Urban Parks in Curitiba as Biodiversity Refuges of Montane Mixed Ombrophilous Forests

: The assessment of the conservation status of urban forest remnants is essential for sustaining tree diversity in metropolitan cities. This study evaluated the composition and structure of forest remnants in four urban parks within the natural distribution of Montane Mixed Ombrophilous Forest in Curitiba (PR, Brazil). We allocated 66 plots of 100 m 2 and recorded all trees with perimeters at breast height ≥ 15 cm. A total of 1256 individuals from 44 botanical families and 117 tree species were identiﬁed, of which three are alien species. Tree density ranged from 1670 to 2095 trees ha − 1 , while the density of standing dead trees varied between 90 and 188 trees ha − 1 . We experienced values for the Shannon diversity index between 3.00 and 3.52 nats ind − 1 , which are comparable to the values for other non-urban forest remnants. Non-pioneer species represented 84.6% of the tree community (99 spp.), and 76% were zoochoric species (89 spp.). The research ﬁndings feed the discussion on the need of speciﬁc and integrated measures for the management of the urban parks reserved to the conservation of tree diversity in a speciﬁc phytoecological formation, while considering aspects of climate change and historical anthropisation.


Introduction
Worldwide, "big cities" provide shelter to more than half of the population, and the world population in urban environments is likely to increase to 68% by 2050 [1].Urban expansion is a phenomenon correlated with the loss and degradation of local vegetation, as it increases the pressure on natural resources and decreases the space available for flora and fauna, altering the composition and structural diversity of terrestrial ecosystems [2][3][4].On the other hand, life quality standards need to be ensured in urban environments for a sustainable future.
Urban forests provide ecological functions that are translated into services and benefits for humans, as they act as green infrastructure that can improve ecosystem health and improve city residents' quality of life.They provide habitat for fauna, are buffer zones for noise pollution and dust deposition, and contribute to carbon sequestration, water infiltration, and climate regulation [5].In the last 30 years, research on urban forests has concentrated on few cities in developed countries (e.g., Europe, North America, Australia) [6].More recently, an increasing amount of research has been observed in South America, for example, Medellín (Colombia), Concepción (Chile), and Curitiba (Brazil) [6].The persistence of urban forests is threatened by climate change; Esperon-Rodriguez et al., 2022 [7] have shown for 164 cities in 78 countries that more than half of tree species are experiencing temperature and precipitation conditions exceeding those typically experienced in their natural geographic ranges.
Curitiba is the largest city in southern Brazil, with nearly 2 million inhabitants (4027.04inhabitants km −2 ), and is the eighth most populated Brazilian city [8].The town is a member of the C40 (network of megacities committed to the cause of climate change), together with three other cities in Brazil.Curitiba is a world reference in urban planning and intelligent environmental initiatives [9].It stands on the national scene for its green urban areas spread over 30 municipal parks (approximately 83 ha, including non-forested areas) [10].In a radius of 40 km from the city centre, forest remnants from varied phytophysiognomies (e.g., subtypes of Mixed Ombrophilous Forests and of Dense Ombrophilous Forests) spread as fragments and are hardly extended over large areas (e.g., ≥1 km 2 ).These are restricted to protected forest areas in small farms or on steep slope terrain and alluvial areas (defined by national laws).
The Curitiba's natural vegetation is characterised by fields (steppe) and tree clusters of Mixed Ombrophilous Forests (MOFs) [11,12] which belong to the Atlantic Rainforest biome (holding 3% of known plant species on Earth) [13].MOFs are restricted to southern Brazil, with disjunctions in higher altitude areas in the Southeast region and neighbouring countries (Paraguay and Argentina) [14,15].This phytoecological formation also receives the name Araucaria Rainforest due to the presence of the species Araucaria angustifolia (Bertol.)O. Kuntze, one of the few Brazilian conifers and threatened to extinction [16].In addition, the tree diversity exceeds 350 species [17,18].The conservation of this phytoecological formation requires special attention, as the primitive remnants of the MOF are no more than 7% of its original area in Brazil [19].Likewise, climate change may further contribute to the extinction process of Araucaria angustifolia in the long-term future [20,21].
In Paraná, where most MOFs were located, well-preserved remnants cover no more than 1% of the original area [22].Remnants are highly fragmented and have undergone different levels of anthropisation, and exploitation, being in areas protected by law, such as public parks and private natural heritage reserves (e.g., "RPPN"), or in farmlands.The elevation of the MOFs determines the classification into the subgroups Alluvial, Sub-Montane, Montane, and High-Montane [23].
This study aimed to assess the composition and structure of the tree vegetation of four urban parks in Curitiba (PR, Brazil) falling into the phytoecological category of Montane Mixed Ombrophilous Forests.We hypothesised that, due to the anthropisation processes and urban drivers, the actual tree diversity in these urban parks is no longer representative of the MOFs.

