ORIGINAL ARTICLE

Modelling the distribution of key tree species used by liontamarins in the Brazilian Atlantic forest under a scenario of futureclimate change

Nima Raghunathan • Louis Francois •

Marie-Claude Huynen • Leonardo C. Oliveira •

Alain Hambuckers

Received: 28 February 2013 / Accepted: 16 April 2014 / Published online: 14 August 2014

� Springer-Verlag Berlin Heidelberg 2014

Abstract We used three IPCC climate change scenarios

(A1B, A2 and B1) in a dynamic vegetation model (CA-

RAIB), to determine the potential future distribution of 75

tree species used by two endemic primate species from the

Brazilian Atlantic Forest (BAF). Habitat conservation is a

vital part of strategies to protect endangered species, and

this is a new approach to understanding how key plant

species needed for survival of golden lion tamarins

(Leontopithecus rosalia) and golden-headed lion tamarins

(L. chrysomelas) might be affected by climate change and

what changes to their distribution are likely. The model

accurately predicted the current distribution of BAF veg-

etation types, for 66 % of the individual tree species with

70 % agreement obtained for presence. In the simulation

experiments for the future, 72 out of 75 tree species

maintained more than 95 % of their original distribution

and all species showed a range expansion. At the biome

level, we note a substantial decrease in the sub-tropical

forest area. There is some fragmentation of the savannah,

which is encroached mostly by tropical seasonal forest.

Where the current distribution shows a large sub-tropical

forest biome, it has been replaced or encroached by tropical

rainforest. The results suggested that the trees may benefit

from an increase in temperature, if and only if soil water

availability is not altered significantly, as was the case with

climate simulations that were used. However, these results

must be coupled with other information to maximise use-

fulness to conservation since BAF is already highly frag-

mented and subject to high anthropic pressure.

Keywords Conservation � Climate change � Modelling �Brazilian Atlantic Forest � Lion tamarins � Dynamic

vegetation model (DVMs)Editor: Wolfgang Cramer.

Electronic supplementary material The online version of thisarticle (doi:10.1007/s10113-014-0625-9) contains supplementarymaterial, which is available to authorized users.

N. Raghunathan (&) � M.-C. Huynen � A. Hambuckers

Behavioural Biology Unit, Department of Biology, Ecology, and

Evolution, University of Liege, Quai Van Beneden, 22,

Liege 4020, Belgium

e-mail: praghunathan@ulg.ac.be

M.-C. Huynen

e-mail: marie-claude.huynen@ulg.ac.be

A. Hambuckers

e-mail: alain.hambuckers@ulg.ac.be

L. Francois

Climate Modelling and Biogeochemical Cycles Unit,

Astrophysics, Geophysics and Oceanography Department,

University of Liege, Allee du 6 Aout, 17, Liege 4000, Belgium

e-mail: louis.francois@ulg.ac.be

L. C. Oliveira

Programa de Pos-Graduacao em Ecologia, Universidade Federal

do Rio de Janeiro, CP. 68020, Rio de Janeiro CEP 21941-590,

Brazil

e-mail: leonardoco@gmail.com

L. C. Oliveira

Programa de Pos-Graduacao em Ecologia e Conservacao da

Biodiversidade, Universidade Estadual de Santa Cruz,

Santa Cruz, Brazil

L. C. Oliveira

Bicho do Mato Instituto de Pesquisa, Belo Horizonte, MG,

Brazil

123

Reg Environ Change (2015) 15:683–693

DOI 10.1007/s10113-014-0625-9

Introduction

Tropical forests all over the world are facing increased risks

of conversion and species extinction. Exacerbating the

problem of deforestation and habitat conversion is climate

change. Studies show that changes in regional climates are

forcing species to adapt their distribution by seeking suit-

able conditions, or altering their phenology, and those that

are unable to react are facing extinction risks (Thuiller et al.

2008; Wright et al. 2009; Lurgi et al. 2012). For more than a

decade, there is convincing evidence of such adaptations

already occurring in the temperate regions and at higher

latitudes (Hughes 2000; Parmesan and Yohe 2003; Root

et al. 2003; Gian-Reto et al. 2005; Walther et al. 2005;

Lenoir et al. 2008, Thomas 2010), but less has been docu-

mented for tropical regions where a particular emphasis has

been put on changes in mountains (Colwell et al. 2008;

Raxworthy et al. 2008; Thomas 2010; Chen et al. 2011).

Among the possible impacts of climate change on the

environment, the 2007 IPCC report suggests that between

20 and 30 % of plant and animal species could be at

increased risk of extinction if global average temperatures

increase by 1.5–2.5 �C (IPCC 2007). To assess changes in

the tropics, several approaches are used, such as detecting

temporal changes in composition of species assemblages

(Gonzalez-Megıas et al. 2008; Bertrand et al. 2011; Chen

et al. 2011), analysing temperature performances of species

(Williams et al. 2008; Berg et al. 2010; Clusella-Trullas

et al. 2011; Huey et al. 2012) or modelling the climatic

range (Thuiller et al. 2008). Modelling the habitat including

biotic interaction of the species is rarely used (Keith et al.

2008; Preston et al. 2008) despite calls for the use of this

approach (Brooker et al. 2007; Van der Putten et al. 2010).

