CONTEMPORARY AND FUTURE POTENTIAL GEOGRAPHIC DISTRIBUTION OF Cedrela fissilis Vell. UNDER CLIMATE CHANGE SCENARIOS

ABSTRACT The objective of the present work was to model the climate niche of Cedrela fissilis Vell. and to project the contemporary and future potential spatial distribution considering different climate change scenarios. Species occurrence data were obtained from the SpeciesLink and Global Biodiversity Information Facility (GBIF) databases. Altitude data, and 19 climate variables for both present and future conditions were obtained from the WorldClim database. The spatial prediction for the year 2070, considering an optimistic scenario (RCP 4.5) and a pessimistic one (RCP 8.5), was defined by Intergovernmental Panel on Climate Change (IPCC) regarding the concentration of greenhouse gases. The climate niche modeling was performed using the Maximum Entropy algorithm (MaxEnt). The results showed that C. fissilis has a wide geographic distribution, occurring in most South American countries. Its distribution showed a high correlation associated with the isothermal and precipitation variables in the humid month. In future scenarios, impacts on the climatic suitability of the areas in which the species occur, will not be spatially homogeneous. Indeed, reductions of about 47% (RCP4.5) and 63% (RCP8.5) are expected. It is recommended the creation and expansion of Conservation Units (CU) in areas that will remain climatically suitable for this species. The areas that will not have a favorable climate in the future should be considered strategic for genetic rescue and establishment of germplasm banks. Areas changing into a favorable climate should be considered as new areas of ecological and forestry interest.


INTRODUCTION
The potential impacts that global climate changes will have on forest ecosystems are unequivocal (RANDIN et al., 2009), as all tree species present a specifi c climate niche in which they can develop more effi ciently . Thus, climate change is expected to modify the fl oristic composition of forests.
This process occurs because the species demonstrate diff erent requirements and tolerance regarding climate conditions, which are circumscribed within minimum and maximum limits (e.g., tolerance to temperature). Therefore, there is a "specifi c range" of favorable climatic conditions for each species (KÖCKE, 2015). According to Walther et al. (2002), these species can respond to climate change in one of four ways: migrating to regions with a climate similar to where they occur; becoming resilient, i.e. tolerating new climate conditions; evolving, which implies improving their resilience or ability to migrate; or simply being extinguished.
On account of future climate changes, it is inferred that tree species may be replaced in certain regions, so those species that do not adapt to the new environmental conditions will be replaced by the better-adapted ones (FIELD et al., 2014). At the same time, some species may have their area of occurrence reduced or even extinct due to being restricted to climatic refuges. On the other hand, other tree species may have their extent of occurrence (EOO) expanded when the new climatic conditions are more favorable (WEBBER and SCOTT, 2012).
In this context, it is of great importance to understand the impacts of potential future climate changes on the extent of occurrence of tree species, particularly those occurring in highly endangered forest formations, such as the Atlantic Forest. Among the tree species that occur in this biogeographic domain, Cedrela fi ssilis Vell. was selected because it is of substantial commercial and conservation interests in South America. As such, it has been widely investigated in ecological and environmental studies (GARCIA et al., 2011). Additionally, C. fi ssilis was prominent in Brazilian economy as one the most exported wood for timber in the 1970's, as well as it is suitable for a wide variety of uses -from musical instruments, carving works, fi ne furniture, and decorative interior design of boats to civil construction in general (CARVALHO, 2003). However, its worth as timber caused intense exploration of this species, which is now on the IUCN Red List of Threatened Species (INTERNATIONAL  UNION FOR CONSERVATION OF NATURE AND  NATURAL RESOURCES, 2013).
This work aimed to understand the geographic distribution patterns of C. fi ssilis, model its climate niche, and predict the potential distribution of this species aff ected by diff erent scenarios of climate change due to its notable environmental, economic, and social importances.  (GBIF, 2017;GBIF, 2019). The data were rigorously verifi ed for the presence of duplicated data and incoherencies related to geographical distribution, e.g., caused by typos in coordinates (HIJMANS e ELITH, 2015).