Location and Study Sites
The study was conducted in Montane Mixed Ombrophilous Forest (MMOF) remnants located in four urban parks in Curitiba (Paraná, Brazil).This phytogeographic formation comprises a type of vegetation of the southern plateau, situated approximately between 400 and 1000 m a.s.l.[23].The city extends over 432.17 km 2 with a mean altitude of 934.6 m a.s.l.(S −25 • 25 40", W −49 • 16 23").The climate is humid temperate (Cfb) according to the Köppen classification [24], with no dry season and cool summer, and is the coldest Brazilian capital.The average annual temperature is 17.7 • C, with averages around 10 • C from June to August, and 27 • C from December to February [25].The average annual precipitation is 1800 mm [25].The studied municipal parks were randomly selected from subgroups defined by the park location on cardinal directions (north, south, east, and west parts of the city), while limited funds were available for this research.The parks cover landscapes with varied characteristics, distinct urbanisation and occupation background (Figure 1): Parque Natural Municipal Barigui (Barigui park, BARI); Parque Natural Municipal de Lazer General Iberê de Mattos (Bacacheri park, BACA); Parque Municipal da Barreirinha (BARE); and Zoológico Municipal (City Zoo, ZOOM).Moreover, the parks differ regarding total forest area, fragmentation, infrastructure, management, and establishment (Table 1); the variability of the vegetation composition and structure are perceptible in Figure S1.The size of the remnants was estimated with QGIS Version 3.16.11,using the database of the Institute of Research and Urban Planning of Curitiba, IPPUC (2022).
(north, south, east, and west parts of the city), while limited funds were available for this research.The parks cover landscapes with varied characteristics, distinct urbanisation and occupation background (Figure 1): Parque Natural Municipal Barigui (Barigui park, BARI); Parque Natural Municipal de Lazer General Iberê de Mattos (Bacacheri park, BACA); Parque Municipal da Barreirinha (BARE); and Zoológico Municipal (City Zoo, ZOOM).Moreover, the parks differ regarding total forest area, fragmentation, infrastructure, management, and establishment (Table 1); the variability of the vegetation composition and structure are perceptible in Figure S1.The size of the remnants was estimated with QGIS version 3.16.11,using the database of the Institute of Research and Urban Planning of Curitiba, IPPUC (2022).

Field Inventory
The field inventory was conducted by simple random sampling, using temporary units and 0.01 ha sample plots (10 × 10 m 2 ) [26].The number of samples varied according to the forest area of each park, totalising 66 plots: 20 plots for ZOOM, 20 plots for BARI, 16 plots in BARE, and 10 plots in BACA (Table 1).Considering the reduced size of the forest remnant in BACA, sampling units were also installed in the associated Private Natural Heritage Reserve, with a forest area of 0.5 ha contiguous to the park.
In the plots, all tree individuals (living and dead) with a circumference at breast height (CBH) ≥ 15 cm (at 1.30 m from the ground) were measured with a measuring tape.The data collected for each tree were: species; CBH; total tree height; phytosociological position (understorey, intermediary, canopy, emergent); phytosanitary status (dead, senescent, physically damaged, or "herbivore damage or defoliation").Tree height was measured with a Vertex Laser Geo 3D (Haglöf Sweden ® , Långsele, Sweden).The vegetation strata were visually evaluated using the local maximum tree height as a reference, relativising the compartmentation of tree crowns at the plot level.Tree basal area was calculated for living and dead trees.
The tree species identification was conducted on the field using their dendrological characteristics, such as leaves, bark properties, and life forms.A branch sample was collected for unidentified species and herborized for comparison to the exsiccata of the herbarium "Escola de Florestas de Curitiba" (EFC) with their experts.These newly produced exsiccata were further processed and added to the collection of EFC.The taxonomic classification followed the system of Angiosperm Phylogeny Group IV [27], and the valid nomenclature has been checked in Flora and Funga of Brazil [28].

