelity, many of the aforementioned factors could be used together. ...
Article
Full-text available
Aim Given that salamanders have experienced large shifts in their distributions over time, we determined how each species of Plethodon in the Pacific Northwest would respond to climate change. We incorporated several greenhouse scenarios both on a species-by-species basis, and also using phylogenetic groups, with the aim to determine the best course of action in managing land area to conserve diversity in this group. Location Pacific Northwest of the United States (northern CA, OR, WA, ID, and MT). Major taxa studied Western Plethodon salamanders. Methods Species distribution models were estimated using MaxEnt for the current time period and for several future climate scenarios using bioclimatic data layers. We used several methods to quantify the change in habitat suitability over time from the models. We explored aspects of the climate layers to determine whether we can expect a concerted response to climate change due to similarity in ecological niche or independent responses that could be harder to manage. Results The distribution of western Plethodon salamander species is strongly influenced by precipitation and less so by temperature. Species responses to climate change resulted in both increases and decreases in predicted suitable habitat, though most species ranges do not contract, especially when taken as a phylogenetic group. Main conclusions While some established habitats may become more or less climatically suitable, the overall distribution of species in this group is unlikely to be significantly affected. Clades of Plethodon species are unlikely to be in danger of extirpation despite the possibility that individual species may be threatened as a result of limited distributions. Grouping species into lineages with similar geographic ranges can be a viable method of determining conservation needs. More biotic and dispersal information is needed to determine the true impact that changes in climate will have on the distribution of Plethodon species.
... With the advent of species distribution models and ecological niche modeling, researchers have been able to make predictions as to how changing conditions, often in the context of climate change, may affect atrisk species in the future (Walls et al. 2013). While a plethora of studies have used modeling methods to predict declines of species in biodiversity hotspots (Malcolm et al. 2006, Fitzpatrick et al. 2008, Esser et al. 2019, there is potentially more value in assessing the vulnerability of widely distributed, common species to climate change because abundance of common species, not species richness, drives ecosystem service delivery and common species often decline rapidly in response to changing environmental conditions (Lindenmayer et al. 2011, Winfree et al. 2015. Additionally, species that are relatively common and exhibit wide geographic ranges are seldom considered at-risk or high priority compared to rare or geographically limited taxa (Gaston 2010). ...
Article
The sensitivity of amphibian species to shifts in environmental conditions has been exhibited through long‐term population studies and the projection of ecological niche models under expected conditions. Species in biodiversity hotspots have been the focus of ample predictive modeling studies, while, despite their significant ecological value, wide‐ranging and common taxa have received less attention. We focused on predicting range restriction of the spotted salamander (Ambystoma maculatum), blue‐spotted salamander (A. laterale), four‐toed salamander (Hemidactylium scutatum), and red‐backed salamander (Plethodon cinereus) under future climate scenarios. Using bias‐corrected future climate data and biodiversity database records, we developed maximum entropy (MaxEnt) models under current conditions and for climate change projections in 2050 and 2070. We calculated positivity rates of species localities to represent proportions of habitat expected to remain climatically suitable with continued climate change. Models projected under future conditions predicted average positivity rates of 91% (89–93%) for the blue‐spotted salamander, 23% (2–41%) for the spotted salamander, 4% (0.7–9%) for the four‐toed salamander, and 61% (42–76%) for the red‐backed salamander. Range restriction increased with time and greenhouse gas concentration for the spotted salamander, four‐toed salamander, and red‐backed salamander. Common, widespread taxa that often receive less conservation resources than other species are at risk of experiencing significant losses to their climatic ranges as climate change continues. Efforts to maintain populations of species should be focused on regions expected to experience fewer climatic shifts such as the interior and northern zones of species' distributions. We modeled the current and future climatic range distributions of 4 widely occurring North American salamander species (spotted salamander, blue‐spotted salamander, four‐toed salamander, and red‐backed salamander) in eastern Canada and the United States. We compared contemporary models with those projected under various climate change scenarios to estimate the loss or addition of climatically suitable habitat in 2050 and 2070.