MATERIALS AND METHODS
Due to the possibility of bias related to the occurrence of spatially autocorrelated points, frequently found in places with high incidence of collection, such as near herbaria and universities, the geographic occurrence data were spatially fi ltered (BORIA et al., 2014) by means of a subsampling, so that the analysis only considered occurrences at every 2.5 km x 2.5 km.
The altitude and 19 climatic variables obtained from WorldClim database (HIJMANS et al., 2005) at a resolution of 2.5 minutes (~5 km) were considered for the present and future conditions as explanatory variables of the geographical distribution of this species.
Aiming to eliminate problems related to multicollinearity of the climatic and altitude explanatory variables (KHANUM et al., 2013), their Variance Infl ation Factor (VIF) values were determined, and highly autocorrelated variable, i.e., those presenting a VIF above 10, were excluded.
The future climatic conditions for 2070 were based on the atmospheric circulation model HADGEM2-ES, by MET OFFICE, of the United Kingdom (UK), which is part of the Coupled Model Intercorporation Project Phase 5 (CMIP5). Two scenarios were considered based on assessments by Intergovernmental Panel on Climate Change (IPCC, 2013). The most optimistic scenario (RCP 4.5 with radiative forcing reaching 4.5 W/m 2 ) assumes the adoption of mitigation policies regarding the concentration of greenhouse gases on a global level. The most pessimistic one (RCP 8.5 with radiative forcing reaching 8.5 W/m 2 ) foresees extreme conditions considering that greenhouse gas emissions will continue to grow whether strategies to reduce them are not adopted.
The geographical distribution of this species was modeled by the Maximum Entropy algorithm (MaxEnt). For such, 100 pseudo-absences were generated for each observation of the species (LOBO e TOGNELLI, 2011) and randomly distributed within a 500 km radius. The data were subdivided into two groups: one was composed of 70% of the data and used for calibration (training) of adjustments, whereas the remaining 30% was used for validation (test) of adjustments, as suggested by Thuiller (2003). For each distribution of pseudo-absence, two rounds of calibration and test were carried out, totalling, six adjustments (three distributions of pseudo-absences x two calibration and test processes).
The performances of the adjustments were evaluated using the Total Sum of Squares TSS Statistics, which range from -1 to 1. (ALLOUCHE et al., 2006). The better adjustment is closer to 1. Then, the response curves were built between the climate suitability estimate and the most explanatory variables, according to the best performance adjustment. In order to reduce the uncertainties inherent to each adjustment, the current and future prediction for geographic distribution was determined by means of a consensus of adjustments with good performance (TSS > 0.4).
To quantify the impact of diff erent climate change scenarios, a binary projection of the potential geographic extents of occurrence was projected for both current and future scenarios, in which estimates with occurrence probability higher than 50% were considered climatically suitable areas. After that, the projections were overlapped, stability areas were determined as those being potentially climatically suitable or not; whereas instability areas would no longer, or it would be, climatically suitable in the future.