Tree Composition
To analyse the tree composition, we used the number of families, genera, and tree species found in the survey of each fragment.The life form (shrub, tree, liana, and tree fern), ecological group (pioneer and non-pioneer), and dispersion syndrome (anemochoric, autochoric, and zoochoric) were based on the information provided by REFLORA (Flora do Brazil) [28] and Swaine and Whitmore [29], Budowski [30] and Van der Pijl [31].We used the red list of the "Centro Nacional de Conservação da Flora" (National Centre for Flora Conservation) [32] to determine the conservation status of species.
The tree species composition was analysed through the Venn Diagram [33].The graph emphasises the number of species shared by the sampled forest remnants while showing the number of exclusive species to each of the four remnants.
The species accumulation curve verified the sampling sufficiency, in which the total number of species for each new sampling unit included in the study is presented graphically [34].We did 100 randomisations in the order of the samples, adopting a confidence interval of 95%, and included in the graphical representations the values of richness estimated by Jackknife 1 with the software EstimateS Version 9.1 [35].The accumulation curve and the Jackknife method represent the species diversity in the assemblage and are reasonable estimates of total richness [36].
Considering that the diversity indices portray the distribution of abundance among the different species in a community, we computed the Shannon-Wiener diversity index (H') [37] as a standard measure for comparison with other studies and vegetation types.Likewise, we calculated the species dominance using Simpson's index (D) [38], the species richness using Margalef's index (M) [39], and the species uniformity using Pielou's index (J) [40].

Phytosociological Parameters
The phytosociological parameters estimated were the absolute and relative values of tree density; dominance; frequency; and the importance value; they were determined according to the references [41,42].To calculate absolute dominance (ADo), we used the stand basal area of a taxon per unit area.In contrast, for relative dominance (RDo), we used the ratio of the absolute dominance of a given taxon by the sum of all absolute dominances.The absolute density (AD) indicates the number of individuals of a given species per unit area, and the relative density (RD) is the number of individuals of a given taxon concerning the total number of individuals sampled.The absolute frequency (AF) indicates the percentage of plots that present a particular taxon, and the relative frequency (RF) is calculated based on the percentage of absolute frequency to the total frequency, representing the sum of all absolute frequencies.
The importance value index (IV) is a synthetic index used to characterise each phytophysiognomy by its most representative species.The IV was calculated using the relative values of dominance, density, and frequency of species, applying the following equation: IV = RDo + RD + RF [41].The standing dead individuals were grouped in the category "dead".This group were considered for calculating the absolute values of density and dominance.The program FITOPAC Version 2.1 [43] was used for the calculations.
The species composition was assembled according to the ecological groups and dispersal syndromes.Likewise, tree individuals were grouped according to their observed position in the strata (understory, intermediate, canopy, and emergent) to define the vertical structure.We calculated the percentage of individuals in each subgroup.The horizontal structure of the forest remnants was evaluated based on the diameter distribution of living trees and organised into diameter classes with a 10 cm interval [44].

Tree Species Composition
A total of 1256 tree individuals were recorded in the sampled area of 0.66 ha encompassing the four studied urban parks with fragments of Montane Mixed Ombrophilous Forest.The local biodiversity was represented by 44 botanical families, 83 genera, and 117 tree species (Table A1).The observed richness was lower than the richness estimated by the Jackknife 1 estimator for all urban remnants studied, ranging from 30.8% (BACA) to 38.9% (BARI), indicating that species richness may be higher than sampled (Figure 2).The urban park with higher species richness was BARI (66 spp.), followed by ZOOM (57 spp.), BARE (51 spp.), and BACA (38 spp.).As shown in Figure 3, the highest number of unique species (22 spp.) was found in ZOOM, followed by BARI (16 spp.).On the other hand, the parks with the highest similarity in tree composition were BARI and BARE, with 13 shared species.The remnants of BARE and BACA shared only one species.Considering the four studied MOF remnants, the botanical families with the most expressive species richness were Myrtaceae (23 spp.), Lauraceae (10 spp.), Fabaceae (9 spp.), Salicaceae (8 spp.), and Rubiaceae (7 spp.).These families together represented 48% of the total observed species.Moreover, 84.6% of the tree species are non-pioneer, and 76.1% have zoochory as seed dispersal strategy.Only 4.3% of the species are shrubby, while 95.7% are trees.Among the arboreal species, 9.4% were common to all urban remnants surveyed (Figure 3  The urban park with higher species richness was BARI (66 spp.), followed by ZOOM (57 spp.), BARE (51 spp.), and BACA (38 spp.).As shown in Figure 3, the highest number of unique species (22 spp.) was found in ZOOM, followed by BARI (16 spp.).On the other hand, the parks with the highest similarity in tree composition were BARI and BARE, with 13 shared species.The remnants of BARE and BACA shared only one species.

