Diversity of the Passiflora L. in the Serra do Mar ecoregion and the relationships with environmental gradients, South and Southeast, Brazil

Abstract The species of the Passiflora (Passifloraceae s.s.) are distributed in tropical and subtropical regions and are of great ecological importance. The Serra do Mar is formed of large escarpments along Brazil’s east coast, between the States of Rio de Janeiro and Rio Grande do Sul. This study aims to point out occurrences of Passiflora species in herbaria along the Serra do Mar, in addition to evaluating the conditions of each specimen found. The richness and sample effort maps were prepared to establish the locations where the greatest numbers of species and records were encountered. We evaluated the relationship of species distribution through Canonical Correlation (CCA) and Quadratic Polynomial Analysis. We also evaluated the floristic composition using a cluster analysis. Calculations of the Extent of Occurrence (EOO) and Area of Occupancy (AOO) were performed for endemic species to the Atlantic Forest area occurring in the Serra do Mar. A total of 53 species, belonging to the Passiflora were found in the Serra do Mar. The States of Rio de Janeiro and São Paulo were the ones that recorded the greatest numbers of species and collection effort. The climatic variables Precipitation of the Driest Quarter and Annual Average Temperature are the most correlated to the distribution of Passiflora in the area. Twenty-two species are endemic to the Atlantic Forest area, and 21 are recorded as having some degree of threat on Official Lists. In this study, we present all the Passiflora species recorded for the Serra do Mar, the richness of species, including those endemics to the Atlantic Forest, demonstrates the importance of the area for conservation.


Introduction
The passionflower species known are part of the genus Passiflora L. (Passifloraceae sensu stricto) and are characterized by woody and herbaceous climbers with tendrils, alternate leave, axillar tendrils, and extrafloral nectaries in the petiole and/or leaf blades; flowers with androgynophore and filament corona, five stamens, three or four carpels, and a locule and berried or capsuled fruit (Milward-de-Azevedo et al. 2012).
The original limits of the Serra do Mar coastal forests ecoregion are inserted entirely in the Atlantic domain, in the Rio de Janeiro, São Paulo, Paraná, Santa Catarina e Rio Grande do Sul states (WWF 2008).This ecoregion is defined as an extensive area, with heterogeneous environmental conditions that determine the occurrence of communities that share critical ecological processes (Dinerstein et al. 1995).Its forest fragments are interconnected by escarpment regions, where agricultural activities are not possible, contributing to it being one of the largest forest extensions of the Atlantic Forest (WWF 2008).
Passiflora species are directly linked to environmental filters and respond significantly to gradients.Studies such as Kessler (2002), Ocampo-Pérez et al. (2007) Ocampo et al. (2010), Moraes et al. (2018;2020), and Fernandes et al. (2022) demonstrate the complex relationships between these taxa and the environment where they occur.Environmental filters that include total area considered, latitude, altitude, precipitation, and the interactions between them demonstrate the complexity of these relationships with Passiflora (Lomolino 2001;McCain 2009).
Therefore, this study aimed to evaluate the distribution patterns of Passiflora in the Serra do Mar and its relationship with environmental variables.Moreover, we assessed the of conservation status of the species found and identified possible collection gaps in the study area.

Material and methods
The original area of the Serra do Mar corresponds to 113,411 km², covering 364 municipalities (Table S1), inserted in the States of Rio de Janeiro, São Paulo, Paraná, Santa Catarina, and Rio Grande do Sul (Fig. 1), in Brazil's South and Southeast regions (WWF 2008).The vegetation classifications vary from Restinga to Highland Fields, passing through Ombrophilous Forest Low Montane, Montane, and Upper Montane Ombrophilous Forests, in addition to Deciduous and Semi-Deciduous Seasonal Forests (WWF 2008).The exact geomorphological delimitation of the Serra do Mar and its geographic limits is still not a consensus among authors.The revised limits of the Serra that include the state of Espírito Santo is a complex approach due to the considerable size of the area and for belonging to another ecoregion (RBMA 2008;WWF 2008).In addition, most studies for the Serra do Mar are carried out on a one-off basis (e.g., Barros et al. 2009;Cortines et al. 2011;Fernandes et al. 2022), which makes it difficult to effectively define the boundaries of the area.Thus, we chose to use the original limit of the Serra do Mar ecoregion, where the area is continuously delimited, and we excluded the disjoint part of the limits revised by WWF (2008) that includes the state of Espírito Santo.Therefore, we used 364 municipalities, and the area was divided into grids corresponding to the points of occurrence.

