Acessibilidade / Reportar erro

Potential distribution of Ocotea odorifera (Vell.) Rohwer in urban areas of Curitiba, Paraná state, Brazil

Distribuição potencial de Ocotea odorifera (Vell.) Rohwer na área urbana de Curitiba, Paraná, Brasil

Abstract

Ocotea odorifera is an endangered tree species, threatened by the urbanization process that occurs in the region of Curitiba, Paraná state, among other factors. Thus, the knowledge about their geographical distribution becomes necessary to plan protected areas that mitigate the effects caused by anthropic pressures on their populations. Therefore, the objective of this research was to develop a model of the potential geographic distribution of Ocotea odorifera and to verify the region of the municipality of Curitiba more relevant to its conservation. First, we georeferenced individuals of the species found in the municipality, while the model of potential distribution of the species was performed using the Maxent® algorithm, using meteorological and topological data as environmental variables. The Jackknife test indicated the most relevant variables to the model and the Area Under the Curve (AUC) indicated its accuracy. The potential distribution map was superimposed on a shapefile file containing the delimitation of 80 protected green areas with fragments of native vegetation. As a whole, 12 occurrence points of the species were located in seven sites. The map generated by Maxent indicated that the central-northern region of Curitiba presents the highest probability of Ocotea odorifera occurrence, corresponding to 19,550.30 ha. The model explained 96.3% of the species distribution in the study area. The relative contribution and the Jackknife test indicated that the most relevant variable to the model was pedology. In future plantings of the species, the pedology should be considered during its planning, as well as other specific characteristics of the soil at each site.

Keywords:
Canela-sassafrás; Maxent; Species distribution modeling; Urban forest

Resumo

Ocotea odorifera é uma espécie arbórea ameaçada de extinção devido ao processo de urbanização que ocorre na região de Curitiba, Paraná. Assim, visando o planejamento de áreas de conservação que mitiguem os efeitos ocasionados pelas pressões antrópicas sobre as suas populações, torna-se necessário o conhecimento sobre a sua distribuição geográfica. Portanto, o objetivo da pesquisa foi desenvolver um modelo de distribuição geográfica potencial de Ocotea odorifera e verificar a região do município de Curitiba mais relevante a sua conservação. Primeiramente, foram georreferenciados indivíduos da espécie encontrados no município, enquanto a modelagem de distribuição potencial da espécie foi realizada por meio do algoritmo Maxent®, sendo utilizados como variáveis ambientais os dados meteorológicos e topo-edáficos. O teste de Jackknife indicou as variáveis mais relevantes ao modelo e a Área Sob a Curva (AUC), a sua acurácia. O mapa de distribuição potencial foi sobreposto a um arquivo shapefile contendo a delimitação de 80 áreas verdes protegidas com remanescentes de vegetação. Ao todo, 12 pontos de ocorrência da espécie foram localizados, em sete locais. O mapa gerado pelo Maxent indicou que a região centro-norte de Curitiba apresenta as maiores probabilidades de ocorrência de Ocotea odorifera, correspondendo a 19.550,30 ha. O modelo explicou 96,3% da distribuição da espécie na área de estudo. A contribuição relativa e o teste de Jackknife indicaram que a variável mais relevante ao modelo foi pedologia. Em futuros plantios da espécie deve-se considerar a pedologia durante o seu planejamento, além de outras características específicas do solo em cada local de plantio.

Palavras-chave:
Canela-sassafrás; Floresta urbana; Maxent; Modelagem de distribuição de espécies

Introduction

Species extinction is a major element of the current environmental crisis, mainly in the context of urbanization. Discussions about biodiversity loss and its consequent extinction rate have been intensified due to its extent worldwide.

In these circumstances, conservation and restauration strategies for threatened species populations depend on the knowledge of their natural geographical distribution, which is determined by variables such as climate, soil and dispersal capacity (YANG et al., 2013YANG, X. Q. et al. Maxent modeling for predicting the potential distribution of medicinal plant, Justicia adhatoda L. in Lesser Himalayan foothills. Ecological Engineering, Amsterdam, v. 51, p. 83-87, 2013.; ZUQUIM et al., 2019ZUQUIM, G. et al. The importance of soils in predicting the future of plant habitat suitability in a tropical forest. Plant Soil, Perth, v. 435, p. 1-20, 2019.). Thus, estimates of suitable areas for species based on their potential distribution is fundamental for a conservation planning that mitigate the effects caused by anthropic pressures over biodiversity (ZUQUIM et al., 2019ZUQUIM, G. et al. The importance of soils in predicting the future of plant habitat suitability in a tropical forest. Plant Soil, Perth, v. 435, p. 1-20, 2019.).

Species Distribution Predictive Models (SDPMs) are important tools to estimate geographical distribution of species. In general, these potential distribution modelings create cartographic representations of a suitable spatial region for the presence of species according to environmental variables that match its distribution requirements (MATEO; FELICÍSIMO; MUÑOZ, 2011MATEO, R. G.; FELICÍSIMO, A. M.; MUÑOZ, J. Modelos de distribución de especies: una revisión sintética. Revista Chilena de Historia Natural, Santiago, v. 84, n. 2, p. 217-240, 2011.). Also, modelings are projected over maps providing numeric values of each point in the landscape to estimate the probability of each pixel as a potential point for the presence of the species (MATEO; FELICÍSIMO; MUÑOZ, 2011MATEO, R. G.; FELICÍSIMO, A. M.; MUÑOZ, J. Modelos de distribución de especies: una revisión sintética. Revista Chilena de Historia Natural, Santiago, v. 84, n. 2, p. 217-240, 2011.).