Indeed, one of the ways to think about conservation of

endangered fauna is to understand how its habitat may shift

as climate regimes change—if the major food sources,

sleeping sites, nesting sites or preferred vegetation type are

able to withstand the impacts of climate change, one might

hope that the fauna will follow. The viability of species in

their natural range comes from the fact that satisfactory

levels are reached for all the requirements, including the

biotic environment furnishing food and shelter. This allows

for considering the ecosystem or the habitat as well, rather

than just understanding a single species’ response to climate

change.

This study focussed on the future of the golden lion

tamarins (GLTs, Leontopithecus rosalia) and golden-headed

lion tamarins (GHLTs, L. chrysomelas), two endangered

endemic species (IUCN 2013) of the Brazilian Atlantic

Forest (BAF). The BAF is a biodiversity hotspot (Myers

et al. 2000; Mittermeier et al. 2005). Its mosaic habitats,

proximity to the coastline and extreme fragmentation makes

it an interesting ecosystem to study. Conservation at the

nexus of high endemism, primary, secondary and degraded

habitats, high fragmentation, and agro-forestry systems

could well be the future; understanding how climate change

could impact this region will be very important to ensure the

conservation of the BAF. We studied the growth of vege-

tation types and tree species, which are important resources

for the tamarins, using a process-oriented vegetation model.

Thus, the vegetation types and the tree species form the

biotic environment of the tamarins. Otherwise, it is difficult

to evaluate the importance of changes in climatic variables

on the biology of the tamarins themselves, because their

distribution is very narrow and may not be limited by cli-

matic constraints alone. While the number of tree species

listed is not exhaustive, this study is, to our knowledge, the

first of its kind to consider climate change impacts on so

many vegetation species (75) at one time.

To model tree species and vegetation type response, we

used the CARAIB (Francois et al. 1998, 2006) dynamic

vegetation model (DVM). DVMs have a long history of

integrating climate model results to understand plants’

response to climate change. DVMs are comprehensive

models, which include vegetation dynamics as well as bio-

geochemical processes (Cramer et al. 2004). DVMs grow

objects representing the plants as plant functional types

(PFTs, Violle et al. 2007), biological affinity groups (BAGs,

Laurent et al. 2004) or individual plant species. PFTs are

defined according to plant life forms (tree, shrub, herb/grass,

etc.) and morpho-physiological and phenological traits

(broadleaves-needle leaves, deciduous, evergreen, etc.)

while BAGs are defined from plant species with similar bio-

climatic and structural characteristics. Tolerances and

threshold values of each type of object to be grown are

inferred from their actual distribution. Here, the CARAIB

model was used to simulate and to compare the growth of

preferred tree species and the growth of the vegetation types

as PFTs under present climate and CO2 conditions and under

future standard scenarios for the period 2071–2100.

Methods

Tree species geographic and climatic distributions

A literature review (Lapenta et al. 2003; Catenacci 2008;

Oliveira et al. 2010) of commonly foraged tree species and

commonly used species for sleeping sites was conducted

for two out of the four lion tamarin species (L. rosalia—

GLTs and L. chrysomelas—GHLTs). A sample of their

distribution was obtained from the Tropicos online data-

base (www.tropicos.com). For some endemic species,

coordinates were obtained from an on-going study—groups

of L. chrysomelas were followed from sleeping site to

sleeping site, and the geographic location of all trees that

684 N. Raghunathan et al.

123

were foraged upon, and/or used as sleeping sites, was

recorded using a GPS (Oliveira et al. 2010, 2011). Where

elevation data were available along with the coordinates,

these data were also collected, and otherwise the GTOPO-

30 database (http://eros.usgs.gov/#/Find_Data/Products_

and_Data_Available/gtopo30_info) was used to estimate

the mean elevation of the 30’’ by 30’’ pixel in which a

given coordinate was present. Using these coordinates,

climate data (mean and diurnal temperature amplitude,

precipitation, wind speed, relative sunshine hours and rel-

ative humidity) were extracted over the geographic distri-

bution of each of the 75 species from New et al. (2002).

Temperatures attributed to each record in a given pixel

were corrected based on its elevation (either as reported in

the database or obtained from GTOPO-30) and using mean

values of monthly lapse rates from NCEP re-analysis

(http://www.esrl.noaa.gov/psd/data/reanalysis/reanalysis.

shtml). No corrections were made on other climate vari-

ables. While most species had a set of coordinates dis-

tributed in a relatively regular way, there was a great deal

of heterogeneity and bias for some of the tree species,

particularly those species which are endemic to the BAF,

where few specimens had been collected, or those species

where extensive sampling was done only in a part of their

full range and was limited in other regions. Of the 75

species, nearly one-third (26) are endemic to Brazil, with

most of them having their range in the BAF and neigh-

bouring caatinga (neighbouring dry shrubland ecosystem

with complex borders with the BAF).

Simulations

The CARAIB model was originally developed to under-

stand the role of vegetation in the carbon cycle. It has been

used to model present-day carbon cycle (Warnant et al.

1994; Gerard et al. 1999), potential vegetation distribution

in past climates (e.g. Miocene, Francois et al. 2006; Henrot

et al. 2010; last glacial maximum, Francois et al. 1998) and

potential vegetation distribution, net primary productivity

(NPP), and wild fires in the future (Dury et al. 2011).

Detailed descriptions of CARAIB modules are found in

Otto et al. (2002), Dury et al. (2011) and Francois et al.