Spatial analysis-
We obtained 19 bioclimatic variables from the WorldClim database (available at: https://www.worldclim.org/)(Fick & Hijmans 2017).The average values of the variables were extracted from each grid of the Serra do Mar (Fig. 1).For the election of the best bioclimatic variables, we used a Factorial Analysis like Sobral-Souza et al. (2015).We choose as final variables, based on the highest absolute values, Annual Mean Temperature (BIO1), Precipitation of Wettest Month (BIO 13), Precipitation Seasonality (Coefficient of Variation) (BIO15), Precipitation of Driest Quarter (BIO17) and Precipitation of Warmest Quarter (BIO18).We also added collected the number of records to test the possible effect of sample bias.
Thus, to evaluate the correlation of the Passiflora composition in the Serra do Mar with the environmental gradients, we used the Canonical Correspondence Analysis (CCA) (Braak 1987).The species abundance matrix consisted of the number of individuals per grid.The matrix of environmental variables included the mean values of the five selected bioclimatic and the collection effort per grid.All previous analyzes were performed in the R program version 4.0.3(R Development Core Team 2020).
To graphically represent the result, we also performed a quadratic polynomial regression analysis to test the relationship of the richness (dependent variable) in the function of the bioclimatic variables of temperature and precipitation (independent variables), obtained from the CCA, using the Past 1.8.1 -Paleontological Statistics software (Hammer et al. 2008).In addition, we performed the overlap of species occurrence with the rasters layers of bioclimatic variables.

Natália Brandão Gonçalves Fernandes, Andreza Magro Moraes and Michaele Alvim Milward-de-Azevedo
Conservation -Calculations of Extent of Occurrence (EOO) and Area of Occupancy (AOO) were performed to evaluate the status of conservation of endemic Atlantic Forest species found in the Serra do Mar and compared to national and state lists as well as other studies, to point out the category that species already evaluated were inserted (Cervi 1997;CONSEMA 2003;Bernacci 2003;SMA-SP 2004;2016;Bernacci et al. 2005;Biodiversitas 2005;MMA 2008;2022;Durigon et al. 2009;CNCFlora 2012a.;CONSEMA 2003;2014;Milward-de-Azevedo & Fernandes 2021).The classification of the species as endemics of the Atlantic Forest and Tropical Rain Forest was based on the Flora do Brasil.The EOO and AOO calculations were performed using the GeoCAT tool (Bachman et al. 2011) (http://geocat.kew.org/editor), with a 2-km grid to calculate AOO, as recommended by the IUCN (2022).For species with populations restricted the Serra do Mar territory we suggest threat categories according to International Union Conservation of Nature (IUCN) Criteria B, based on the geographical distribution of the organisms (IUCN 2022).Concerning species with one or two points of occurrence, these parameters were not evaluated, because they do not form the polygon required to perform the calculations.These species were then classified as Data Deficient (DD).
Similarity -To evaluate the floristic composition of the Passiflora in the Serra do Mar, the UPGMA cluster analysis with the Jaccard index was used.The occurrence grids (0.50° × 0.50°) were compared using the Biodiverse 2.1 software, which allows cluster analysis of biogeographic data (Laffan et al. 2010).