... Climate projections point to a greater risk of extinction and modifications to the spatial distribution of several species (Esser et al., 2019;Joly et al., 2014). However, there is still uncertainty about how tropical tree species and the ecosystems they live in will respond to climate change (Corlett & Westcott, 2013). ...
Article
Climate change will affect the distribution of many tropical plant species. However, the understanding of how dioecious tropical species cope with different environmental conditions is still limited. To address this issue, we investigated how secondary trait attributes in populations of the dioecious tropical tree Myrsine coriacea change along an altitudinal gradient. Eighty individual plants (40 male and 40 female) were selected among seven natural populations. Leaf variation in morphological and stomatal traits, and carbon and nitrogen isotopic compositions were analyzed. Female plants had greater isotopic leaf carbon composition (δ13C) and nitrogen content than male plants, increasing their carboxylation capacity. Plants of both sexes had smaller stomata, greater water‐use efficiency (greater δ13C), and greater nitrogen isotopic composition (δ15N) at higher altitudes. They also showed lower δ15N and had greater carbon:nitrogen ratios at lower altitudes. There was a lack of coordination between stomatal and vein traits, which was compensated for by variation in specific leaf area. This mechanism was essential for increasing plant performance under the limiting conditions found by the species at higher altitudes. This article is protected by copyright. All rights reserved.
... fernambucensis) occurred during interglacial periods(Franco, Silva, et al., 2017). Examples of floristic interconnections between rock outcrops in Cerrado and inselbergs/Restinga forest have been proposed previously for plant species(Antonelli et al., 2010;Esser et al., 2019) and our study adds empirical evidence for this historical floristic interchange. ...
Article
Full-text available
The interconnectedness and biotic interchange among Neotropical biomes are thought to play an important role in driving adaptation and diversification. However, how these processes are in synteny to trait evolution in species of open and xeric areas is poorly studied. Here, we investigate the spatial and temporal dimensions of evolution and candidate traits associated with biome shifts in xeric vegetation, focusing on the family Cactaceae. Xeric and open areas of South America. Genus Cereus Mill. (Cactaceae, Cereeae). We applied biogeographical reconstructions on a time‐calibrated phylogeny inferred from multilocus data (ddRAD‐Seq) using Bayesian analyses on BEAST2, species distribution modelling in Maxent, the reconstruction of biome affinities and niche shift analyses based on abiotic traits (climate and soil) using Mk‐model in BioGeoBEARS, and phenotypic trait‐based analysis in Mesquite. The Cerrado domain is the ancestral area of Cereus, with most diversification events occurring in a time of intense orogenesis, climatic changes, and marine regressions within the last 5 Mya. Events of biome transition from the seasonally dry tropical forest (SDTF) were also associated with trait and niche shifts. The diversification of the xerophyte genus Cereus is associated with the climatic and geomorphological instabilities of the Pliocene and Pleistocene epochs. The Cerrado domain states an important region of dispersal for the genus. Some geographical range movements involved biome shifts associated with niche evolution while others were restricted to a simple biogeographical transition without niche change. Particular clades that experienced biome shifts displayed some phenotypic state changes, suggesting a role of biotic traits for environment transition. The results observed in Cereus may be a biogeographical pattern that should be tested with other cactus species, such as Pilosocereus spp., or species of xeric habitats, such as Annonaceae and Vochysiaceae.
... Isto porque o isolamento das populações leva à depreciação do fluxo gênico, aumento da endogamia e perda da heterozigose, favorecendo a ocorrência da extinção local ou até mesmo regional (Porto et al, 2018;Cruz et al, 2018). Tais consequências são ainda mais preocupantes quando ocorrem em ambientes que apresentam fragilidade ambiental, ou seja, quando os recursos naturais têm sido explorados durante séculos, incluindo, por exemplo, a ocupação antrópica na região da Mata Atlântica, o que proporcionou modificações na paisagem em larga escala e fragmentação florestal (Santos et al, 2020;Laurance, 2009;Ribeiro et al, 2009).Os altos níveis de fragmentação ainda observados atualmente, assim como, a contínua perda de habitat em toda a Mata Atlântica, têm gerado várias preocupações na comunidade científica (Joly et al, 2014;Esser et al, 2019). ...