Diversity Indices
The species diversity descriptors for all studied parks are summarised in Table 2.All parks had a Shannon diversity index (H') higher than 3 nats ind −1 , with BARI presenting the highest value (3.52 nats ind −1 ).All the remnants showed an equitable distribution among the sampled species, with values of Pielou's index varying between 0.81 and 0.85.Moreover, the Simpson's diversity index (D) exposed low values in all remnants, indicating a low dominance of individuals of some tree species.
The Margalef index (M) estimates diversity based on the numerical distribution of individuals of different species.Thus, the highest values were found in BARI and ZOOM (10.93 and 9.71, respectively), indicating a greater diversity in these remnants.The urban park with higher species richness was BARI (66 spp.), followed by ZOOM (57 spp.), BARE (51 spp.), and BACA (38 spp.).As shown in Figure 3, the highest number of unique species (22 spp.) was found in ZOOM, followed by BARI (16 spp.).On the other hand, the parks with the highest similarity in tree composition were BARI and BARE, with 13 shared species.The remnants of BARE and BACA shared only one species.

Diversity Indices
The species diversity descriptors for all studied parks are summarised in Table 2.All parks had a Shannon diversity index (H') higher than 3 nats ind −1 , with BARI presenting the highest value (3.52 nats ind −1 ).All the remnants showed an equitable distribution among the sampled species, with values of Pielou's index varying between 0.81 and 0.85.Moreover, the Simpson's diversity index (D) exposed low values in all remnants, indicating a low dominance of individuals of some tree species.The Margalef index (M) estimates diversity based on the numerical distribution of individuals of different species.Thus, the highest values were found in BARI and ZOOM (10.93 and 9.71, respectively), indicating a greater diversity in these remnants.

Phytosociological Parameters
Phytosociological parameters are presented in Table 3 and refer to the fifteen species with the highest IV for each park.In ZOOM, the most important species are large trees of moderate to low growth (Araucaria angustifolia and Myrcia hatschbachii).A similar situation to ZOOM was observed in BARI, as Palicourea sessilis and A. angustifolia were the most important species.At BACA, the species Mollinedia clavigera, which is a shrub, had the most significant importance value, followed by Erythrina falcata, a large-size non-pioneer tree species whose presence was only recorded in this park.On the other hand, in BARE, the most important species are Casearia sylvestris and Myrcia splendens, which can occupy different environments and generally show significant variation in height and can occur in shrub and tree forms.Still, in BARE, the presence of Pittosporum undulatum, an alien species, was observed among the fifteen most important species.Considering the eleven tree species shared by all remnants, only the species Casearia sylvestris was highlighted in all the remnants researched.The species Allophylus edulis, Araucaria angustifolia, Cabralea canjerana, Casearia decandra, and Ocotea nutans stood out in this ranking in three remnants.
The species C. sylvestris can occupy different environments, and shows great variation in relation to height, being able to occur in shrub and arboreal forms, presenting zoochory dispersion mainly by birds.The species preferentially occurs in altered primary forests [45,46] and its group is discussed by different authors [45,[47][48][49][50], being classified in this study as a non-pioneer species.
In BARE, C. sylvestris presented the highest IV (36.6%).The species also presented the highest IV in the urban remnant studied by Rondon Neto, et al. [51], in Curitiba.Despite being well distributed in the remaining part of the BARE (100% AF), the species showed an absolute dominance of 4.19 m 2 ha −1 .This species had an IV of 9.77% in the remnant of BACA, 7.06% in the ZOOM, 6.05% in BARI.The lowest RF of C. sylvestris was in the ZOOM, with 35% and ADo of 0.64 m 2 .ha −1 .
The species A. angustifolia presented the highest IV in ZOOM (33.9%) and the second highest in BARI (20.4%).Although A. angustifolia is the most characteristic species of the studied phytophysiognomy, appearing with the highest IV in the other works [51][52][53][54][55][56][57][58][59], this species was not positioned in the BARE ranking.In BACA, A. angustifolia had an IV of 10.7% and was ranked 11th, behind Mollinedia clavigera.
The species of shrubby habit M. clavigera presented FR of 8.1% and was well distributed in the sampling units in BACA, standing out with the highest IV (34.3%), followed by the species of tree habit Erythrina falcata (IV = 19.1%).Another species of shrub habit, Rudgea jasminoides, had an importance value index of 11.95% in the remainder of BARI.Two distinct pioneer species with anemochory dispersion were Jacaranda puberula and Vernonanthura discolor, with an importance value index of 10.5% and 7.92%, respectively.
Myrtaceae presented the highest number of species, confirming it as one of the leading families among the Neotropical angiosperms [60].Yet, only Myrcia hatschbachii was found in all studied parks.However, the species presented the highest IV only in ZOOM (24.4%), also with the highest AD (210 trees ha −1 ) and ADo of 3.6 m 2 ha −1 , followed by Eugenia uniflora (IV = 19.5%).
The species Myrcia splendens (Myrtaceae) had the second highest IV (22.64%) in BARE.Still in this park, the pioneer species Clethra scabra and Jacaranda puberula appear with IV of 18.72 and 8.64%, while the invasive alien species, also pioneer, Pittosporum undulatum ("pau-incense") presented RF of 1.1% and RD of 2.2%, standing out among the fifteen most important species in BARE.In this park, the presence of the alien species Hovenia dulcis, which is not among the fifteen species with the highest IV, showed AD of 18.8 trees ha −1 and AF of 18.8%, indicating that it is not well distributed in the sampled area.In BARI, the presence of an alien species, Eriobotrya japonica, was also recorded, but with a low density of individuals (5 trees ha −1 ).