(2011). The model calculates carbon flows between

atmosphere and ecosystems on grid spreading at selected

spatial scales, global to regional. Monthly climatic fields

(mean and diurnal temperature amplitude, precipitation,

wind speed, relative sunshine hours and relative humidity)

and soil texture and colour were used as inputs; a stochastic

generator produced daily values of climatic variables. It

calculates photosynthesis, respiration and finally NPP of a

predefined set of objects representing plants (PFTs, BAGs

or species) by taking into account competition for space,

light, water and threshold values of climatic variables

controlling germination or increasing mortality under stress

conditions (Otto et al. 2002; Dury et al. 2011). For ger-

mination, we used the following thresholds: minimum of

yearly sum of daily temperature above 5 �C (GDD5) and

maximum of driest month’s soil water (SWmaxg) allowing

germination. We used SWmaxg only for the drought-

deciduous trees, and shrubs (broadleaf evergreen xeric and

sub-desertic) PFTs and not for the individual BAF tree

species, because this is not a constraint in tropical moist

forests. To initiate stress conditions, we used the following

thresholds: coldest day temperature (Tmins) and soil water

of the driest month (SWmins). During conditions of severe

hydric stress, the model also considers the response related

to stomatal closure (under lower relative air humidity or

decreased soil water), as well as a progressive decrease in

the leaf area index, when soil water decreases. Thus, NPP

and growth tend to decrease, when soil water levels

decrease. For the tree species, the climatic threshold values

for germination and growth were the 1 % or the 99 %

quantile of the obtained climatic distributions. BAF vege-

tation was simulated using a 26 PFT classification, which is

an upgrade from the 15 PFTs originally described in Ute-

scher et al. (2007) and Francois et al. (2011). For the

simulations described here, we were not able to use spe-

cies-specific morpho-physiological values, because they

are not available uniformly for many species. While the fire

module in CARAIB was run, this version of CARAIB did

not use its output, so that impacts of fire events on standing

biomass related to increased dryness could not be

accounted for. Land-use changes, principally conversion of

forest to agricultural or urban use, which can impact NPP

(usually a decrease, e.g. DeFries 2002), were not also taken

into account in this version.

We simulated the present-day distributions of the tree

species and the PFTs in the BAF, using 330 ppmv of

carbon dioxide and corresponding climate variables,

between 1961 and 1990 (New et al. 2002). For future-

scenario simulations, we used three ARPEGE-CLIMATE

scenarios (CMIP3) as outlined by the 2007 IPCC report

(IPCC 2007) to reach equilibrium for an average climate

between 2071 and 2100 (A1, A1B and A2). The model

chosen for these future simulations is coherent with the

mean values of the CMIP5 scenarios, which were not

available when these simulations were performed. For

example, in the most pessimistic A2 scenario, the annual

precipitation for this region has a slight increase that is

consistent with the mean values for annual precipitation

from CMIP5. However, the precipitation in the driest

month(s) can still decrease and lead to hydric stress, the

effects of which can be diminished due to fertilisation by

increased CO2 concentrations. Within the A1 scenario,

A1B includes balancing sources of fuel (fossil and non-

fossil based), and of the three scenarios being modelled,

Modelling the distribution of key tree species 685

123

this shows the second highest carbon dioxide levels at

660 ppmv. The B1 is the most optimistic scenario which

assumes emissions levels up to 530 ppmv. The A2 sce-

nario assumes low population and economic growth, but

also low technological change and caps carbon dioxide

levels at 750 ppmv. None of the IPCC scenarios comes

with a probability value. Other, more regional models, or

global models projected from Latin America were not

available at that time either. Models such as the BESM

(Brazilian Model of Global Climate System) were

developed later and do not necessarily guarantee higher

precision, as the spatial resolution is lower than some of

the CMIP5 models. The ARPEGE CMIP3 does not have a

uniform grid; for the Atlantic Forest in our simulated

zone, it has an effective resolution of 1.2–1.8�, compared

to 1.875� for the BESM.

We generated a ‘‘present-day’’ distribution scenario,

showing the full areas where the trees and the region’s

biomes could theoretically be present. To test the accu-

racy of predicted current distribution, we overlaid the

known coordinates per species over maps showing

modelled species NPP distribution within the simulated

area. An NPP cut-off point at 100 g C/m2/year was

considered as presence of the species. To determine the

level of agreement between the model’s current distri-

bution and actual distribution, we calculated a kappa

value, J = [Pr_a - Pr_e]/[1 - Pr_e], where Pr_a is the

proportion of agreement between the model’s predicted

presence and the actual presence, and Pr_e is the

hypothetical probability of chance agreement. Pr_e is

obtained as the proportion of presence calculated by the

model over the entire area of the simulation (which

includes the entire BAF region and extends westwards to

65�W). Kappa values above 0.41 indicate moderate

levels of agreement, between 0.61 and 0.80 substantial

agreement and almost perfect agreement above 0.8

(Landis and Koch 1977) (Table 1 in Electronic Supple-

mentary Material).

For the future simulations, the potential distribution of

the same trees was re-plotted, based on the changes in the

climatic variables. Similar simulations were conducted

based on PFTs to highlight potential biome-level changes.

Additionally, the areas of overlap between the current and

future distributions, and the overall increase or decrease in

distribution were calculated.

Results

Present-day distributions

The superposition of the distribution of observed individual

trees species on the maps of the simulated NPP showed that

the model predicted the distribution accurately for 66 % of

the individual tree species (43 of 75) with more than 70 %

agreement obtained for presence. Of the remaining 32

species with less than moderate agreement, 16 tree species

are endemic to Brazil (the comprehensive analysis with

tree usage by the primates and threshold values is given by

species in Table 1 of Electronic Supplementary Material).