Article
Full-text available
This study aimed to describe and compare the floristic and structural component of the tree component, as well as the alpha and beta diversity, as well as the floristic similarity of a remnant of Semideciduous Forest. The tree vegetation was sampled at two edges in contact with pasture (BP) and coffee growing (BC), and inside the fragment (INT). 2.840 individuals were sampled, identified in 56 families, 144 genera and 271 species (94 BC, 128 BP and 178 INT). The border stretches were characterized by higher density and smaller basal area than the interior, indicating the occurrence of disturbances. The variations in the structural floristic composition of the tree communities wereinfluenced by regional and local environmental variations, as well as the historical use of the area, according to Niche and Intermediate Disturbance Theories.
... However, this response was spatially heterogeneous with rainforests suffering more losses than dry forests and the dry forest PSR was more stable under a moderate climate change scenario. Esser et al. (2018) modeling distinct climate change scenarios for the Atlantic Forest also observed a spatially heterogeneous response, although with a distinct pattern where there was the occurrence of relatively stronger potential loss of suitable environment for semideciduous forest (Seasonally Dry Atlantic Forest), with semideciduous and rainforest species showing a lower degree of overlap in climate adequacy (6.7% in the current climate), which decreases with the climate change scenario (1.2% in Representative Concentration Pathway-RCP8.5/2070). Tropical dry forests are not necessarily more resilient than tropical rainforests, but they may be more resistant to specific disturbances such as fire and drought (Pulla et al. 2015). ...
Article
Full-text available
Key message For better categorization of species according to foliar habit, a set of leaf and wood traits must be observed. Abstract Tropical forests are influenced by distinct regional rainfall regimes, microclimates, and dynamics of nutrient cycling, which are responsible for creating key biodiversity patterns and differences in leaf deciduousness to drought. Functional traits studies have improved understanding of the functioning and heterogeneity of complex ecosystems. We have reviewed the literature focusing mainly on tropical dry forests and relationships among leaf habits (evergreen and deciduous) and other leaf and wood traits. Thus, we have compiled 121 original papers, 2 reports, and 9 book chapters published since 2000. We also provide a meta-analysis of these traits from Neotropics. Tropical deciduous species often have high photosynthetic rates per mass and specific leaf area and traits that improve water flow throughout the plant, such as wide xylem diameters and high hydraulic conductivity, maximizing resource capture during a limited growing season because of an acquisition strategy. The opposite is observed in evergreen species, namely as conservative species. Regardless of the plant organ, more morphological than physiological traits are available to compare leaf habits. For better categorization of species according to foliar habit, a set of leaf and wood traits must be observed. However, while local comparisons based on one or few traits may group species according to leaf habit, multivariate analyses for large spatial scales can reveal a different pattern. We have identified some open questions that can be further addressed in this research field to contribute to the improvement of theoretical frameworks as well as the consequences of a changing climate for tropical dry forests.
... Williams et al., 2004), past novel or no-analogue assemblages have been documented from Amazonia to south-eastern Brazil, generally characterised by the co-occurrence of cold-and warmadapted pollen taxa during (Late) Glacial times (Behling, 1998;Bush et al., 2004;Colinvaux et al., 1997Colinvaux et al., , 2000De Oliveira, 1992;De Oliveira et al., 2020;Francisquini et al., 2020;Haberle and Maslin, 1999;Hermanowski et al., 2012;Hor ak-Terra et al., 2020;Lima et al., 2018;Raczka et al., 2013;Whitney et al., 2011). Southern Brazil's forests are characterised by floristic gradients e each contains significant compositional diversity, and differences between forest types are generally marked most by gradual species turnover (Bergamin et al., 2017;Brown et al., 2020;Duarte et al., 2014;Esser et al., 2019;Oliveira-Filho et al., 2014;Oliveira-Filho and Fontes, 2000). These characteristics mean there is a significant chance that any past or future reassembly of these communities would be poorly captured by modern-day ecosystem classifications. ...