Previous studies of Species Distribution Predictive Models, however, have often focused on areas of great geographic extension wherein accuracy is decreased by its size, besides of considering weakly regional variations of environmental conditions and its interactions (MURPHY; LOVETT-DOUST, 2007MURPHY, H. T.; LOVETT-DOUST, J. Accounting for regional niche variation in habitat suitability models. Oikos, New Jersey, v. 116, n. 1, p. 99-110, 2007.; FRANKLIN, 2010FRANKLIN, J. Mapping species distributions: spatial inference and prediction. Cambridge University Press, 2010. 340p.). As a consequence, few researchers have addressed modeling in smallest areas such as municipalities or protected areas.

The municipality of Curitiba is in the Mixed Ombrophilous Forest (MOF) domain, as part of forestry formations and associated ecosystems that compose Brazil’s Atlantic Forest. This region has several threatened species, such as Ocotea odorifera (Vell.) Rohwer (canela-sassafrás), species highlighted here since it is an indicator for this forest ecosystem. A reasonable explanation for Ocotea odorifera extinction risk is their high market value, inappropriate species management, and extensive urban development. Hence, these factors may be responsible for Ocotea odorifera being classified as “Endangered” (EN) according to the Official List of Brazilian Flora Threatened Species (BRASIL, 2014BRASIL. Ministério do Meio Ambiente. Portaria n. 443, de 17 de dezembro de 2014. Reconhece como espécies da flora brasileira ameaçadas de extinção aquelas constantes da “Lista Nacional Oficial de Espécies da Flora Ameaçadas de Extinção”. Ministério do Meio Ambiente. Brasília, 17 dez. 2014.).

Taking into account that the greater detail of species potential distribution in a municipality may provide the greater relevant information to its conservation, the aim of this research was to develop a model of the potential geographic distribution of Ocotea odorifera and verify which regions of Curitiba are more important to conserve.

Material and methods

Study area

The municipality of Curitiba is situated in the Paraná State, Brazil, between geographical coordinates 25°25’46” S and 49°16’16” W, comprising a total area of 435.27 km2 with 934.6 meters of altitude (Figure 1) (IPPUC, 2015INSTITUTO DE PESQUISA E PLANEJAMENTO URBANO DE CURITIBA. IPPUC. Nossa Curitiba: perfil físico-territorial e socioeconômico da cidade de Curitiba, 2015. IPPUC, 2015. 36p.). The forest area is mostly covered by the Atlantic Forest Biome and the terrain is flat or gently undulated, consisting of Guabirotuba formation sediments (IPPUC, 2015INSTITUTO DE PESQUISA E PLANEJAMENTO URBANO DE CURITIBA. IPPUC. Nossa Curitiba: perfil físico-territorial e socioeconômico da cidade de Curitiba, 2015. IPPUC, 2015. 36p.).

Figure 1
Map of Curitiba’s municipality and their administrative regions in the state of Paraná, Brazil
Figura 1
Mapa do município de Curitiba e suas regiões administrativas no Estado do Paraná, Brasil

According to the climatic classification of Köeppen-Geiger, this region exhibits Cfb characteristics (subtropical and mesothermal) (IPPUC, 2015INSTITUTO DE PESQUISA E PLANEJAMENTO URBANO DE CURITIBA. IPPUC. Nossa Curitiba: perfil físico-territorial e socioeconômico da cidade de Curitiba, 2015. IPPUC, 2015. 36p.). Its average annual temperature and rainfall are 17. 4°C and 1,486.5 mm/year, respectively (IPPUC, 2015INSTITUTO DE PESQUISA E PLANEJAMENTO URBANO DE CURITIBA. IPPUC. Nossa Curitiba: perfil físico-territorial e socioeconômico da cidade de Curitiba, 2015. IPPUC, 2015. 36p.).

Grassland and forest formations with species characteristic of MOF (both Montane and Alluvial) occurred in Curitiba. The MOF regions had a high physiognomic relevance of species Araucaria angustifolia (Bertol.) Kuntze (Paraná pine), Ocotea porosa (Nees & Mart.) Barroso (Brazilian-walnut), and Ocotea odorifera, besides other species of the Ocotea genus and the Fabaceae and Myrtaceae botanic families (RODERJAN et al., 2002RODERJAN, C. V. et al. As unidades fitogeográficas do Estado do Paraná. Ciência & Ambiente, Santa Maria, v. 24, n. 1, p. 75-92, 2002.). In most of the municipality area, there was grassland composed mainly of species of Poaceae, interspersed by small forest nuclei wherein A. angustifolia, O. porosa and Ocotea odorifera predominated. The mentioned Ocotea forms dense sub-forests which characterized the plant physiognomy of the region (RODERJAN et al., 2002RODERJAN, C. V. et al. As unidades fitogeográficas do Estado do Paraná. Ciência & Ambiente, Santa Maria, v. 24, n. 1, p. 75-92, 2002.).

In regard to the current municipality vegetation in terms of its urban forest, the area covered by vegetation corresponds to 43.69% of the city, in which 34.70% is private urban forest inserted within the blocks and 8.98% is public urban forest, considering the streets, green areas and water bodies (GRISE; BIONDI, ARAKI, 2016aGRISE, M. M.; BIONDI, D.; ARAKI, H. A floresta urbana da cidade de Curitiba, PR. Floresta, Curitiba, v. 46, n. 4, p. 425-438. 2016a. ). Additionally, 4.99% of Curitiba’s vegetation area is corresponded by flower beds, 3.23% by green areas and 0.76% by riparian areas (GRISE; BIONDI, ARAKI, 2016aGRISE, M. M.; BIONDI, D.; ARAKI, H. A floresta urbana da cidade de Curitiba, PR. Floresta, Curitiba, v. 46, n. 4, p. 425-438. 2016a. ).

Data collection and analyses

Firstly, we researched Ocotea odorifera on specialized bibliography, botanical collections and observation records within the study area, such as herbalized material available at the SpeciesLink database (CRIA, 2017); floristic surveys conducted in Curitiba; management plans of protected areas; and reports from the Municipal Secretary of Environment and experts from the Federal University of Paraná.