Results
Spatial analysis -A total of 53 species were recorded for the Serra do Mar (Tab.1), from a total of 3,246 obtained records.Of these, 107 were discarded due to the inability to identify the material.The state with the most significant number of records was Rio de Janeiro (1,138 collections), followed by São Paulo, (1,110 collections), Santa Catarina (464 collections), Paraná (400 collections) and Rio Grande do Sul (134 collections), which obtained the smallest number of records.
Regarding species richness, the largest number was found in grids 7 and 18, in the areas more to the northeast of the Serra do Mar (the municipalities of Rio de Janeiro, Petrópolis, Teresópolis and Nova Friburgo, in the State of Rio de Janeiro) (Fig. 2a).At the same time, grid 18 showed the greatest number of efforts (in the municipalities of Rio de Janeiro, Petrópolis and Teresópolis, in the State of Rio de Janeiro), followed by grids 4 and 5, also to the northeast of the Serra do Mar (Fig. 2b).The most representative species in Serra do Mar considering the number of records, is P. amethystina Mikan with 345 occurrences, and P. porophylla Vell.with 256.Among the species recorded in the Serra do Mar, 22 are endemic to the Atlantic Forest and 12 are restricted to Tropical Rain Forest environments (Tab.1).
The quadratic polynomial analysis relating Passiflora species richness to altitude showed a decrease in the number of species as a function of the increase in altitude gradient (Figure 3 a-b) (R²=80%, p=0.001).The greatest richness was found between an altitude range between 1-199 m.a.s.l.presenting 40 species, followed by the range between 600-899 m.a.s.l. with 39 species.Furthermore, the lowest richness was recorded above 2,000 m.a.s.l. with only one species.
The results of the canonical correlation analysis show the highest eigenvalues of the first axis, representing a strong gradient.Despite having a lower value, the second axis was also highly correlated (Table 2).Five corresponding variables were found, where two of them showed the highest correlation, the Annual Average Temperature (BIO 1) highly explained on axis 1 and the Precipitation of the Driest Quarter (BIO 17) highly explained on axis 2 (Fig. 4).The Monte Carlo permutation test indicated a relationship between species composition and environmental variables in the ordination axes (p=0.001).

Natália Brandão Gonçalves Fernandes, Andreza Magro Moraes and Michaele Alvim Milward-de-Azevedo
The polynomial analysis of the bioclimatic variables showed that Annual Mean Temperature had a close relationship with the number of Passiflora species (R²= 80%, p=0.001) (Fig. 5 a-b), and the highest richness was recorded where the average is close to 20 ºC; at low mean temperatures the richness was lower, in addition, it also tended to decrease when this average was above 23 ºC.Another, of the Precipitation of the Driest Quarter, showed a negative correlation with the number of species (R²= 66%, p=0.01), indicating a decline in richness as the average precipitation of this period increases (Fig 6 a-b).Conservation -We found 17 threatened species among the endemic species that have already been evaluated (Table 3), P. setulosa Killip and P. ischnoclada Harms were the only ones found as "Extinct in the Wild" and "Critically Endangered" respectively.In addition, six species were presented in the "Endangered" categories and another six appeared as "Vulnerable", two species are listed as "Rare".We also suggest the categories of six species, two not yet evaluated, P. cervii as "Least Concern" for Extent of Occurrence and "Threatened" for Area of Occupancy, and P. elliptica Gardner as "Threatened" for both calculations.
Six species have their distribution restricted to the Serra do Mar territory (Tab.3): P. elliptica, P. farneyi Pessoa & Cervi and P. imbeana Sacco in the State of Rio de Janeiro, P. ischnoclada in the State of São Paulo, P. reitzii Sacco in the State of Santa Catarina and P. truncata Regel with occurrences recorded along the entire Serra do Mar.Four of the species that showed the greatest range of distribution (AOO and EOO values) in the Serra do Mar area were the most representative: P. actinia Hook.had the greatest range of distribution, followed by P. mediterranea Vell., P. truncata, and P. vellozoi Gardner.Meanwhile, seven species showed the smallest distribution range: P. junqueirae Imig and Cervi (2014) with a small range of distribution, then P. farneyi, P. filamentosa Cav., P. imbeana, P. ischnoclada, P. loefgrenii Vitta, and P. marginata Mast.
Similarity -The similarity analysis considering overall composition of Passiflora species in the Serra do Mar (Fig. 7a), generated a dendrogram with four groups and two isolated grids (Fig. 7b).The main group (Group 3, red) concentrates a larger of grids extends all along the Serra do Mar area and groups broad-distribution species, such as P. alata Curtis, P. amethystina Mikan, P. capsularis L., P. edulis Sims, P. haematostigma Mart.ex Mast., P. mediterranea Vell., P. porophylla Vell.and P. suberosa subsp.Natália Brandão Gonçalves Fernandes, Andreza Magro Moraes and Michaele Alvim Milward-de-Azevedo