Vegetation Structure
The density of individuals in the urban remnants surveyed ranged between 1670 and 2095 trees ha −1 (Table 4), with the highest density found in BARI.The BACA remnant presented the highest stand basal area (69.39 m 2 ha −1 ), while the highest average tree height was observed in ZOOM (10.80 m).In the BARE relict, the species of the ecological group non-pioneer presented the highest density of individuals compared to the others, corresponding to 294 trees ha −1 .The total basal area of the species in this ecological group represents 26.24% (12.45 m 2 ha −1 ) of the total sampled.
Regarding the analysis of standing dead individuals, BARE and BARI presented the highest AD of 187.5 and 180 trees ha −1 , respectively.The standing dead trees were well distributed among the sampling units.In BARI, the basal area of standing dead individuals was 13.14% of the total sample, the highest percentage among the areas studied.
The tree diameters ranged from 4.8 cm to 98.7 cm, with the largest average diameter found in the ZOOM (17 cm), while BARI and BARE had the highest mean diameters for standing dead trees of 16 and 11 cm, respectively.These values were higher than the average of the living community in these remnants.
The tree diameter distribution had the traditional inverted "J" pattern (Figure 4), where the total number of individuals with a minor diameter is the largest one.This pattern was also presented in other studies of Ombrophilous Mixed Forest [61][62][63].The tree diameter distribution had the traditional inverted "J" pattern (Figure 4), where the total number of individuals with a minor diameter is the largest one.This pattern was also presented in other studies of Ombrophilous Mixed Forest [61][62][63].The vertical structure of the MMOF remnants is summarised in Table 5.The urban remnant of BARI recorded the highest heights in the emergent (28.8 m), canopy (26.0 m), and intermediate (19.0 m) strata, while the highest height for the understory was observed in the remnant of the ZOOM (14.0 m).The species Araucaria angustifolia was the only one that was present composing the emerging stratum in all the remnants.In ZOOM, Cedrela The vertical structure of the MMOF remnants is summarised in Table 5.The urban remnant of BARI recorded the highest heights in the emergent (28.8 m), canopy (26.0 m), and intermediate (19.0 m) strata, while the highest height for the understory was observed in the remnant of the ZOOM (14.0 m).The species Araucaria angustifolia was the only one that was present composing the emerging stratum in all the remnants.In ZOOM, Cedrela fissilis was also observed in the emerging stratum.
In the canopy layer of ZOOM, Myrcia hatschbachii had the highest percentage of individuals (15.24%), followed by Eugenia uniflora with 44.4%, and a similar situation was observed in the intermediate stratum (17.4% and 8.3%, respectively).In the BACA, Matayba elaeagnoides and Luehea divaricata stood out in the canopy stratum with 16.7% and 13.3% of individuals, respectively, while Allophylus edulis had 21.4% of the trees in the intermediate stratum.In BARI, Palicourea sessilis had the highest percentage of individuals in the canopy (20%) and intermediate strata (17.7%).In BARE, the species Casearia sylvestris stood out both in the canopy (13.64%), the intermediate (29.1%), and understory (20.2%) stratum.In the understory, the tree species Allophylus semidentatus had 20% of the individuals in ZOOM.The shrub species Mollinedia clavigera and Rudgea jasminoides had the highest percentages of individuals in BARI and BACA, 41.18% and 18.50%, respectively.Where: zoochory (ZO); anemochory (AN); autochory (AU); pioneer (PI); and not pioneer (NP) in percentage of tree species.* Presence of individuals whose dispersal syndrome was not identified.
When analysing the composition of the four parks, the species Casearia sylvestris and Cabralea canjerana, which can vary between tree and shrub habit, were observed composing three strata (canopy, intermediate, and understory) of the remnants ZOOM, BARI, and BARE.The average height of individuals of these species was 13.6 m (C.canjerana) and 12.6 m (C.sylvestris).
The predominance of zoochory as dispersion syndrome and the ecological group of non-pioneer species was noted in all strata of the urban parks.BARE had the highest presence of pioneer species, mainly composing the remaining canopy.As for the dispersion syndrome, it is observed that BACA concentrated the highest number of autochory species distributed among the strata.BARI had the highest number of anemochory species, mainly in the understory.Finally, standing dead individuals had average height ranges between 5.2 and 6.1 m, with the highest and lowest value recorded in ZOOM (13.0 m and 1.5 m).