Thereafter, we conducted field collections as recommended by Kamino et al. (2012KAMINO, L. H. Y. et al. The application of species distribution models in the megadiverse Neotropics poses a renewed set of research questions. Frontiers of Biogeography, Oakland, v. 4, n. 1, p. 7-10, 2012.), where we recorded the geographic location; altitude; and the following habitat’s characteristics: topography, area conservation status, and vegetal cover; of each Ocotea odorifera found in the city. The habitat’s characteristics are important information due to its fast transformation in the urban context. The geographic location of each individual was registered using Garmin GPS® 78s, with as approximate error of 10 meters. All specimens were recorded for this study, including the young ones.

We model the potential distribution of Ocotea odorifera in Curitiba through the Maxent® algorithm (PHILLIPS; DUDÍK; SCHAPIRE, 2004PHILLIPS, S.; DUDÍK, M.; SCHAPIRE, R. A maximum entropy approach to species distribution modeling. In: 21st International Conference on Machine Learning, New York, 2004. Proceedings… ACM Press: New York, 2004, p. 655-662. ), version 3.4.1. of 2017. Maxent uses species presence data and environmental variables from the confirmed geographic location to estimate a range of probabilities for the species potential distribution. A logistic function is used by the model setup which provides an estimate probability from 0 to 1% of species presence (PHILLIPS; DUDÍK; SCHAPIRE, 2004PHILLIPS, S.; DUDÍK, M.; SCHAPIRE, R. A maximum entropy approach to species distribution modeling. In: 21st International Conference on Machine Learning, New York, 2004. Proceedings… ACM Press: New York, 2004, p. 655-662. ), indicating the habitat’s environmental suitability for the species occurrence (KHANUM; MUMTAZA; KUMAR, 2013KHANUM, R.; MUMTAZA, A. S.; KUMAR, S. Predicting impacts of climate change on medicinal asclepiads of Pakistan using Maxent modeling. Acta Oecologica, Paris, v. 49, p. 23-31, 2013.). We established 1,000 combinations with 10 replications in cross-validate mode and adjusted its radius with “-30” according to the working scale.

We selected the environmental variables most relevant to the species distribution as a function of spatial scale, aiming to avoid using variables that would provide unnecessary restrictions to the model (PEARSON et al., 2007PEARSON, R. G. et al. Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. Journal of Biogeography, Oxford, v. 34, n. 1, p. 102-117, 2007.; FIGUEIREDO et al., 2015FIGUEIREDO, S. D. M. et al. Predição da distribuição de espécies florestais usando variáveis topográficas e de índice de vegetação no leste do Acre, Brasil. Acta Amazonica, Manaus, v. 45, n. 2, p. 167-174, 2015.). Thereby, we evaluated the meteorological and topological data, with 30 meters of spatial resolution (Table 1).

Table 1
Abiotic variables used for the elaboration of the potential distribution modelling of Ocotea odorifera
Tabela 1
Variáveis abióticas utilizadas para a elaboração do modelo de distribuição potencial de Ocotea odorifera

The precipitation values refer to annual collections from four monitoring points within the municipality limits, during the period from 1993 to 1999 (ÁGUAS PARANÁ, 2017ÁGUAS PARANÁ. Sistema de Informações Hidrológicas. 2017. Available at: http://www.aguasparana.pr.gov.br/pagina-264.html . Accessed in: July 18th 2017.
http://www.aguasparana.pr.gov.br/pagina-...
). The temperature data were recorded during the four seasons of 2011, from 44 collection points distributed in four transects in the urban area of Curitiba (LEAL; BIONDI, BATISTA, 2014LEAL, L.; BIONDI, D.; BATISTA, A. C. Condições meteorológicas e diferenças térmicas diárias extremas no clima urbano da cidade de Curitiba. Scientia Plena, Aracajú, v. 10, n. 1, p. 1-11, 2014.). The Inverse Distance Weighting (IDW) procedure was used to interpolate the meteorological variables.

We converted the environmental layers corresponding to the environmental variables to the ASCII format (.asc), cut out based on the geographical boundaries of the municipality of Curitiba, and defined with the same spatial resolution of 30 m due to the resolution of the topographic data. We used the QGIS® software, version 2.18, to perform these analyses.

In order to avoid multicollinearity, when there are high correlations (close to 1) among the abiotic variables, we analyzed the environmental data in pairs through the Principal Component Analysis (PCA) in software R, considering on the analysis values lower than 0.7 on Pearson correlation (FIGUEIREDO et al., 2015FIGUEIREDO, S. D. M. et al. Predição da distribuição de espécies florestais usando variáveis topográficas e de índice de vegetação no leste do Acre, Brasil. Acta Amazonica, Manaus, v. 45, n. 2, p. 167-174, 2015.).

We used the Jackknife test to verify the most relevant variables to the model and the estimation of its accuracy was performed through the Receiver Operating Characteristic (ROC), which evaluates its performance through a single value that represents the Area Under the Curve (AUC) (KHANUM; MUMTAZA; KUMAR, 2013KHANUM, R.; MUMTAZA, A. S.; KUMAR, S. Predicting impacts of climate change on medicinal asclepiads of Pakistan using Maxent modeling. Acta Oecologica, Paris, v. 49, p. 23-31, 2013., FIGUEIREDO et al., 2015FIGUEIREDO, S. D. M. et al. Predição da distribuição de espécies florestais usando variáveis topográficas e de índice de vegetação no leste do Acre, Brasil. Acta Amazonica, Manaus, v. 45, n. 2, p. 167-174, 2015.). The closer to 1 the AUC value, the better the model ability to estimate the presence of environmental conditions for the species (MARCO JÚNIOR; SIQUEIRA, 2009MARCO JÚNIOR, P.; SIQUEIRA, M. F. Como determinar a distribuição potencial de espécies sob uma abordagem conservacionista? Megadiversidade, Belo Horizonte, v. 5, n. 1, p. 65-76, 2009.). We used the model which AUC values displayed both the total set and sub-set of occurrence (Training data), chosen randomly to test the predictive capacity of the model (Test data). The results were closer to 1, as recommended by Gomes (2012GOMES, V. H. F. Modelagem de distribuição de espécies comerciais da flora ameaçadas no Estado do Pará: um enfoque sobre as unidades de conservação. 2012. 51 f. Dissertação (Mestrado em Gestão de Recursos Naturais e Desenvolvimento Local na Amazônia) - Universidade Federal do Pará, Belém. 2012.).