Article
Brazil's Atlantic Forest biome is one of the world's biodiversity hotspots, whose heterogeneous ecosystems are threatened by habitat loss and climate change. Palaeoecological research can provide essential context for the impacts of anthropogenic climate change in the 21st Century and beyond, but existing studies have notable limitations in the insights they can provide: vegetation proxy data are spatially and temporally skewed with inconsistent taxonomic resolution; existing modelling studies typically overlook individualistic species-level responses, are limited in temporal coverage, and lack close integration with empirical palaeoecological data. Here, we investigate the impact of major climate changes upon the species-level floristic composition of southern Brazil's Atlantic Forest, from the Last Glacial Maximum (LGM) to the late 21st century, by modelling the distributions of 30 key species at seven time slices since the LGM and comparing the assemblages they form with an unprecedented dataset of palaeoecological proxy data. We find notable compositional changes through time across our study area, especially during the early Holocene, which was characterised by extensive no-analogue plant communities. Aspects of these modelled floristic changes are captured in proxy records but many occur in data-sparse regions, highlighting geographic foci for future palaeoecological investigation to test these model predictions. Our findings highlight the individualistic responses of Atlantic Forest plant species to climate change and help resolve long-standing palaeoecological questions – explaining the dominance of highland grasslands at the Last Glacial Maximum (likely due to low atmospheric CO2 concentrations), clarifying the LGM extent of coastal tropical forest (probably in a grassland matrix on exposed continental shelf), and explaining the origins of Araucaria angustifolia's western populations (from climatic (micro-)refugia rather than human-mediated dispersal). Our results also set the 21st Century's impending climate and vegetation changes in a 21,000-year temporal context, revealing that, under a high emissions scenario, more than 100,000 km² of the southern Atlantic Forest will experience more climate-driven floristic change in the coming decades than it has in the last 21 millennia.
... These studies typically replicate altitudinal gradient studies at the community level carried decades ago, revealing upward range shifts and contraction (e.g., Forero-Medina et al. 2011), and could be carried out in the Atlantic Forest. The lack of studies on observed or predicted climate change impacts on Atlantic Forest freshwater ecosystems is also worrisome, given their high diversity and vulnerability (Collen et al. 2013;Roland et al. 2012; but see Esser et al. 2019). Finally, the Atlantic Forest has many associated coastal ecosystems, such as restingas and mangroves, which are also vulnerable to climate change, especially sea-level rise, but there is blatant lack of studies on the topic (Godoy and Lacerda 2015;Oliveira et al. 2016;Copertino et al. 2010). ...
Chapter
Ongoing anthropogenic climate change is becoming one of the major threats to biodiversity. Studies that aim at projecting the future impacts of ongoing climate change on biodiversity should use general circulation models (GCMs) that show a good performance in the region of study, an information that is lacking for the Atlantic Forest. Here, we evaluated the performance of different GCMs over the Atlantic Forest, describe the predicted climatic changes for the regions based on the best performing GCMs, review the literature on observed and predicted impacts of climate change on the Atlantic Forest biodiversity, and discuss adaptation strategies to reduce the negative impacts of climate change on the region’s biodiversity.
Article
Full-text available
While there is a clear demand for scenarios that provide alternative states in biodiversity with respect to future emissions, a thorough analysis and communication of the associated uncertainties is still missing. Here, we modelled the global distribution of ~11,500 amphibian, bird and mammal species and project their climatic suitability into the time horizon 2050 and 2070, while varying the input data used. By this, we explore the uncertainties originating from selecting species distribution models (SDMs), dispersal strategies, global circulation models (GCMs), and representative concentration pathways (RCPs). We demonstrate the overwhelming influence of SDMs and RCPs on future biodiversity projections, followed by dispersal strategies and GCMs. The relative importance of each component varies in space but also with the selected sensitivity metrics and with species’ range size. Overall, this means using multiple SDMs, RCPs, dispersal assumptions and GCMs is a necessity in any biodiversity scenario assessment, to explicitly report associated uncertainties.