Discussion
Spatial analysis -In this study, we found the greater richness of Passiflora in the areas comprised of the massifs inside the Serra do Mar and exactly where the main Conservation Units are (Tijuca National Park, Serra dos Órgãos National Park, Serra da Bocaina National Park, Serra do Mar State Park and Três Picos State Park).The so-called "mar de morros", in the Serra do Mar, shows that hillockes complexes represent the main landscapes in the Atlantic Forest (Graeff 2015).The distribution of species in an altitude gradient, defined by terrain characteristics, can show evident distinctions in diversity along a mountain (Grytnes 2003).Thus, the Serra do Mar terrain structure becomes a relevant factor in the distribution of climbing plants, since that Passiflora species significantly respond to altitude gradients (Ocampo et al. 2010;Moraes et al. 2018;2020;Fernandes et al. 2022).Fernandes et al. (2022), in a study concerning the species richness of Passiflora as a function of altitude in an area within the Serra Mar, found higher species richness in the middle part of the gradient.Meanwhile, in this study, we found a negative correlation, where richness decreased with increasing altitude.However, our results corroborate with the demonstrated by Moraes et al. (2020), in which there is Passiflora richness reduction with increasing altitude, and other lianas studies also found similar patterns, such Malizia et al. (2004), Hernández et al. (2012), Carvalho and Melis (2013) and Sainge et al. (2019).
The temperature was a relevant factor in understanding the distribution of Passiflora, we found the highest species richness where the average annual temperature varies between 19 ºC and 23 ºC.Places with very low temperatures present a deficit in the water balance because they do not maintain the same humidity as warmer areas and cause lianas to establish themselves better in places with higher temperatures (Stephenson 1990;Barry 1992;Hu and & Riveros-Iregui 2016).An example of this is those very low temperatures can impede the water conductance of liana species, which explains the low richness found in this study (Ewers 1985).
In contrast to the result found for temperature, the Precipitation of Driest Quarter showed a negative correlation with Passiflora species richness.We observed a decrease in the number of species as we moved south of the Serra do Mar, where the highest rainfall was recorded in the driest period.Even so, the areas in this region suffer less influence from the variation in relief (Pandolfo et al. 2002), while the north portion of the Serra do Mar sustains the areas humidity in these periods through orographic rains and fogs.
Studies such as the one by Arcova et al. (2021) and Cortines et al. (2011) point to fog as a water source for vegetation during all months of the year, which maintains the area's humidity in drier periods.In addition, water droplets suspended in the air and intercepted by the canopy commonly occur in Ombrophilous Forests and are called hidden rain, contributing to the maintenance of the area's water resource (Pereira et al. 2016).
Conservation -According to studies and Official National and State Lists (SEMA/ GTZ-PR 1995;Cervi 1997;CONSEMA 2003;Bernacci 2003;SMA-SP 2004;Bernacci et al. 2005;Durigon et al. 2009 Passiflora elliptica is solely restricted to the municipality of Rio de Janeiro, occurring in the areas of Tijuca National Park (Mezzonato-Pires et al. 2020) and no evaluation was made as to its status of endangerment.Thus, we indicate this species as "Endangered", due to the quantity of specimens found and the small number of locations where they were encountered following the IUCN category criteria (IUCN 2022).Of the species found which are endemic to the Atlantic Domain seven others, in addition to P. elliptica were not evaluated as being endangered nor were they mentioned on any national or state list or study.
Another species presented as rare, according to Cervi (1997), was P. reitzii, found in the forests on the slopes of the Serra do Mar and is represented in this study by only one record and no calculation was possible to determine its status of endangerment.However, the date of the last record is 1957, thus the species can be considered as CR*, a classification which indicated that, in addition to the species being "Critically Endangered", there is also the possibility of extinction, as there are no new records of collection over the last 30 years, ensuring its legal protection and conservation actions (IUCN 2022;Penedo et al. 2015;Martinelli & Moraes 2013).