The final map of Ocotea odorifera potential distribution in the municipality of Curitiba overlapped a shapefile containing the delimitation of protected green areas: parks, forests, Private Reserves of Municipal Natural Patrimony (PRMNP, IUCN category IV), Urban Biodiversity Conservation Forests (UBCF), Ecological Station (IUCN category Ia) and Wildlife Refuge (IUCN category III) (Figure 4). This approach aimed to verify the most suitable sites to conserve the species (GOMES, 2012GOMES, V. H. F. Modelagem de distribuição de espécies comerciais da flora ameaçadas no Estado do Pará: um enfoque sobre as unidades de conservação. 2012. 51 f. Dissertação (Mestrado em Gestão de Recursos Naturais e Desenvolvimento Local na Amazônia) - Universidade Federal do Pará, Belém. 2012.). The Municipal Botanical Garden, the Zoological Garden of Curitiba and the Public Walk were categorized as parks.

We used the typologies of protected green areas because they reveal the greatest amount of remaining vegetation coverage among the typologies of green areas of the municipality, in the case of parks and forests (GRISE; BIONDI, ARAKI, 2016bGRISE, M. M.; BIONDI, D.; ARAKI, H. Distribuição espacial e cobertura de vegetação das tipologias de áreas verdes de Curitiba, PR. Floresta e Ambiente, Seropédica, v. 23, n. 4, p. 498-510. 2016b.), as they have fewer anthropic impacts and because they offer greater conservation possibilities of the species in relation to other typologies, once they already have their using consolidated by the specific legislation (CURITIBA, 2000CURITIBA. Lei n. 9804, de 03 de janeiro de 2000. Cria o sistema de unidades de conservação do município de Curitiba e estabelece critérios e procedimentos para implantação de novas unidades de conservação. Diário Oficial do Município. Curitiba, 2000.; 2015CURITIBA. Lei n. 14.587, de 14 de janeiro de 2015. Reestrutura o programa das Reservas Particulares do Patrimônio Natural Municipal - RPPNM no município de Curitiba, revoga as leis nº 12.080, de 19 de dezembro de 2006 e lei nº 13.899, de 9 de dezembro de 2011. Diário Oficial do Município. Curitiba, 2015.). However, since all the typologies of the protected green areas in Curitiba were not selected, the term Selected Protected Green Areas (SPGAs) was chosen in the present study, considering those that meet the criteria mentioned above.

A shapefile for the SPGAs was created based on the vectoral files overlapping provided by the Institute for Research and Urban Planning of Curitiba (IPPUC), dated July 2018, containing the delimitations of 48 parks and forest areas; 21 Private Reserves; 9 BCBUs; one Ecological Station and one Wildlife Refuge, totalizing 80 protected areas.

Results

Twelve individuals of Ocotea odorifera were found in Curitiba, located in seven green areas of the city: the PRMNP Airumã (1 specimen) and Bosque da Coruja (2); the Forest Boa Vista (1); the Parks Barigui (1) and Barreirinha (1); the Barreirinha Municipal Garden (1); and one area of public interest belonging to the Municipality of Curitiba (5); all presented in Table 2.

Table 2
Ocotea odorifera occurrence in Curitiba, Paraná, Brazil
Tabela 2
Ocorrência de Ocotea odorifera em Curitiba, Paraná, Brasil

PCA analysis did not indicate Pearson correlation values higher than 0.7 (Figure 2). Thus, all the variables selected for the study were used to compose the potential distribution model, being considered pertinent to its autoecology. Most of the variables used to compose the model were non-climatic (10 variables of the 13 used).

Figure 2
Analysis of main components for the environmental variables used for modeling
Figura 2
Análise de Componentes Principais para as variáveis ambientais utilizadas para a modelagem

The map generated by Maxent (Figure 3) indicated that the north-central region of Curitiba, which includes the Boa Vista, Matriz, Santa Felicidade and Cajuru regions, presents the highest probability of occurrence of Ocotea odorifera, corresponding to 19,550.30 ha.

Figure 3
Potential distribution of Ocotea odorifera in Curitiba, Paraná state, Brazil
Figura 3
Distribuição potencial de Ocotea odorifera em Curitiba, Paraná, Brasil

The AUC values ranged from 0.954 to 0.988 for the Training data, and from 0.285 to 1.0 for the Test data. We chose the models which AUC values were 0.963 for the Training data, and 1.0 for the Test data (Figure 4).

Figure 4
Area Under the Curve (AUC) for Ocotea odorifera in Curitiba, Paraná state, Brazil
Figura 4
Área Sob a Curva (AUC) para Ocotea odorifera em Curitiba, Paraná, Brasil

The analysis of the variable relative contributions to the model (Table 3) indicated that the most relevant variable was pedology, contributing with 48.9% of the response indicated by the model. The second and third variables that contributed the most were terrain shape and precipitation, with 12.7% and 11.3%, respectively. Minimum temperature and terrain orientation did not contribute to the model.