Discussion
Although more than 350 species have been reported to belong to the MOF [17,64], many species are rarely found in phytosociological studies at a single locality (Table 6).Kozera et al. [65], when studying the phytosociology of the arboreal component in BARI, reported 97 species, the highest richness recorded in comparison with other studies in the metropolitan region of Curitiba.Similarly to other studies in Curitiba, Rondon Neto et al. [66] observed 77 species and Mielke et al. [67] 57 species, while Geraldi et al. [53] reported 88 species in Tijucas do Sul and Eisfeld et al. [68] found 65 species in Campo Largo.
In our study, we observed 117 species throughout all parks.In BARI, 66 species were present, a lower species number than that observed by Kozera et al. [65] in 2006.Such differences may relate to the different sampling methods (point-centred quarters vs. plots), and potential vegetation changes.In BARE, the observed richness was lower than that observed by Mielke et al. [67] in 2015 (51 vs. 57 spp.).
However, the values recorded by these authors are within the range estimated by Jackknife 1.The differences may be due to the total sampling size, the locations, and the methods applied, as the available resources and funds ruled our sampling efforts.However, other works on MOF in the southern region of Brazil have adopted total sampling areas smaller than one hectare (Table 6).
Studies conducted outside urban centres to analyse the tree component showed lower species richness than observed in this study (with the same inclusion criterion, DBH ≥ 5 cm or CBH ≥ 16 cm).In the state of Santa Catarina (Brazil), Passos et al. [69] recorded 106 species in the Araucarias State Park, while Higuchi et al. [58] reported 95 species in the Municipality of Campos Novos.In Rio Grande do Sul, Rondon Neto et al. [51] observed 80 species in a MOF remnant in Criúva.In Paraná state, Martins et al. [61] and Silvestre et al. [70] reported 47 and 65 species, respectively.
The Shannon index is a means of comparing different areas widely used by the scientific literature, whose common values are between 1.50 and 3.50 and rarely exceed 4.0 [36].However, the sampling effort and the inclusion criterion are essential aspects to be considered, as they can influence its values [36].In other studies in Curitiba and the metropolitan region [65,66,68], the values for the Shannon index were similar to those observed in ZOOM and BARI.Shannon index values for BACA and BARE were higher than those observed by Seger et al. [52], who reported values of 2.37 nats ind −1 and 2.18 nats ind −1 .Similarly, the values were higher than those of the National Forest São Francisco do Sul (RS), with 1.2 nats ind −1 and 2.2 nats ind −1 [55,71].
Rondon Neto et al. [66] suggested that the variation found for Shannon index values in MOF areas may be related to environmental heterogeneity (e.g., topographic, edaphic, anthropic levels, and sunlight incidence).Despite the natural variation in landscapes and MOF forests or fragments, it is possible to note that the studied forest remnants presented moderate to high diversity for the tree component.
The comparative analysis of species richness between the remnants and MOF did not consider the successional stage or aspects related to the landscape characteristics (e.g., slope and aspect) or soil conditions.However, it focused on the tree species' diversity and its stratification.Thus, the lower number of species found in the studied parks is explained by the fact that urban remnants located in public areas are more subjected to anthropic influence, either by the occupation history, previous land use or even by actual and varied urban conditions.
As for the specific richness of the botanical families, it was found that the most frequent families observed together had approximately half of the species observed.The leadership of the families Myrtaceae, Lauraceae, Fabaceae, Salicaceae, and Rubiaceae in the tree composition agrees with the studies conducted in this forest typology in many localities [51,52,55,56,58,59,61,63,[68][69][70]72,73].
We observed a wide range of unique tree species occurring exclusively in each park (Figure 4), with a total of 60 unique species (from 11 to 22 unique species per park).On the one hand, this shows that the parks are complementary to each other in the conservation of tree diversity, as they safeguard more species together.On the other hand, this exposes a risk of biodiversity loss if local populations are not maintained and regeneration is not ensured.The physical distance from the parks would also not allow the natural regeneration or migration of these unique species.Thus, the local vegetation can only be sustained with active measures in the parks (e.g., planting new individuals of key species and enrichment planting).Kersten and Galvão [75] pointed to the dominance of pioneer and shade tolerant species in remnants of the Araucaria Forest in Curitiba (Guabirotuba Formation), such as Casearia sylvestris, Allophylus edulis, Luehea divaricata, Solanum sanctae-catharina, Styrax leprosus, and Matayba elaeagnoides.Similarly, the non-pioneer and light-dependent species Ocotea puberula was predominant.We recorded the presence of all these species in the sampled areas (Table 2).