RESULTS
Cedrela fissilis occurs widely across the Neotropical region ( Figure 1). In terms of latitude, its distribution extends from the equatorial region (±10º N) to the south (±30º S) of the Tropic of Capricorn. As to longitude, its geographical distribution is cisandine, occurring in the range between 37 and 83º W.
Regarding the distribution of this species in the American continent, there are data supporting its presence in most countries of South America and also in Central America, Panama, and Costa Rica. Nevertheless, its occurrence in most of the territory of Bolivia, Peru, Paraguay, and particularly in Brazil, is notable. In the latter, the species is distributed across most of its territory, occurring with higher density in the Southeast and South regions of the country.
The most explanatory variables in terms of the geographical distribution of the species, according to best adjustment (TSS = 0.523), were isothermality (Bio3) and precipitation in the wettest month (Bio13). In the case of Bio3, the species presented a higher estimate of suitability in areas where the isothermality ranged from 50% to 80%. With values under or above this threshold, suitability is linearly reduced ( Figure 2a). As to Bio13, the suitability estimate of the species was higher in the range between 150 and 500 mm, presenting a sharp and linear drop below this range and having a constant decrease, however, with stability points, above this range (Figure 2b).
Regarding the current and future climate suitability of the species (Figure 3a, b, d), the most favorable areas for the occurrence of C. fi ssilis in Brazil are mainly in the South and Southeast region, however, to a lesser extent, in the Central-West and Northeast regions. The North region is relatively less favorable for the species occurrence.
In other countries of South America, the east and northeast of Paraguay and Argentina, respectively, as well as the surroundings of Andes in Bolivia and Peru, are the most favorable regions for the species occurrence in terms of climate. In Central America, although there are records of the species occurring in Panama and Costa Rica, the countries further to the north (Nicaragua, Honduras, Guatemala, Belize, and Mexico) presented a higher proportion of more favorable regions, which can also be observed to a lesser extent in Caribbean countries.
In future scenarios, the most optimistic projection (RCP 4.5) foresees a reduction of about 47% in climatic suitability for species occurrence (Figure  3b), considering a loss of about 50% and a gain of 6.55% (Table 2) in relation to the current favorable area. The areas that remained climatically suitable    Of the total extent of favorable area for future species occurrence in this scenario (2,704,475 km 2 ), about 88% will remain favorable for the species occurrence, even after climate changes. About 12% of the remaining areas will be composed of areas that are not currently favorable to this species, nevertheless these area will be climatically suitable for C. fi ssilis in the future. Among these areas, the southernmost region of Brazil and northeastern Uruguay are notable, as well as other smaller and sparse areas located in northern and northeastern Brazil, Venezuela, Bolivia, and Peru.
According to the pessimistic scenario (RCP 8.5), a reduction of about 63% of climatic suitability area for the species is expected ( Table 2). As for the optimistic scenario, the continent's regions that will In fi gures c and e, the dark gray color represents the stable areas, which present favorable climate and will continue to present in the future; the light gray color represents the stable areas not presenting favorable climate and that will remain as such in the future; the black color represents unstable areas that will no longer present favorable conditions for occurrence of the species; the white color represents unstable areas that will begin to present favorable climatic conditions for occurrence of the species. Nas fi guras c e e, em cinza escuro são áreas estáveis, que apresentam o clima favorável e continuarão apresentando no futuro; cinza claro são áreas estáveis, que não apresentam o clima favorável e continuarão não apresentando no futuro; preto são áreas instáveis, que deixarão de apresentar condições favoráveis para a ocorrência da espécie; branco são áreas instáveis, que passarão a apresentar condições climáticas favoráveis para a ocorrência da espécie.
be less impacted, in this sense, are mainly the South of Brazil, followed by the Southeast. In the rest of the continent, the reduction will be more pronounced (Figure 3e).

DISCUSSION
Although Cedrela fi ssilis reaches Central America, it is restricted to areas where isothermality is between 50% and 80% and without long periods of low rainfall. Isothermality values above 80% indicate that daily temperature variation is relatively close to annual temperature variation. This pattern can occur in environments tending to tropicality, especially in regions near the Equator. This pattern occurs because, in these places, both daily and annual variation have small amplitudes. On the other hand, isothermality values below 50% suggest locations where the daily temperature range is higher than the annual.
According to the other species of Cedrela genus, C. fi ssilis is the species with the widest range of geographic distribution throughout South and Central America (KÖCKE, 2015). This fact may be associated with the species' ability to occur in diff erent vegetation formations, being peculiar in dry forests, subsistent in open fi elds and even in Cerrado (RIZZINI, 1971). With regard to edaphic conditions, this species has a preference for well-drained clay and loam soils with high water retention from the bottom of valleys and along rivers and lowlands, however, not developing properly in poorly drained and shallow soils, or in soils with shallow groundwater table (INOUE et al., 1984;REITZ, 1984). The high capacity of physiological adaptation to the light conditions (PINHEIRO et al., 1990), the deciduous behavior, which allows adaptation to environments with water defi cit (MUELLNER et al., 2010), and also the high dispersion capacity of propagules (CARVALHO, 2003) are autecological aspects that may explain the wide geographical distribution of this species.
Based on the results found, our study observed that, despite its wide geographical distribution, C. fi ssilis is more present across the entire Atlantic Forest Biome, mainly covering the South and Southeast of Brazil. This region has temperate and subtropical climate, from mild to hot summers and without long periods of drought. This pattern can be explained by the fact that greater climatic suitability for the species and greater historical climate stability are observed in this region. Thus, providing refuge areas for C. fi ssilis during climate cycles that aff ected other regions more drastically (INOUE et al., 1984;CARNAVAL et al., 2009). On the other hand, the probability of occurrence is lower in other biomes, such as Cerrado and Amazon. The predominant climate types of these regions are Aw, Am, and Af, according to the Köppen classifi cation; overall, these biomes present the average temperature of the coldest month exceeding 18 ºC.
Regarding the impacts of climate change, projections of climatically suitable areas for this species in future climate scenarios indicate that the Table 2 -Impact of climate changes on the absolute (in km2) and relative (in %) potential areas regarding climatic suitability for the occurrence of Cedrela fi ssilis in diff erent climate change scenarios for 2070. Tabela 2 -Impacto de mudanças climáticas sobre as áreas potencias absolutas (em km2) e relativas (em %), no que se refere à adequabilidade climática para a ocorrência de Cedrela fi ssilis, em diferentes cenários de mudanças climáticas para o ano de 2070. northernmost and lower elevation regions will be less favorable to the occurrence of C. fi ssilis ( Figure  3). Notably, the favorable area for C. fi ssilis will be reduced in regions of Cerrado Biome and the northern portion of Atlantic Forest, while the more southern and high altitude regions, especially those located in the subtropical portion of Atlantic Forest, will be more climatically suitable for the species. Similarly, Estrada-Contreras et al. (2016) could fi nd in several scenarios of future climate change for Mexico, that the occurrence of C. odorata L. will be favored, especially in high altitude regions, such as the mountain ranges. Silva (2016), in projecting the climate suitability of Ilex paraguariensis A.St.-Hil. and Mimosa scabrella Benth. in climate change scenarios in Brazil, has similarly verifi ed that regions with high altitudes will tend to be more favorable for the species development, highlighting the regions with typically humid temperate climate, which is also observed for C. fi ssilis.