Table 3
Percentage contribution of each abiotic variable used for modeling
Tabela 3
Contribuição percentual de cada variável abiótica utilizada para a modelagem

The Jackknife test (Figure 5) indicated that the occurrence of Ocotea odorifera seems to be influenced by the pedology variable, since it was the most relevant one to the model also when used alone. The second most important variable when used alone was altitude. The terrain orientation did not present relevance to the generated model, obtaining null value, being able to be excluded from the database of the model without any damage to its accuracy.

Figure 5
Jackknife test indicating the importance of different variables adjusted to the model
Figura 5
Teste Jackknife indicando a importância de diferentes variáveis ajustadas ao modelo

Discussion

The Ocotea odorifera occurrence points, limited to the northern region of the study area, does not represent a modeling deficiency of the potential distribution of the species, since it provides indications about the most favorable sites for its occurrence. Not all places where individuals of a particular species are present, however, represents the best environments for their establishment, and may offer incomplete environmental conditions to meet their auto-ecological demands. Thus, although individuals can develop, they may not be healthy. Therefore, the potential distribution algorithm verifies the most relevant conditions for the species, independently of the vitality of the sampled individuals.

Franklin (2010FRANKLIN, J. Mapping species distributions: spatial inference and prediction. Cambridge University Press, 2010. 340p.) reinforces this idea by stating that a distribution model can be effective, even if the information used is from part of the whole study area, if the main objective is to predict habitat adequacy of non-inspected sites, especially in areas modified by human activities. In addition, sites with no occurrence records represent a background environment for the species environmental suitability, that is, with medium conditions for its establishment and which, for some auto-ecological reason, were never colonized (FRANKLIN, 2010FRANKLIN, J. Mapping species distributions: spatial inference and prediction. Cambridge University Press, 2010. 340p.).

However, models developed using few samples of biotic data, with less than 15 points of confirmed occurrence of the species, should be interpreted as indicators of regions with environmental conditions similar to those where the species is known (PEARSON et al., 2007PEARSON, R. G. et al. Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. Journal of Biogeography, Oxford, v. 34, n. 1, p. 102-117, 2007.). The authors suggest that such models should be used in the segmentation of field surveys, as a basis for locating unknown populations and individuals, but not as estimates of the actual range limits for a species.

Cruz-Cárdenas et al. (2014)CRUZ-CÁRDENAS, G. et al. Selección de predictores ambientales para el modelado de la distribución de especies en Maxent. Chapingo, Texcoco, v. 20, n. 2, p. 187-201, 2014., recommend that the main variables for the elaboration of SDPMs are topography and pedology, while climatic variables should be considered auxiliary. In addition, we highlight the number of top-soil variables used in the present study, higher than the amount commonly used in SDPMs’ studies. Figueiredo et al. (2015FIGUEIREDO, S. D. M. et al. Predição da distribuição de espécies florestais usando variáveis topográficas e de índice de vegetação no leste do Acre, Brasil. Acta Amazonica, Manaus, v. 45, n. 2, p. 167-174, 2015.) used four non-climatic variables: declivity, exposure, vertical distance to the nearest drainage, and Normalized Difference Vegetation Index (NDVI) to estimate the potential distribution of forest species in the state of Acre, Brazil.

The central-northern region of the city of Curitiba was indicated to be the most relevant for the occurrence of Ocotea odorifera due to the fact that the north region concentrates the highest altitudes of approximately 1,000 m and slopes, in addition to soils groups Latosols and Cambisols, indicated by Martins (2016MARTINS, E. M. Conservação de Ocotea catharinensis, Ocotea odorifera e O. porosa: espécies de Lauraceae ameaçadas de extinção. 2016. 154 f. Tese (Doutorado em Botânica) - Instituto de Pesquisas Jardim Botânico do Rio de Janeiro/ Escola Nacional de Botânica Tropical, Rio de Janeiro, 2016.) as preferred by the species. Besides these aspects, according to Grise, Biondi and Araki (2016bGRISE, M. M.; BIONDI, D.; ARAKI, H. Distribuição espacial e cobertura de vegetação das tipologias de áreas verdes de Curitiba, PR. Floresta e Ambiente, Seropédica, v. 23, n. 4, p. 498-510. 2016b.), the Boa Vista and Matriz regions, located in the north of the city, have the largest amounts of protected areas in the municipality. Thus, the results corroborate the places where the individuals of the species were recorded, and those with the highest probability of species’ distribution are located closer to the northern limit of Curitiba and at higher altitudes.

Even in a relatively small area, the Maxent algorithm indicated a limited potential distribution area for Ocotea odorifera to a small portion of Curitiba, probably due to the scale of the abiotic variables used, which favors the greater detail of the adequacy for the species.

The model of potential distribution of Ocotea odorifera made by Martins (2016MARTINS, E. M. Conservação de Ocotea catharinensis, Ocotea odorifera e O. porosa: espécies de Lauraceae ameaçadas de extinção. 2016. 154 f. Tese (Doutorado em Botânica) - Instituto de Pesquisas Jardim Botânico do Rio de Janeiro/ Escola Nacional de Botânica Tropical, Rio de Janeiro, 2016.) for Brazil as a whole indicated that part of the Curitiba region should be one of the priority areas for the species’ conservation, due to the greater environmental suitability it provides. Thus, we highlight the importance of modeling in a smaller scale, since, as noted, there may be significant differences in the occurrence probability. In addition, this approach can contribute to species conservation efforts by providing guidance to the best sites for the establishment of protected areas, and for the reintroduction of their populations. Thus, if the distribution of Ocotea odorifera is considered only on a greater scale, addressing all its ecological limits, conservation strategies may be impaired.