In this study, tree species considered less frequent were found [75].These were, for example, Podocarpus lambertii (BARI and ZOOM), Curitiba prismatica (BARI and BARE), and Jacaranda puberula (all parks).Araucaria trees were recorded in all surveyed areas, with the highest number of individuals in the ZOOM (90 trees ha −1 ).Ilex paraguariensis ("erva mate") was not found in Bacacheri but in others.In addition, we observed Lafoensia pacari (Lythraceae) in the ZOOM and Ocotea nutans (Lauraceae) in all parks, which are described as rare species by the Inventário Florístico Florestal Florestal do Estado de Santa Catarina [18].The incidence of rare and less frequent species in the studied parks indicates a potential for conserving tree diversity and the need for a specific management plan to perpetuate the existence of critical species.
Management guidelines should consider controlling invasive alien species in the forest remnants of the urban parks, assuming they are supportive conservation units.Invasive alien species were reported in other studies in urban forest remnants in Curitiba [66][67][68].Biondi and Muller [76] recorded 13 invasive alien species in five parks in Curitiba, noticing that the presence of adult individuals may precede the creation of these conservation units.Exotic species may occur in other remnants though not recorded in the sampling units, but also considering parks' anthropisation processes and story.Moreover, Mielke et al. [67] observed a high frequency of P. undulatum in the BARE remnant, which is associated with human influence since part of the land was used for agriculture before the park's establishment.As alien species take advantage of disturbed areas, this indicates that the remnant of BARE is the least conserved, compared to the other studied areas, once it has the highest density of alien species.
Regarding the horizontal structure, we found tree densities that are many times higher than in other MOF remnants located in protected areas, such as the "Parque Estadual das Araucárias" (SC, Brazil), with a density of 1368 trees ha −1 [64], and in FLONA (Irati, PR) with 567 trees ha −1 [63] and 561 trees ha −1 [59].However, the basal area of the studied urban forests was smaller than that observed in the FLONA in São Francisco de Paula (RS, Brazil; 79.05 m 2 ha −1 ), although the density of 1445 trees ha −1 was smaller than that observed in the studied remnants [55].Moreover, compositional and structural variations between remnants of MOF are highlighted in Table 6.Environmental conditions (e.g., precipitation and insolation), topography, altitudinal gradient, soil type, vegetation successional stage, and anthropisation history can influence the heterogeneity of the areas since the MOF is formed by multiple associations and groupings [59,63].
During the forest succession process, several biotic and abiotic factors influence the competitive capacity of individuals of different species, influencing mortality rates.Tree death within remnants contributes to various ecosystem functions, such as nutrient cycling, soil conservation, and maintenance of fauna diversity (e.g., saproxylic organisms) [77][78][79].When summing the values of the four parks, it was observed that the total number of standing dead individuals corresponded to 8.2% of the sample (156 dead trees ha −1 ).BARI and BARE presented the highest density of standing dead trees, or 180 and 187 dead trees ha −1 , respectively.Similarly, these remnants recorded the highest densities of individuals of pioneer species, with 294 trees ha −1 in BARE and 200 trees ha −1 in BARI, indicating that the vegetation is on a more intense successional process.Evaluating the ecological and successional aspects of an urban MMOF remnant, Machado et al. [80] found that a reduction in tree density was more intense in the pioneer species, considering the behaviour of all involved ecological groups.
Inventory studies of the tree diversity and structure are knowledge-and labourintensive; thus, proper funding must be allocated to this type of research.Additional studies regarding forest regeneration could provide essential insights into understanding species' successional processes and dynamics, for example, whether new species are entering these remnants.Likewise, studying fauna populations (e.g., insects and birds) may help elucidate interactions and interrelations of fauna and flora regarding sustaining biodiversity.Concerning the vegetation structure, function, services, and benefits, a detailed assessment of the three-dimensional arrangement of the vegetation (with terrestrial LiDAR) would support further analysis dealing with the vertical profile and the identification of tree habitats [81].Although dead trees provide shelter and food for animal species, there is a lack of studies on dead wood and the diversity of saproxylic organisms in urban and non-urban systems in Curitiba and the whole of South America.Finally, further investigation of shading effects would help finding relationships between forest regeneration and vertical stratification [82], as well as that tree pruning strategies could be evaluated for the regulation of insolation and microclimate within the remnants [83].
Strategies to overcome the effects of climate change on urban vegetation are missing in Curitiba; park management plans dealing with the vegetation are only available for BACA and BARI [10].These are supported by the national law 9985/2000 and are devoted to the management of conservation units.High-altitude areas were identified [20,21] as most relevant for the maintenance of Araucaria trees, which meets the conditions found in Curitiba to a great extent.In this respect, the connection of urban forest remnants through corridors will play a major role in the genetic flow between individuals.In Brazil, natural spaces in urban environments are objects of monetary valuation incorporated into the real estate market, as they provide ecological services [84][85][86].In Curitiba's case, energy efficiency and landscape issues are being aligned to real estate supply for the urban upper-middle class [85].