Pointed out as "Endangered" for the State of Rio Grande do Sul, on the Official List of Native Flora Species in Danger Natália Brandão Gonçalves Fernandes, Andreza Magro Moraes and Michaele Alvim Milward-de-Azevedo of Extinction (SEMA 2014), P. actinia was one of the species with the largest number of records along the Serra do Mar (11 grids).However, most of these records are in Brazil's Southeast Region, which explains the presence of this species in the endangered categories in the South of the country.
In this study, we suggested four endemics species for the Serra do Mar: P. elliptica and P. ischnoclada recorded in the Ombrophilous Forest, P. farneyi in Restinga vegetation, and P. imbeana Sacco in ecotones between Highland Fields and Upper Montane Ombrophilous Forest.The heterogeneity of vegetation in the Serra do Mar ecoregion contributes to the occurrence of several endemic taxons (Garcia & Pirani 2005).Characteristics of these phytophysiognomes are determined by terrain formation and may differ according to the altitude and latitude (Ivanauskas & Assis 2009;IBGE 2012).
Similarity -In a study performed by Moraes et al. (2020), for the Serra da Mantiqueira, a similarity analysis of floristic composition indicated two well-defined groups (Northern and Southern) and an isolated group, which corroborates with the geomorphological division of the range (Machado-Filho et al. 1983).The Northern portion presents greater richness and a more heterogeneous composition of Passiflora species, while the Southern portion points to less variation of richness and greater homogeneity in composition (Moraes et al. 2020).Compared with the Serra do Mar, the Northeast and North areas, adjacent to the Northern and Southern portions of the Serra da Mantiqueira, showed material heterogeneity, also related to the greater richness of Passiflora species.The homogeneity in the floristic composition of the Serra do Mar is observed only in the areas more to the Central-West and Southwest of the range.
In this study performed by Moraes et al. (2020) 42 species were found in common with this study, most of them belonging to the Ombrophilous Forest at altitudes above 500 m.a.s.l, which can be explained by the proximity between the two ranges.The change in vegetation composition and questions covered related to terrain explain the heterogeneity of the species clusters in the Northeastern part of the Serra do Mar and point to the fact that the Ombrophilous Forest regions are made up of more homogeneous groups.This can be observed by the number of species restricted to Brazil's Southeast Region, where terrain is rugged.
Considerations -We presented the analysis of all Passiflora species recorded thus far in data bases for the Serra do Mar.Rainfall levels and terrain conditions are factors that influence the distribution of these species.In this study, we point to the fact that the Southeast Region showed the greatest richness and collection effort in the area, precisely in the regions where there is the greatest rainfall and variation in altitude.The State of Rio de Janeiro recorded the largest number of endemic species in the Serra do Mar.
A considerable portion of the species found in this study were presented on lists indicating endangered flora taxons.We also suggest including P. elliptica and P. reitzii, which are found only in Rio de Janeiro and Santa Catarina, respectively, on the lists of endangered species.It is also important to consider the species with insufficient data to make AOO and EOO calculations and revise them regarding their degree of threat.
The richness of Passiflora species, including those which are endemic to the Atlantic Forest, reveals the importance of the Serra do Mar area for conservation.Filling collection gaps in the South Region is also necessary to express and understand the diversity of these species in the range areas.
Diversity of the Passiflora L. in the Serra do Mar ecoregion and the relationships with environmental gradients, South and Southeast, Brazil Through the locations provided by the collectors, the geographical coordinates obtained were plotted superimposed on the shapefile of the Serra Diversity of the Passiflora L. in the Serra do Mar ecoregion and the relationships with environmental gradients, South and Southeast, Brazil do Mar (WWF 2008), in the QGIS software version 3.16 (QGIS DEVELOPMENT TEAM 2020), and occurrence was compared with the area boundaries.Within these aspects, we evaluated the distribution pattern of the species utilizing the same software with 0.50º × 0.50º squares, to analyze Passiflora richness and the collection effort in the Serra do Mar area.The altitude values were extracted from the information contained in the descriptions of the exsiccates.When the values were not presented, the altitude data were extracted from the point of occurrence, with the help of the WorldClim (Fick & Hijmans 2017) elevation raster, together with QGIS software version 3.16 (QGIS DEVELOPMENT TEAM 2020).