The non-occurrence modeling of Ocotea odorifera throughout the study area can be justified by the arguments of Murphy and Lovett-Doust (2007MURPHY, H. T.; LOVETT-DOUST, J. Accounting for regional niche variation in habitat suitability models. Oikos, New Jersey, v. 116, n. 1, p. 99-110, 2007.), which explain that individuals of a species tend to respond in exactly the same way to environmental conditions. However, a regional variation in the distribution of these conditions may influence the development of individuals. Thus, the results indicate that, although the municipality of Curitiba presents environmental factors relevant to the Ocotea odorifera occurrence, environmental variability may favor or hinder its establishment. Thus, other endangered species with environmental sensitivity may also benefit if more detailed studies of their potential distributions are made.

Therefore, the most important value of AUC indicates that the model generated using such environmental variables explains 96.3% of the species distribution in the study area. According to Yang et al. (2013YANG, X. Q. et al. Maxent modeling for predicting the potential distribution of medicinal plant, Justicia adhatoda L. in Lesser Himalayan foothills. Ecological Engineering, Amsterdam, v. 51, p. 83-87, 2013.), the AUC values found in our study are considered excellent because they are higher than 0.9. The AUC values found in this paper are similar to those of other studies for species of the genus Ocotea, corroborating the ability of the algorithm to predict the potential occurrence of this group of species. Using the algorithm Maxent®, Coelho, Carvalho and Gomide (2016COELHO, G.; CARVALHO, L. M. T.; GOMIDE, L. R. Modelagem preditiva de distribuição de espécies pioneiras no Estado de Minas Gerais. Pesquisa Agropecuária Brasileira, Brasília, v. 51, n. 3, p. 207-214, 2016.) found a similar value of AUC (0,964) for Ocotea corymbosa (Meisn.) Mez, in Minas Gerais. Martins (2016MARTINS, E. M. Conservação de Ocotea catharinensis, Ocotea odorifera e O. porosa: espécies de Lauraceae ameaçadas de extinção. 2016. 154 f. Tese (Doutorado em Botânica) - Instituto de Pesquisas Jardim Botânico do Rio de Janeiro/ Escola Nacional de Botânica Tropical, Rio de Janeiro, 2016.) found that Ocotea catharinensis Mez, O. odorifera and O. porosa presented AUC values of 0,940, 0,910 and 0,960, respectively, in a study that includes part of the Atlantic Forest, from Bahia to Rio Grande do Sul.

The importance of pedology, both in its relative contribution and in its predictive capacity when used alone, is explained by the migration of the species, which may be restricted by soil suitability in its potential propagation pathway, as suggested by Zuquim et al. (2019ZUQUIM, G. et al. The importance of soils in predicting the future of plant habitat suitability in a tropical forest. Plant Soil, Perth, v. 435, p. 1-20, 2019.). Ocotea odorifera has a preference for shallow soils and fast drainage, in addition to high chemical fertility (CARVALHO, 2005CARVALHO, P. E. R. Canela-sassafrás. Circular Técnica. Empresa Brasileira de Pesquisa Agropecuária. n. 110, 2005. 12p.; POTTKER et al., 2016POTTKER, G. S. et al. Padrão espacial de espécies florestais: estudo de caso com Ocotea odorifera (Vell.) Rohwer. Ciência Florestal, Santa Maria, v. 26, n. 4, p. 1097-1106, 2016.).

When the soils have inadequate conditions for the species’ establishment, the area is outside its tolerance and with low probability of its occurrence, even if the climatic conditions are favorable (ZUQUIM et al., 2019ZUQUIM, G. et al. The importance of soils in predicting the future of plant habitat suitability in a tropical forest. Plant Soil, Perth, v. 435, p. 1-20, 2019.). Furthermore, the most pertinent variables to the distribution of the species are only indicative of the best areas for their establishment, so that in the case of planting, other specific parameters to each location should also be considered.

Other studies on the modeling of potential distribution of species of the Lauraceae family did not find a relation between their occurrence and the topography and pedology variables, possibly due to the scale of work, which covers large areas where species distribution tends to be more influenced by climatic variables. Additionally, estimates of potential distribution using only climatic parameters are imprecise (ZUQUIM et al., 2019ZUQUIM, G. et al. The importance of soils in predicting the future of plant habitat suitability in a tropical forest. Plant Soil, Perth, v. 435, p. 1-20, 2019.). Velazco et al. (2017VELAZCO, S. J. E. et al. Using worldwide edaphic data to model plant species niches: an assessment at a continental extent. PloS One, São Francisco, v. 12, n. 10, p. 1-24, 2017.) thus recommend the inclusion of topography and pedology data along with climatic variables to the potential distribution models, aiming to reach more precise forecasts, with the increase of the model accuracy.

The second most relevant variable to the model, the altitude, demonstrates the autoecological preferences of the species, corroborating the assertion of Pottker et al. (2016POTTKER, G. S. et al. Padrão espacial de espécies florestais: estudo de caso com Ocotea odorifera (Vell.) Rohwer. Ciência Florestal, Santa Maria, v. 26, n. 4, p. 1097-1106, 2016.), in which Ocotea odorifera prefers high and steep slopes to develop. This variable is important because it systematically affects climatic behavior, especially temperature and orographic precipitation (FRANKLIN, 2010FRANKLIN, J. Mapping species distributions: spatial inference and prediction. Cambridge University Press, 2010. 340p.). Thus, possibly the slope regulates microclimatic differences in the central-northern region of Curitiba in relation to the rest of the municipality, favoring the abiotic requirements of the species.

Bispo, Valeriano and Kuplich (2010BISPO, P. C.; VALERIANO, M. M.; KUPLICH, T. M. Relação entre as variáveis morfométricas extraídas de dados SRTM (Shuttle Radar Topography Mission) e a vegetação do Parque Nacional de Brasília. Acta Botanica Brasilica, Belo Horizonte, v. 24, n. 1, p. 96-103. 2010.) add that altitude is also related to the pedological distribution gradient, conditioning different vegetation patterns in the landscape, which matches with the vegetation original distribution in Curitiba. Arboreal vegetation was predominant in the central-north portion of the city, as opposed to herbaceous vegetation in the rest of the municipality’s area.