For these ones, the number of occurrences obtained was

very low, though we do not know whether this is because

of a natural rarity or a lack of sampling over the entire

distributions. Additionally, there is no information to

indicate whether the endemism is linked to climatic factors

or other variables such as dispersion or soil factors. Since

the model does not take into account such factors, it is

difficult to obtain accurate predictions for these species.

However, for six (6) endemic species, despite the low

number of occurrences (B10), the model was able to pre-

dict correctly the distribution (J C 0.41). Overall, con-

sidering the limitations in the actual occurrence data, the

model was able to accurately predict presence for 57.3 %

of the species with a moderate or higher level of agreement

(Fig. 1). At biome level (Fig. 2), statistical comparisons are

difficult, given that there are different interpretations of

biomes (extent and class) and the existing vegetation maps

are often not concordant between themselves, e.g. Otto

et al. (2002) reported an agreement of only 74 % between

two global maps of natural vegetation. In addition, the

usefulness of simulating biomes is mostly to detect changes

in the potential vegetation distribution under future climate

change scenarios. Nevertheless, a visual comparison of the

biome distribution simulated by the model and the potential

vegetation distribution in the same area showed substantial

concordance along the coast, tropical rainforest, sub-trop-

ical forest and some tropical seasonal forest in the BAF

area and northeast of this region, savannahs in the caatinga

area. Further west under the correctly predicted Amazon

rainforest, the model uniformly simulated tropical seasonal

forest over the cerrado. This area, despite its exposition to

tropical seasonal climate with annual rainfall averaging

1,500 mm, is covered by an ecotone consisting of sa-

vannahs integrated with more or less deciduous forests.

The main determinants are soil dystrophy, aluminium

toxicity and fire (Mistry 2000). Since CARAIB does not

take into account soil chemistry, it was unable to predict

this particular vegetation but this had limited consequence

to our conclusions since the cerrado was outside our area

of direct interest.

Future distributions

For the distribution of individual species, the predicted

future area of all the individual species increased in the

three scenarios (B1, A1B and A2), with respect to the

686 N. Raghunathan et al.

123

simulation of the present-day area of distribution. For 72

out of 75 species, more than 95 % of the original area

remained in the future simulations, and only three species

showed any substantial change ([15 % decrease) (Fig. 4;

Table 2 in Electronic Supplementary Material). Consider-

ing that the modelled area was just a little over

7,200,000 km2, the species that were already distributed

over a majority of the modelled area naturally showed a

smaller per cent increase in their future distribution.

Additionally, considering the importance of these tree

species to the endemic primates, they also remained within

the BAF biome.

With regard to the biome, in all three future scenarios,

we note in the BAF area a substantial decrease in the sub-

tropical forest replaced by tropical rainforest and towards

the north, a slight thinning of the tropical rainforest

replaced by tropical seasonal forest or possibly by

savannahs (Fig. 3d). Additionally, there is some fragmen-

tation of the savannahs in the caatinga area (towards the

north), which is replaced mostly by tropical seasonal forest.

In all three scenarios, where the current distribution shows

a large sub-tropical forest biome, it has been replaced by

tropical rainforest, the A2 scenario showing the maximum

change. Whether this is a product of ‘‘fully replacing’’ the

biome or merely an increased number of tropical species

‘‘moving in’’ to the area is hard to say.

To understand the changes in the vegetation distribution,

we looked at soil water availability, which under climate

warming is the principal limiting factor in the model on

germination through SWmaxg and on mortality through

SWmins and temperature, which impacts mortality through

Tmins and controls germination through GDD5. For

instance, the analysis of annual mean temperature anoma-

lies between the future scenarios and the current

Fig. 1 Examples of

superposition of actual

coordinates indicating tree

presence and distribution

predicted through CARAIB.

a Substantial (kappa [ 0.8)

agreement for species with large

distribution through tropical

South America; b substantial

(kappa [ 0.8) agreement for

endemic species; c slight to no

agreement for endemic species

with few coordinates

Modelling the distribution of key tree species 687

123

distribution showed most drastic differences in the A2

scenario (maximum anomaly less than 5 �C), which is

concordant with the high emissions scenario (Fig. 3a).

There is a slight overall increase in the mean annual pre-

cipitation for a large part of the modelled area (B300 mm

difference between scenarios); however, the precipitation

in the driest month decreases in all three scenarios, com-

pared to the present day, leading to increased dryness

during the dry season (Fig. 3b). Figure 3c shows minimum

monthly soil water availability anomalies for the three

scenarios (maximum difference at -0.3), and we see some

increased dryness in the caatinga region as the most

drastic, with some slight decreases as we move towards the

Amazon region. CARAIB’s fire module predicts a signifi-

cant increase in fire frequency, due to increased droughts in

the dry period, but this is most prevalent in the caatinga

region, and is much less frequent in the Atlantic Forest. It

can be expected that reduction in biomass linked to fires in

the Atlantic Forest will be limited in the future.