Figure 1 .
Figure 1.Original boundaries of the Serra do Mar ecoregion, South and Southeast Brazil, with 0.5° × 0.5° cells utilized in the analyses.

Figure 2 .
Figure 2. a. Richness of Passiflora L., in the Serra do Mar, in the South and Southeast Regions of Brazil; b.Collect effort of Passiflora L. species in the Serra do Mar, South and Southeast Brazil.

Figure 4 .
Figure 4. Results of Canonical Correspondence Analysis (CCA) with the records and five most correlated bioclimatic variables in the Serra do Mar, South and Southeast Brazil.BIO 1= Annual Mean Temperature; BIO 13= Precipitation of Wettest Month; BIO 15 = Precipitation Seasonality (Coefficient of Variation); BIO 17= Precipitation of Driest Quarter; BIO 18 = Precipitation of Warmest Quarter.
Figure 5. a. Annual Mean Temperature in the Serra do Mar, Brazil; b.Passiflora L. (Passifloraceae s.s.) species richness by annual mean temperature level in the Serra do Mar, South and Southeast Brazil.

Figure 7 .
Figure 7. a. Similarity map in the composition of Passiflora L. (Passifloraceae s.s.) species obtained by the UPGMA method with grids (0.50º x 0.50º) of occurrence in the Serra do Mar, Brazil; b.Similarity dendrogram generated through the Jaccard index.
; CNCFlora 2012a; Martinelli & Moraes 2013; List of Endangered Flora in the States of Rio Grande do Sul (CONSEMA 2003; SEMA 2014) and Santa Catarina (CONSEMA 2014; Martinelli et al. 2018; Milwardde-Azevedo & Fernandes 2021), at least 21 species found in the Serra do Mar area are classified in some level of threat.An example is P. farneyi, restricted to Restinga vegetation.It is worth highlighting that this type of vegetation has been suffering significant losses of habitat due to urban expansion and real estate speculation in the Restinga areas, further threatening endemic and restricted species (Leme 2000; Guerra 2005; Ribeiro & Oliveira 2009).

Table 1 .
Occurrence of Passiflora L. (Passifloraceae s.s.) species in the Serra do Mar, in the South and Southeast Regions of Brazil.PR = Paraná; RJ = Rio de Janeiro; RS = Rio Grande do Sul; SC = Santa Catarina; SP = São Paulo.
Passiflora foetida L. 11 21 RJ, SP, PR, RS, SC Highland Fields and Ombrophilous Forest Diversity of the Passiflora L.

Table 2 .
Results of the Canonical Correlation Analysis (CCA)with the five variables most correlated with the species richness of Passiflora L., in Serra do Mar, Brazil.

Table 3 .
Distribution and threat assessment of the endemic species of Passiflora L. from the Atlantic Rain Forest occurring in the Serra do Mar, Southeast and South, Brazil.Being: EOO = Extent of Occurrence; AOO = Area of Occupancy; CR = Critically Endangered; DD = Data Deficient; EN = Endangered; LC = Least Concern; NT = Nearly Threatened; VU = Vulnerable; EW = Extinct in the Wild; NE = Not Evaluated.

of the Passiflora L. in the Serra do Mar ecoregion and the relationships with environmental gradients, South and Southeast, Brazil restricted
distribution such as P. catharinensis, P. farneyi, P. filamentosa, P. imbeana and P. ischnoclada.It is possible to observe that the northeast portion of the Serra do Mar encompasses more heterogeneous areas regarding species composition.Meanwhile, the Center-West and Southwest areas of the range are characterized as presenting more homogeneity.Groups five and six presented a few species concerning other areas, which is why they turned into distinct groups.