Thus, the abiotic variables that present near zero gains by the Jackknife test, represent the prediction of a random model, whereas those with gains close to 1, being able to be superior to the unit, present information highly correlated to the occurrences, indicating good predictions (COELHO; CARVALHO; GOMIDE, 2016COELHO, G.; CARVALHO, L. M. T.; GOMIDE, L. R. Modelagem preditiva de distribuição de espécies pioneiras no Estado de Minas Gerais. Pesquisa Agropecuária Brasileira, Brasília, v. 51, n. 3, p. 207-214, 2016.).

Other studies have shown that the terrain orientation had no influence on the establishment and development of Ocotea odorifera. Martins (2016MARTINS, E. M. Conservação de Ocotea catharinensis, Ocotea odorifera e O. porosa: espécies de Lauraceae ameaçadas de extinção. 2016. 154 f. Tese (Doutorado em Botânica) - Instituto de Pesquisas Jardim Botânico do Rio de Janeiro/ Escola Nacional de Botânica Tropical, Rio de Janeiro, 2016.) also found a low value in this variable for the prediction of the area occupied by Ocotea odorifera in Brazil, with a value of 1.09% of contribution to their model. However, the terrain orientation influences the amount of solar radiation received by the slope and regulates the local water and energy regimes, affecting the soil moisture available to the plants (BISPO; VALERIANO; KUPLICH, 2010BISPO, P. C.; VALERIANO, M. M.; KUPLICH, T. M. Relação entre as variáveis morfométricas extraídas de dados SRTM (Shuttle Radar Topography Mission) e a vegetação do Parque Nacional de Brasília. Acta Botanica Brasilica, Belo Horizonte, v. 24, n. 1, p. 96-103. 2010.; FRANKLIN, 2010FRANKLIN, J. Mapping species distributions: spatial inference and prediction. Cambridge University Press, 2010. 340p.). In fact, the locations where the Ocotea odorifera specimens were found have a predominantly southern orientation, which tends to provide less sunshine. This finding is evidenced by Carvalho (2005CARVALHO, P. E. R. Canela-sassafrás. Circular Técnica. Empresa Brasileira de Pesquisa Agropecuária. n. 110, 2005. 12p.), which points out the shadow-tolerance characteristic of the species that, in the juvenile phase, needs shadow to develop.

Conclusion

The potential distribution and possible conservation areas for Ocotea odorifera in Curitiba are mainly located in the central-northern region of the city, coinciding with the greater amount of its protected green areas.

Pedology and altitude were the most relevant variables to the species’ occurrance, as indicated by the potential distribution model.

Pedology alone was responsible for approximately half of Ocotea odorifera environmental suitability.

Therefore, we recommend that pedology should be considered by the municipality during future species’ planting plans, in addition to other specific soil features at each planting site.

The study confirmed the preference of the species for higher terrain and the presence of soils of the Latosols and Cambisols classes.

Moreover, we recommend that further studies on the potential distribution of endangered species should use environmental variables with detailed information, aiming a more precise delimitation of priority conservation areas.