Discussion

In situ conservation is the method of choice for protecting

biodiversity because it provides maximal viability, it is

often the most cost-effective method, and it covers

unknown species with the species of direct interest (Possiel

et al. 1995). The extent of the supposed displacement of

species climate envelopes owing to future climate changes

raises the question of the adequacy between protected area

limits and the future distribution of the species. The results

of growth simulation with the CARAIB DVM showed that

most of the tree species supporting the activity of two

endemic primate species would rather remain in the same

area and increase their range a little in the same biomes

under future predicted climate. This bodes well for con-

serving the associated fauna. DVMs running at continental

or global scale broadly indicate a decrease in species

diversity, but with respect to NPP, their results are variable

(White et al. 1999; Alo and Wang 2008; Sitch et al. 2008;

Reu et al. 2011). Modelling studies focussing on tropical

regions of limited extension are rather rare because there

are little data available (Feeley and Silman 2011). Never-

theless, our results are not fully in accordance with niche-

based model studies. For instance, Colombo and Joly

(2010) evaluate, with two niche-based models (GARP and

MaxEnt), the future distribution of 38 well-represented

trees species typically found in the BAF using climate

simulations produced under conditions similar to B1 and

A1FI scenarios. Their results show a substantial decrease in

the future distribution of many of the species modelled and

a shift towards southern areas of Brazil. For our simula-

tions, the southward shift is reproduced for many species,

but the overall distribution of the species’ is increased

rather than reduced. In the cerrado region adjacent to the

cFig. 3 Annual mean temperature anomalies degree Celsius (a),

precipitation anomalies in the driest month mm/month (b), minimum

monthly soil water availability anomalies, in relative units (SW–WP/

FC–WP; SW soil water, WP wilting point, FC field capacity)

(c) between present-day simulation at 330 ppmv and the three future

climate change scenarios. Simulated biome distribution (d) for the

same future climate change scenarios

Fig. 2 Visual comparison between actual vegetation map and biome distribution simulated by the model for present-day (1961–1990) scenario.

Map on right adapted from Instituto Brasileiro de Geografia e Estatısticas (http://www.ibge.gov.br/home/presidencia/noticias/21052004bio

mashtml.shtm)

688 N. Raghunathan et al.

123

Modelling the distribution of key tree species 689

123

BAF, Siqueira and Peterson (2003) also found a substantial

reduction ([50 %) in the original ranges with a similar

method. By contrast, processes-based model studies

focussing on tropical areas rather concluded to increased

growth of the plant under future conditions. For example,

using direct observations and models with parametrisation

of leaf-level photosynthesis incorporating known temper-

ature sensitivities, Lloyd and Farquhar (2008) concluded

that the tropical forests are not ‘‘dangerously close’’ to their

optimum and that the new conditions of temperature, water

availability and CO2 concentration would produce higher

growth rate. Gopalakrishnan et al. (2011) using a DVM to

model climate change impacts on Tectona grandis in India

found that despite increased vulnerability of the species to

climate, there would be an increase in its NPP and biomass

due to increased CO2 concentration. These conclusions

agree with Lewis et al. (2009) review over recent decade

changes in the tropical forests. Analysing the results of

several complementary approaches such as permanent plot

studies or physiology experiments, they reached the con-

clusion that both gross and net primary productivity have

increased, as have tree growth, recruitment, and mortality

rates, and forest biomass.

The scope of conclusions of modelling studies dealing

with organisms and climate change is limited by the

ability to reproduce the current distributions but also by

the hypothesis of the physiological response to new

conditions to simulate the future distributions. Concerning

ability to reproduce current distribution, a significant

proportion of the species modelled in this study obviously

suffered from too few calibration data. The CARAIB

model does not include important processes such as

migration, biotic interactions or preference for particular

soil conditions. However, the question of the physiologi-

cal responses to climate change is a cornerstone to make

predictions of the future distribution of the organisms, but

only the processes-based models explicitly treat the

problem. Resistance of plants to water stress increases

with atmospheric CO2 concentration because CO2 can be

more easily absorbed minimising transpiration by stoma-

tal closure (Fitter and Hay 2002). It becomes obvious that

enhanced CO2 concentration in the atmosphere could

effectively stimulate tree growth in temperate regions

(Leaky et al. 2009). These effects are clearly reproduced

by the CARAIB model over Europe when comparing the

results of simulations for future climates with or without

increasing atmospheric CO2 concentration (Dury et al.