Article
Full-text available
Increasingly complex research questions and global challenges (e.g., climate change and biodiversity loss) are driving rapid development, refinement, and uses of technology in ecology. This trend is spawning a distinct sub‐discipline, here termed “technoecology.” We highlight recent ground‐breaking and transformative technological advances for studying species and environments: bio‐batteries, low‐power and long‐range telemetry, the Internet of things, swarm theory, 3D printing, mapping molecular movement, and low‐power computers. These technologies have the potential to revolutionize ecology by providing “next‐generation” ecological data, particularly when integrated with each other, and in doing so could be applied to address a diverse range of requirements (e.g., pest and wildlife management, informing environmental policy and decision making). Critical to technoecology's rate of advancement and uptake by ecologists and environmental managers will be fostering increased interdisciplinary collaboration. Ideally, such partnerships will span the conception, implementation, and enhancement phases of ideas, bridging the university, public, and private sectors.
Article
Full-text available
Compositional changes in Himalayan vegetation in response to the major drivers of biodiversity loss, climate change and land-use change, are barely documented. We quantify temporal changes in the alpine vegetation of central Nepal and attribute these changes to temporally varying climatic and landuse factors. We re-surveyed the alpine vegetation of two locations within Langtang National Park, central Nepal, after 25 yr using 127 plots of 100 m2. Using ordination, regression, and weighted average regression and calibration techniques, we analyzed the changes in terms of species abundance, frequency, and elevational shift in relation to changing atmospheric temperature, precipitation, and livestock grazing. We found a significant increase in the frequency and relative abundance of the majority of species, which was significantly related to the temporal trends in climatic factors and grazing intensity. Out of 12 species with unimodal responses along the elevation gradient during both surveys, the optima of eight species decreased over the time period. The observed elevations of 62 out of 92 sample plots (hence, species composition) in 2014 were lower than the elevations calibrated from species composition and elevation of 1990, indicating an overall downward shift of species assemblages. However, an upward shift of assemblages was also observed at higher elevations. These results indicate that the observed temporal changes in alpine vegetation, largely contrasting the expected upslope shift of species due to climate warming, are driven most likely by interactions of contemporary climate and land-use changes, especially reduced grazing. The complex interactions and feedback mechanisms between warmer winters, increased precipitation, reduced grazing pressure, and thereby altered species interactions most likely facilitated the downslope shift of alpine species assemblages. Climatic and land-use responses of plant species assemblages should therefore be studied focusing on the potential interactions between both the climatic and the land-use factors because such interactions and feedback mechanisms have potential to mask or modify the expected climatic or land-use response of biodiversity.
Article
Full-text available
Climate change triggers poleward shifts in species distribution leading to changes in biogeography. In the marine environment, fish respond quickly to warming, causing community-wide reorganizations, which result in profound changes in ecosystem functioning. Functional biogeography provides a framework to address how ecosystem functioning may be affected by climate change over large spatial scales. However, there are few studies on functional biogeography in the marine environment, and none in the Arctic, where climate-driven changes are most rapid and extensive. We investigated the impact of climate warming on the functional biogeography of the Barents Sea, which is characterized by a sharp zoogeographic divide separating boreal from Arctic species. Our unique dataset covered 52 fish species, 15 functional traits, and 3,660 stations sampled during the recent warming period. We found that the functional traits characterizing Arctic fish communities, mainly composed of small-sized bottom-dwelling benthivores, are being rapidly replaced by traits of incoming boreal species, particularly the larger, longer lived, and more piscivorous species. The changes in functional traits detected in the Arctic can be predicted based on the characteristics of species expected to undergo quick poleward shifts in response to warming. These are the large, generalist, motile species, such as cod and haddock. We show how functional biogeography can provide important insights into the relationship between species composition, diversity, ecosystem functioning, and environmental drivers. This represents invaluable knowledge in a period when communities and ecosystems experience rapid climate-driven changes across biogeographical regions.