References

  • ÁGUAS PARANÁ. Sistema de Informações Hidrológicas. 2017. Available at: http://www.aguasparana.pr.gov.br/pagina-264.html Accessed in: July 18th 2017.
    » http://www.aguasparana.pr.gov.br/pagina-264.html
  • BISPO, P. C.; VALERIANO, M. M.; KUPLICH, T. M. Relação entre as variáveis morfométricas extraídas de dados SRTM (Shuttle Radar Topography Mission) e a vegetação do Parque Nacional de Brasília. Acta Botanica Brasilica, Belo Horizonte, v. 24, n. 1, p. 96-103. 2010.
  • BRASIL. Ministério do Meio Ambiente. Portaria n. 443, de 17 de dezembro de 2014. Reconhece como espécies da flora brasileira ameaçadas de extinção aquelas constantes da “Lista Nacional Oficial de Espécies da Flora Ameaçadas de Extinção”. Ministério do Meio Ambiente. Brasília, 17 dez. 2014.
  • CARVALHO, P. E. R. Canela-sassafrás. Circular Técnica. Empresa Brasileira de Pesquisa Agropecuária. n. 110, 2005. 12p.
  • COELHO, G.; CARVALHO, L. M. T.; GOMIDE, L. R. Modelagem preditiva de distribuição de espécies pioneiras no Estado de Minas Gerais. Pesquisa Agropecuária Brasileira, Brasília, v. 51, n. 3, p. 207-214, 2016.
  • CRUZ-CÁRDENAS, G. et al Selección de predictores ambientales para el modelado de la distribución de especies en Maxent Chapingo, Texcoco, v. 20, n. 2, p. 187-201, 2014.
  • CURITIBA. Lei n. 9804, de 03 de janeiro de 2000. Cria o sistema de unidades de conservação do município de Curitiba e estabelece critérios e procedimentos para implantação de novas unidades de conservação. Diário Oficial do Município. Curitiba, 2000.
  • CURITIBA. Lei n. 14.587, de 14 de janeiro de 2015. Reestrutura o programa das Reservas Particulares do Patrimônio Natural Municipal - RPPNM no município de Curitiba, revoga as leis nº 12.080, de 19 de dezembro de 2006 e lei nº 13.899, de 9 de dezembro de 2011. Diário Oficial do Município. Curitiba, 2015.
  • FIGUEIREDO, S. D. M. et al Predição da distribuição de espécies florestais usando variáveis topográficas e de índice de vegetação no leste do Acre, Brasil. Acta Amazonica, Manaus, v. 45, n. 2, p. 167-174, 2015.
  • FRANKLIN, J. Mapping species distributions: spatial inference and prediction. Cambridge University Press, 2010. 340p.
  • GOMES, V. H. F. Modelagem de distribuição de espécies comerciais da flora ameaçadas no Estado do Pará: um enfoque sobre as unidades de conservação. 2012. 51 f. Dissertação (Mestrado em Gestão de Recursos Naturais e Desenvolvimento Local na Amazônia) - Universidade Federal do Pará, Belém. 2012.
  • GRISE, M. M.; BIONDI, D.; ARAKI, H. A floresta urbana da cidade de Curitiba, PR. Floresta, Curitiba, v. 46, n. 4, p. 425-438. 2016a.
  • GRISE, M. M.; BIONDI, D.; ARAKI, H. Distribuição espacial e cobertura de vegetação das tipologias de áreas verdes de Curitiba, PR. Floresta e Ambiente, Seropédica, v. 23, n. 4, p. 498-510. 2016b.
  • INSTITUTO DE PESQUISA E PLANEJAMENTO URBANO DE CURITIBA. IPPUC. Nossa Curitiba: perfil físico-territorial e socioeconômico da cidade de Curitiba, 2015. IPPUC, 2015. 36p.
  • INSTITUDO DE TERRAS, CARTOGRAFIA E GEOLOGIA DO PARANÁ. ITCG. Arquivos shapefile de fitogeografia e geologia de Curitiba. Curitiba, 2017. Fornecido.
  • INSTITUTO NACIONAL DE PESQUISAS ESPACIAIS. INPE. TopoData. 2011. Available at: http://www.webmapit.com.br/inpe/topodata /. Accessed in: July 10 2017.
    » http://www.webmapit.com.br/inpe/topodata
  • KAMINO, L. H. Y. et al The application of species distribution models in the megadiverse Neotropics poses a renewed set of research questions. Frontiers of Biogeography, Oakland, v. 4, n. 1, p. 7-10, 2012.
  • KHANUM, R.; MUMTAZA, A. S.; KUMAR, S. Predicting impacts of climate change on medicinal asclepiads of Pakistan using Maxent modeling. Acta Oecologica, Paris, v. 49, p. 23-31, 2013.
  • LEAL, L.; BIONDI, D.; BATISTA, A. C. Condições meteorológicas e diferenças térmicas diárias extremas no clima urbano da cidade de Curitiba. Scientia Plena, Aracajú, v. 10, n. 1, p. 1-11, 2014.
  • MARCO JÚNIOR, P.; SIQUEIRA, M. F. Como determinar a distribuição potencial de espécies sob uma abordagem conservacionista? Megadiversidade, Belo Horizonte, v. 5, n. 1, p. 65-76, 2009.
  • MARTINS, E. M. Conservação de Ocotea catharinensis, Ocotea odorifera e O. porosa: espécies de Lauraceae ameaçadas de extinção. 2016. 154 f. Tese (Doutorado em Botânica) - Instituto de Pesquisas Jardim Botânico do Rio de Janeiro/ Escola Nacional de Botânica Tropical, Rio de Janeiro, 2016.
  • MATEO, R. G.; FELICÍSIMO, A. M.; MUÑOZ, J. Modelos de distribución de especies: una revisión sintética. Revista Chilena de Historia Natural, Santiago, v. 84, n. 2, p. 217-240, 2011.
  • MURPHY, H. T.; LOVETT-DOUST, J. Accounting for regional niche variation in habitat suitability models. Oikos, New Jersey, v. 116, n. 1, p. 99-110, 2007.
  • PEARSON, R. G. et al Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. Journal of Biogeography, Oxford, v. 34, n. 1, p. 102-117, 2007.
  • PHILLIPS, S.; DUDÍK, M.; SCHAPIRE, R. A maximum entropy approach to species distribution modeling. In: 21st International Conference on Machine Learning, New York, 2004. Proceedings… ACM Press: New York, 2004, p. 655-662.
  • POTTKER, G. S. et al Padrão espacial de espécies florestais: estudo de caso com Ocotea odorifera (Vell.) Rohwer. Ciência Florestal, Santa Maria, v. 26, n. 4, p. 1097-1106, 2016.
  • RODERJAN, C. V. et al As unidades fitogeográficas do Estado do Paraná. Ciência & Ambiente, Santa Maria, v. 24, n. 1, p. 75-92, 2002.
  • VELAZCO, S. J. E. et al Using worldwide edaphic data to model plant species niches: an assessment at a continental extent. PloS One, São Francisco, v. 12, n. 10, p. 1-24, 2017.
  • YANG, X. Q. et al Maxent modeling for predicting the potential distribution of medicinal plant, Justicia adhatoda L. in Lesser Himalayan foothills. Ecological Engineering, Amsterdam, v. 51, p. 83-87, 2013.
  • ZUQUIM, G. et al The importance of soils in predicting the future of plant habitat suitability in a tropical forest. Plant Soil, Perth, v. 435, p. 1-20, 2019.

Data availability

Data citations

INSTITUTO NACIONAL DE PESQUISAS ESPACIAIS. INPE. TopoData. 2011. Available at: http://www.webmapit.com.br/inpe/topodata /. Accessed in: July 10 2017.

Publication Dates

  • Publication in this collection
    10 Feb 2020
  • Date of issue
    Oct-Dec 2019

History

  • Received
    03 June 2019
  • Accepted
    04 Oct 2019
  • Published
    10 Dec 2019
Universidade Federal de Santa Maria Av. Roraima, 1.000, 97105-900 Santa Maria RS Brasil, Tel. : (55 55)3220-8444 r.37, Fax: (55 55)3220-8444 r.22 - Santa Maria - RS - Brazil
E-mail: cienciaflorestal@ufsm.br