2011). Concerning plant response to temperature increase,

two phenomena have to be considered, heat injury and net

photosynthesis (NP). Heat injury is observed more on

stems than on leaves, close to the soil when surface

temperatures reach between 52 and 66 �C. It eventually

leads to plant death and is caused by direct solar radiation

or by heat conducted or radiated from soil having low

heat capacity and conductivity. This response, which

could be associated with the effect of mortality inducing

maximal temperature, is not considered in CARAIB

because it is rare in nature, compared to cold injury or

death associated with a minimal temperature (Kozlowski

and Pallardy 1997). The maximal soil temperatures cal-

culated by CARAIB never reached the lower end of the

threshold (52 �C), even in the driest regions of the sim-

ulated area, where highest temperatures are found. The

NPP response to temperature increase follows an asym-

metric bell curve with optimum response depending on

bio-climatic origin of the species and adaptation to season

or station (Berry and Bjorkman 1980; Fitter and Hay

2002). The components of NPP, gross photosynthesis

(GPP) and plant respiration (Ra) have contrasted respon-

ses to temperature. GPP rate gently increases with tem-

perature to peak at values between 10 and 39 �C at least,

following values reviewed by Cannell and Thornley

(1998). On the contrary, Ra rate sharply increases in the

region of 20 �C until a compensation temperature where

Ra overtakes GPP and there is no NPP or growth. These

mechanisms are simulated in CARAIB with standard

parameterisation, i.e. the same values for all the simulated

objects. The importance of the above consideration

regarding temperature effects is sustained by the results of

Vetaas (2002). This author demonstrated that four Rho-

dodendron species are able to withstand a broader range

of temperatures at the warmer end but not at their colder

end of their natural tolerance range. The author suggests

that cold temperatures may set a boundary in terms of

range expansion, but species may be more likely to

Fig. 4 Fertilisation effect of increased CO2 concentrations in the A2

scenario. Three species lost more than 15 % of their original

distribution in the A2 scenario with high emissions, compared to 38

species that lost more than 15 % of the original distribution in the

modified A2 scenario with 330 ppmv of CO2

690 N. Raghunathan et al.

123

survive warmer temperatures in situ. In our simulation,

the future climates imposed higher temperatures (up to

5 �C difference), and low change in precipitation (mini-

mum precipitation did not fall by more than 40 mm),

when combined resulted in some water stress. The plants

were still able to resist this stress thanks to stomatal

closure. Indeed, this resistance is not seen when CO2

concentration is fixed to 330 ppmv in the A2 pessimistic

scenario, where 38 species showed a reduction C15 % in

the original area of distribution compared to the three

species in A2 with 750 ppmv. Niche-based models are

generally only susceptible to capture natural tolerance

ranges, which could be a serious limitation of their ability

for predicting the future range of species sensitive to

temperature. For instance, Zelazowski et al. (2011) found

some risk of forest retreat of humid tropical forest zone

using a niche-based model under 17 climate projections

but, when the results are refined by considering physio-

logical responses to atmospheric CO2, they found possible

niche expansion in many regions.

Finally, another limitation within the results is the

potential increased risk of fire, as outlined by Nepstad et al.

(2008) and Alo and Wang (2008). Even though the plants

could resist increased temperatures and water stress under

future climate, the forest litter would become drier, which

fosters conditions favouring fire ignition. However, barring

an exceptional level of drought (leading to lowered soil

water levels), both in severity and frequency—which was

not foreseen in these simulations over the Atlantic Forest—

the impacts appear to be relatively limited on tree growth.

If our results are correct, in the long term an increase in

temperature may not negatively impact the distribution of

BAF trees. It would be advantageous to test these same

scenarios with climate data from other general circulation

models in a transient way—especially those built specifi-

cally for South America, (e.g. Brazilian Model of Global

Climate System; Nobre et al. 2009, 2013)—because there

are important differences in the future climate predictions

for this region (IPCC 2007). In the meantime, short-term

conservation strategies must focus on the immediate threats

of land-use patterns and land conversion, to ensure that the

BAF habitat and the monkeys are able to survive in the

long term.

Acknowledgments We would like to acknowledge Michel Deque

CNRM-GAME (Meteo-France, CNRS) who provided the ARPEGE

climatic data for the future. Part of this work was funded by the

BIOSERF project of BELSPO.

References

Alo C, Wang G (2008) Potential future changes of the terrestrial

ecosystem based on climate projections by eight general

circulation models. J Geophys Res 113:1–16

Berg M, Kiers T, Driessen G, van der Heijden M, Kooi B, Kuenen F,

Liefting M, Verhoef H, Ellers J (2010) Adapt or disperse:

understanding species persistence in a changing world. Glob

Change Biol 16:587–598

Berry J, Bjorkman O (1980) Photosynthetic response and adaptation

to temperature in higher plants. Ann Rev Plant Physiol

31:491–543

Bertrand R, Lenoir J, Piedallu C, Riofrıo-Dillon G, de Ruffray P,

Vidal C, Pierrat JC, Gegout JC (2011) Changes in plant

community composition lag behind climate warming in lowland

forests. Nat 479:517–520

Brooker R, Travis J, Clark E, Dytham C (2007) Modelling species’

range shifts in a changing climate: the impacts of biotic

interactions, dispersal distance and the rate of climate change.

J Theor Biol 245:59–65

Cannell M, Thornley J (1998) Temperature and CO2 response of leaf

and canopy photosynthesis: a clarification using non-rectangular

hyperbola model of photosynthesis. Ann Bot 82:883–892

Catenacci L (2008) Ecologia alimentar do mico-leao-da-cara-dourada

(primates: Callitrichidae) em areas degradadas da Mata Atlantica

do Sul da Bahia. Dissertation, Universidade Estadual de Santa

Cruz, Brazil

Chen I, Hill J, Ohlemuller R, Roy D, Thomas C (2011) Rapid range

shifts of species associated with high levels of climate warming.

Sci 333:1024–1026

Clusella-Trullas S, Blackburn T, Chown S (2011) Climatic predictors

of temperature performance curve parameters in ectotherms imply

complex responses to climate change. Am Nat 177:738–751

Colombo A, Joly C (2010) Brazilian Atlantic Forest lato sensu: the

most ancient Brazilian forest, and a biodiversity hotspot, is

highly threatened by climate change. Braz J Biol 70:697–708

Colwell R, Brehm G, Cardelus C, Gilman A, Longino J (2008) Global

warming, elevational range shifts, and lowland biotic attrition in

the wet tropics. Sci 322:258–261

Cramer W, Bondeau A, Schaphoff S, Lucht W, Smith B, Sitch S

(2004) Tropical forests and the global carbon cycle: impacts of

atmospheric carbon dioxide, climate change and rate of defor-

estation. Philos Trans R Soc (Biol Sci) 359:331–343

DeFries R (2002) Past and future sensitivity of primary production to

human modification of the landscape. Geophys Res Lett

29:1132–1136

Dury M, Hambuckers A, Warnant P, Henrot A, Favre E, Ouberdous

M, Francois L (2011) Response of the European forests to

climate change: a modelling approach for the 21st century.

iForest Biogeosci For 4:82–99

Feeley K, Silman M (2011) The data void in modelling current and

future distributions of tropical species. Glob Change Biol

17:626–630

Fitter A, Hay R (2002) Environmental physiology of plants, 3rd edn.