Article
Full-text available
Aim: To propose and compare priority sites for conservation and restoration of woody plants under diverse climate and land use scenarios, considering socio-economic costs, presence of protected areas and distribution of forest remnants. Location: The Atlantic Forest Biodiversity Hotspot, Brazil. Methods: We used ecological niche modelling to estimate geographical distributions for 2,255 species under current and future climate scenarios, which we analysed in relation to spatially explicit land use projections, maps of forest remnants derived from remote sensing and socio-economic variables for each municipality within the Atlantic Forest region. We identified spatial priorities that complement the current network of protected areas under three different prioritization scenarios: (1) conservation of existing forest remnants only; (2) conservation of remnants followed by restoration of degraded habitat; and (3) unconstrained actions, in which management location is not defined a priori. We compared our results under different levels of land protection, with targets of 10%, 17% and 20% of the Atlantic Forest extent. Results: Current forest remnants cover only 12% of the Atlantic Forest, so targets of 17% and 20% were achieved only through active restoration. Targets of 17% and 20% captured most species and represented on average 26%-34% of species' distributions. The spatial pattern of degraded habitats negatively affected representation of biodiversity and implied higher costs and reduced efficiency of planning. We did not observe major differences between conservation prioritizations based on contrasting climate change scenarios. Main conclusions: Protection of forest remnants alone will not suffice to safeguard woody plant species under climate and land use changes; therefore, restoration actions are urgently needed in the Atlantic Forest. With integrated management actions and multicriterion nationwide planning, reaching the 17% of land protection of Aichi biodiversity targets will constitute an important step towards protecting Atlantic Forest biodiversity.
Article
Full-text available
Aim: We aimed to assess the contribution of marginal habitats to the tree species richness of the Mata Atlântica (Atlantic Forest) biodiversity hotspot. In addition, we aimed to determine which environmental factors drive the occurrence and distribution of these marginal habitats. Location: The whole extension of the South American Atlantic Forest Domain plus forest intrusions into the neighbouring Cerrado and Pampa Domains, which comprises rain forests (‘core’ habitat) and five marginal habitats, namely high elevation forests, rock outcrop dwarf-forests, riverine forests, semideciduous forests and restinga (coastal white-sand woodlands). Methods: We compiled a dataset containing 366,875 occurrence records of 4,431 tree species from 1,753 site-checklists, which were a priori classified into ten main vegetation types. We then performed ordination analyses of the species-by-site matrix to assess the floristic consistency of this classification. In order to assess the relative contribution of environmental predictors to the community turnover, we produced models using 26 climate and substrate-related variables as environmental predictors. Results: Ordination diagrams supported the floristic segregation of vegetation types, with those considered as marginal habitats placed at the extremes of ordination axes. These marginal habitats are associated with the harshest extremes of five limiting factors: temperature seasonality (high elevation and subtropical riverine forests), flammability (rock outcrop dwarf-forests), high salinity (restinga), water deficit severity (semideciduous forests) and waterlogged soils (tropical riverine forests). Importantly, 45% of all species endemic to the Atlantic Domain only occur in marginal habitats. Main conclusions: Our results showed the key role of the poorly protected marginal habitats in contributing to the high species richness of the Atlantic Domain. Various types of environmental harshness operate as environmental filters determining the distribution of the Atlantic Domain habitats. Our findings also stressed the importance of fire, a previously neglected environmental factor.
Article
Full-text available
The rocky montane savannas of South America, known as campos rupestres in Brazil, where they largely occur, represent a megadiverse habitat housing c.15% of the Brazilian vascular flora in less than 1% of the Brazilian territory. Amongst other factors, the remarkable plant diversity in campos rupestres has been attributed to its occurrence as many isolated patches and to floristic influences from surrounding habitats, including lowland woody savannas (cerrado), Atlantic rain forests, seasonally dry woodlands and Amazonian rain forests. However, no study has assessed the degree to which the putative floristic influence from surrounding habitats drives compositional variation in campos rupestres. Here, we used a dataset on the composition of South American woody plant communities (>4,000 community surveys, with >100 representing campos rupestres), combined with environmental data, with the aim of characterising and explaining compositional variation of the campos rupestres woody flora. Our results showed that all campos rupestres, including the sites occurring in Amazonian ironstone formations, are more similar to cerrado woody savannas than to any other South American vegetation formations covered in our dataset. Also, multiple campo rupestre floristic groups may be recognized based on distinct species composition and environmental conditions, primarily related to substrate and climate. We stress the importance of considering this floristic heterogeneity in conservation, management and research planning.