Scenarios
Forecasts for both climate change scenarios indicate that part of Southern Brazil, where the Pampa Biome occurs -which is not currently climatically suitable for C. fi ssilis -will present favorable conditions for the species in the future.
In general, the displacement of plant species towards colder regions of higher latitude and altitude has been observed with the recent increase in temperature in temperate regions (WALTHER et al., 2002). In addition, species that are heat-or droughttolerant and adapted to fi re will be favored in regions subject to reduced rainfall, which will consequently decrease species richness in most biomes (WILLIAMS et al., 2007).
The impact of climate change on the occurrence of species is part of a process associated with evolutionary history regarding anthropic action, so that forecasting the ability of species to adjust their favorable areas of occurrence is highly complex. In the case of C. fi ssilis, it is key to understand the impacts of anthropic nature, such as forest fragmentation, because the species' populations in Brazil and other countries are threatened by deforestation and selective exploitation (STEINER, 2014).
The impacts of climate change on the geographic occurrence of C. fi ssilis will not take place homogeneously in the territory it occupies. Thus, mitigation measures aimed at the conservation of the species should be analyzed according to the type of impact observed. In this sense, the following possible alternatives are suggested: a) Stable areas, which will remain in the appropriate climate for species, such as those in Southern and Southeastern Brazil, should be considered strategic for in situ conservation by the creation and expansion of protected areas; b) Unstable areas that will no longer have a favorable climate in the future, such as those in Central Brazil and the northern portion of Atlantic Forest, should be considered strategic for genetic rescue and the establishment of germplasm banks; c) Unstable areas that will have a favorable future climate, such as those in southernmost region of Brazil, should be considered as potential new areas of ecological and silvicultural interest; d) For all circumstances considered, the sensitivity of C. fi ssilis to the moth Hypsipyla grandella, when cultivated as pure stands, should not be ignored (CUSATIS et al., 2018). Therefore, planting in the form of consortia or forest enrichment in recovery programs or managed forests is recommended (CUSATIS et al., 2018).

CONCLUSION
Cedrela fi ssilis is widely geographically distributed, occurring in South and Central America, especially in Brazil, Paraguay, Bolivia, and Peru. The occurrence of this species was observed to be more related to regions where climatic conditions presented intermediate values of isothermality (~ 60%) and the wettest month presents rainfall of at least 150 mm.
The climatically suitable area for the potential species occurrence will be reduced in future scenarios across South and Central America, and this is expected to be more pronounced in the pessimistic climate projection by IPCC. At the same time, certain areas, especially those located in the southernmost region of Brazil and northeastern Uruguay, will become climatically more favorable for the species occurrence.

ACKNOWLEDGEMENTS
To FAPESC, for the scholarship granted to the fi rst author and support by means of PAP/UDESC.