Academic Press, San Diego

Francois L, Delire C, Warnant P, Munhoven G (1998) Modelling the

glacial-interglacial changes in the continental biosphere. Glob

Planet Change 17:37–52

Francois L, Ghislain M, Otto D, Micheels A (2006) Late Miocene

vegetation reconstruction with the CARAIB model. Palaeogeogr

Palaeoclimatol Palaeoecol 238:302–320

Francois L, Utescher T, Favre E, Henrot A, Warnant P, Micheels A,

Erdei B, Suc J, Cheddadi R, Mosbrugger V (2011) Modelling

late Miocene vegetation in Europe: results of the CARAIB

model and comparison with palaeovegetation data. Palaeogeogr

Palaeoclimatol Palaeoecol 304:359–378

Gerard J, Nemry B, Francois L, Warnant P (1999) The interannual

change of atmospheric CO2: contribution of subtropical ecosys-

tems? Geophys Res Lett 26:243–246

Gian-Reto W, Beißner S, Burga C (2005) Trends in the upward shift

of alpine plants. J Veg Sci 16:541–548

Modelling the distribution of key tree species 691

123

Gonzalez-Megıas A, Menendez R, Roy D, Brereton T, Thomas C

(2008) Changes in the composition of British butterfly assem-

blages over two decades. Glob Change Biol 14:1464–1474

Gopalakrishnan R, Jayaraman M, Swarnim S, Chaturvedi R, Bala G,

Ravindranath N (2011) Impact of climate change at species

level: a case study of teak in India. Mitig Adapt Strat Glob

Change 16:199–209

Henrot A, Francois L, Favre E, Butzin M, Ouberdous M, Munhoven

G (2010) Effects of CO2, continental distribution, topography

and vegetation changes on the climate at the Middle Miocene: a

model study. Clim Past 6:675–694

Hughes L (2000) Biological consequences of global warming: is the

signal already. Trends Ecol Evol 15:56–61

Huey R, Kearney M, Krockenberger A, Holtum J, Jess M, Williams S

(2012) Predicting organismal vulnerability to climate warming:

roles of behaviour, physiology, and adaptation. Philos Trans R

Soc Lond Biol Sci 367:1665–1679

IPCC (2007) Climate change 2007: the physical science basis. In:

Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt

KB, Tignor M, Miller HL (eds) Contribution of working group

I to the fourth assessment report of the intergovernmental

panel on climate change. Cambridge University Press,

Cambridge

IUCN (2013) IUCN red list of threatened species. Version 2013.2.

www.iucnredlist.org. Downloaded on 08 April 2014

Keith D, Resit Akcakaya H, Thuiller W, Midgley G, Pearson R,

Phillips S, Regan H, Araujo M, Rebelo T (2008) Predicting

extinction risks under climate change: coupling stochastic

population models with dynamic bioclimatic habitat models.

Glob Change Biol 4:560–563

Kozlowski T, Pallardy S (1997) Growth control in woody plants. In:

Mooney HA (ed) Physiological ecology. Academic Press, San

Diego, p 641

Landis R, Koch G (1977) The measurement of observer agreement for

categorical data. Biometrics 33:159–174

Lapenta M, Procopio de Oliveira P, Kierluff M, Motta-Junior J (2003)

Fruit exploitation by Golden Lion Tamarins (Leontopithecus

rosalia) in the Uniao Biological Reserve, Rio das Ostras, RJ-

Brazil. Mammalia 67:41–46

Laurent J, Bar-hen A, Francois L, Ghislain M, Cheddadi R (2004)

Refining vegetation simulation models: from plant functional

types to bioclimatic affinity groups of plants. J Veg Sci

15:739–746

Lenoir J, Gegout J, Marquet P, de Ruffray P, Brisse H (2008) A

significant upward shift in plant species optimum elevation

during the 20th century. Science 320:1768–1771

Lewis S et al. (2009) Increasing carbon storage in intact African

tropical forests. Nat 457:1003–1007

Lloyd J, Farquhar G (2008) Effects of rising temperatures and [CO2]

on the physiology of tropical forest trees. Phil Trans Roy Soc

Lond Biol Sci 363:1811–1817

Lurgi M, Lopez B, Montoya J (2012) Climate change impacts on

body size and food web structure on mountain ecosystems.

Philos Trans Roy Soc 367:3050–3057

Mistry J (2000) World savannas, Ecology and human uses. Pearson

Education Limited, Harlow, UK

Mittermeier R, Robles Gil P, Hoffman M, Pilgrim J, Brooks T,

Mittermeier C, Lamoreux J, Fonseca G (eds) (2005) Hotspots

revisited: earth’s biologically richest and most endangered

terrestrial ecoregions. University of Chicago Press, Chicago

Myers N, Mittermeier R, Mittermeier C, Fonseca G, Kent J (2000)

Biodiversity hotspots for conservation priorities. Nature

403:853–858

Nepstad D, Stickler C, Soares-Filho B, Merry F (2008) Interactions

among Amazon land use, forests and climate: prospects for a

near-term forest tipping point. R Soc 36:1–10

New M, Lister D, Hulme M, Makin I (2002) A high-resolution data

set of surface climate over global land areas. Clim Res 21:1–25

Nobre P, Malagutti M, Urbano D, Almeida R, Giarolla E (2009)