Article
Recent developments in geographic information systems and their application to conservation biology open doors to exciting new synthetic analyses. Exploration of these possibilities, however, is limited by the quality of information available: most biodiversity data are incomplete and characterized by biased sampling. Inferential procedures that provide robust and reliable predictions of species' geographic distributions thus become critical to biodiversity analyses. In this contribution, models of species' ecological niches are developed using an artificial-intelligence algorithm, and projected onto geography to predict species' distributions. To test the validity of this approach, I used North American Breeding Bird Survey data, with large sample sizes for many species. I omitted randomly selected states from model building, and tested models using the omitted states. For the 34 species tested, all predictions were highly statistically significant (all P < 0.001), indicating excellent predictive ability. This inferential capacity opens doors to many synthetic analyses based on primary point occurrence data. Predicción de Áreas de Distribución de Especies con Pase en Modelaje de Nichos Ecológicos Resumen. Avances recientes en los sistemas de información geográfica y su aplicación en la biología de conservación presentan la posibilidad de analisis nuevos y sintéticos. La exploración de estas posibilidades, de todas formas, se limita por la calidad de información disponible: la gran mayoria de datos respecto a la diversidad biológica son incompletos y sesgados. Por eso, procedimientos de inferencia que proveen predicciones robustas y confiables de distribuciones de especies se hacen importantes para los análisis de la biodiversidad. En esta contribución, se desarrollan modelos de los nichos ecológicos por medio de un algoritmo de inteligencia artificial, y los proyeccionamos en la geografía para predecir las distribuciones geográficas de especies. Para probar el método, se usan los datos del North American Breeding Bird Survey, con tamaños de muestra grande. Se construyeron modelos con base en 30 estados unidenses seleccionados al azar, y se probaron los modelos con base en los 20 estados restantes. De las 34 especies que se analizaron, todos mostraron un alto grado de significanza estadística (todos P < 0.001), lo cual indica un alto grado de predictividad. Esta capacidad de inferencia abre la puerta a varios analisis sintéticos con base en puntos conocidos de ocurrencia de especies.
Article
We assessed the effectiveness of a protected area (PA) network in representing tree taxonomic and phylogenetic uniqueness in subtropical Atlantic Forests (Rain, Mixed and Seasonal Forests). Using a network of plots distributed over ∼95,000 km2 in southern Brazil, we first map the distribution of species richness (SR), beta diversity (BD) and phylogenetic diversity (PD) across the extent of remaining forest in the region. We then tested whether areas of taxonomic and phylogenetic uniqueness are either over- or under-protected based on the existing PA network (3% coverage) and at least 10% coverage, and assessed whether protection is distributed equally for each uniqueness area type. Here, areas of taxonomic uniqueness were defined as those with higher contribution than the mean to the total BD, and areas of phylogenetic uniqueness as those with higher or lower PD than expected by chance given their SR, and sites exhibiting spatial congruence or mismatch between PD and SR. We found a high percentage of representation of both areas of uniqueness across the extent of remaining forest. However, our analyses showed that these areas are poorly and unequally captured by the PA network; they are on average less protected than expected based on at least 10% coverage and have high inequality of protection. Our results suggest that both beta diversity and evolutionary history of angiosperm trees are not adequately protected, and indicate relevant areas to extend the current PA network. We also emphasize the need to consider a multifaceted approach to maximize protection of the Atlantic Forest biodiversity.
Article
The Amazon and Atlantic Forest are considered the world's most biodiverse biomes. Human and climate change impacts are the principal drivers of species loss in both biomes, more severely in the Atlantic Forest. In response to species loss, the main conservation action is the creation of protected areas (PAs). Current knowledge and research on the PA network's conservation efficiency is scarce, and existing studies have mainly considered a past temporal view. In this study, we tested the efficiency of the current PA network to maintain climatically stable areas (CSAs) across the Amazon and Atlantic Forest. To this, we used an ecological niche modeling approach to biome and paleoclimatic simulations. We propose three categories of conservation priority areas for both biomes, considering CSAs, PAs and intact forest remnants. The biomes vary in their respective PA networks' protection efficiency. Regarding protect CSAs, the Amazon PA network is four times more efficient than the Atlantic Forest PA network. New conservation efforts in these two forest biomes require different approaches. We discussed the conservation actions that should be taken in each biome to increase the efficiency of the PA network, considering both the creation and expansion of PAs as well as restoration programs.