Amazon deforestation and climate change in a couple model

simulation. J Clim 22:5686–5697

Nobre P, Siqueira L, Almeida R, Malagutti M, Giarolla E, Castelao G,

Bottino M, Kubota P, Figueroa S, Costa M, Baptista M, Irber L,

Marcondes G (2013) Climate simulation and change in the

Brazilian Climate Model. J Clim 26:6716–6732

Oliveira L, Hankerson S, Dietz J, Raboy B (2010) Key tree species for

the golden-headed lion tamarin and implications for shade-cocoa

management in southern Bahia, Brazil. Anim Conserv 13:60–70

Oliveira L, Neves L, Raboy B, Dietz J (2011) Abundance of Jackfruit

(Artocarpus heterophyllus) affects group characteristics and use

of space by Golden-Headed Lion Tamarins (Leontopithecus

chrysomelas) in cabruca agroforest. Environ Manage 48:248–262

Otto D, Rasse D, Kaplan J, Warnant P, Francois L (2002) Biospheric

carbon stocks reconstructed at the Last Glacial Maximum:

comparison between general circulation models using prescribed

and computed sea surface temperatures. Glob Planet Change

33:117–138

Parmesan C, Yohe G (2003) A globally coherent fingerprint of

climate change impacts across natural systems. Nature 42:35–42

Preston K, Rotenberry J, Redak R, Allen M (2008) Habitat shifts of

endangered species under altered climate conditions: importance

of biotic interactions. Glob Change Biol 14:2501–2515

Possiel W, Saunier R, Meganck R (1995) In-situ conservation of

biodiversity. In: Saubier RE, Meganck RA, (eds) Conservation

of biodiversity and the regional planning. Organization of

American States and the IUCN–The World Conservation Union

Raxworthy C, Pearson R, Rabibisoa N, Rakotondrazafy A, Ramana-

manjato J, Raselimanana A, Wu S, Nussbaum R, Stone D (2008)

Extinction vulnerability of tropical montane endemism from

warming and upslope displacement: a preliminary appraisal for

the highest massif in Madagascar. Glob Change Biol

14:1703–1720

Reu B, Proulx R, Bohn K, Dyke J, Kleidon A, Pavlick R, Schmidtlein

S (2011) The role of climate and plant functional trade-offs in

shaping global biome and biodiversity patterns. Glob Ecol

Biogeogr 20:570–581

Root T, Price J, Hall K, Schneider S, Rosenzweig C, Pounds J (2003)

Fingerprints of global warming on wild animals and plants.

Nature 421:57–60

Siqueira M, Peterson A (2003) Consequences of global climate

change for geographic distributions of Cerrado tree species.

Biota Neotropica 3:1–14

Sitch S, Huntingford C, Gedney N, Levy P, Lomas M, Piao S, Bett R,

Ciai P, Cox P, Friedlingstein P, Jones C, Prentice I, Woodward F

(2008) Evaluation of the terrestrial carbon cycle, future plant

geography and climate-carbon cycle feedbacks using five

Dynamic Global Vegetation Models (DGVMs). Glob Change

Biol 14:1–25

Thomas C (2010) Climate, climate change and range boundaries.

Divers Distrib 16:488–495

Thuiller W, Albert C, Araujo M, Berry P, Cabeza M, Guisan A,

Hickler T, Midgley G, Paterson J, Schurr F, Sykes M,

Zimmermann N (2008) Predicting global change impacts on

plant species’ distributions: future challenges. Perspect Plant

Ecol Evol Syst 9:137–152

Utescher T, Erdei B, Francois L, Mosbrugger V (2007) Tree diversity

in the Miocene forests of Western Eurasia. Palaeogeogr Palae-

oclimatol, Paleaoecol 253:226–250

Van der Putten W, Macel M, Visser M (2010) Predicting species

distribution and abundance responses to climate change: why it

is essential to include biotic interactions across trophic levels.

Phil Trans of Roy Soc Lond Biol Sci 365:2025–2034

692 N. Raghunathan et al.

123

Vetaas O (2002) Realized and potential climate niches: a comparison

of four Rhododendron tree species. J Biogeogr 29:545–554

Violle C, Navas M, Vile D, Kazakou E, Hummel C, Garnier E (2007)

Let the concept of trait be functional. Oikos 116:882–892

Walther G, Beissner S, Burga C (2005) Trends in the upward shift of

alpine plants. J Veg Sci 16:541–548

Warnant P, Francois L, Strivay D, Gerard J (1994) CARAIB: a global

model of terrestrial biological productivity. Global Biogeochem

Cy 8:255–270

White M, Running S, Thornton P (1999) The impact of growing

season length variability on carbon assimilation and

evapotranspiration over 88 years in the eastern US deciduous

forest. Int J Biometeorol 42:139–145

Williams S, Shoo L, Isaac J, Hoffmann A, Langham G (2008)

Towards an integrated framework for assessing the vulnerability

of species to climate change. PLOS Biol 6:2621–2626

Wright J, Muller-Landau H, Schipper J (2009) The future of tropical

species on a warmer planet. Conserv Biol 23:1418–1426

Zelazowski P, Malhi Y, Huntingford C, Sitch S, Fisher J (2011)

Changes in the potential distribution of humid tropical forests on

a warmer planet. Philos Trans R Soc 369:137–160

Modelling the distribution of key tree species 693

123