Conclusions
The four studied urban parks have comparable tree diversity and richness in their MMOF remnants to many other MOFs in non-urban conditions.There was the predominance of non-pioneer species with zoochoric dispersal on the tree composition.Uncommon, unique, and alien tree species were found, indicating the need for active management with the goal of preserving the tree community.Traces of historical anthropisation (e.g., wood exploitation) were present, as Araucaria trees were not always present in the different stratum of the parks.Moreover, results highlighted the importance and need for park management plans strategically, including integrated measures (with other parks) to sustain the tree diversity of the urban environments of metropolitan cities. Subsequent studies on the tree regeneration and fauna will help highlighting the dependencies of populations and the likely perpetuation of the MMOF remnants on urban parks in Curitiba.Where: CST-conservation status (NE-species not assessed, LC-least concerned, EN-endangered, VU-vulnerable, AL-alien species); LF-life form (T-tree, TF-tree fern, SH-shrub, P-palm tree, DRA-dracenoid); DS-dispersal syndrome (ZO-zoochory, A-anemochory, Au-autochory); EG-ecological group (PI-pioneer, NP-non pioneer).

Figure 1 .
Figure 1.Study locations in Curitiba (PR, Brazil) with park boundaries in red and MMOF areas in lime green.

Figure 2 .
Figure 2. Observed species richness and Jackknife 1 of the tree component in MMOF remnants of urban parks in Curitiba (PR, Brazil).

Figure 2 .
Figure 2. Observed species richness and Jackknife 1 of the tree component in MMOF remnants of urban parks in Curitiba (PR, Brazil).

Figure 2 .
Figure 2. Observed species richness and Jackknife 1 of the tree component in MMOF remnants of urban parks in Curitiba (PR, Brazil).

Figure 3 .
Figure 3. Number of exclusive and shared tree species in MMOF remnants of urban parks in Curitiba (PR, Brazil).

Figure 3 .
Figure 3. Number of exclusive and shared tree species in MMOF remnants of urban parks in Curitiba (PR, Brazil).

Figure 4 .
Figure 4. Diameter distribution of the live tree component in MMOF remnants in urban parks in Curitiba (PR, Brazil).

Figure 4 .
Figure 4. Diameter distribution of the live tree component in MMOF remnants in urban parks in Curitiba (PR, Brazil).

Table 1 .
Description of the studied urban parks with MMOF remnants (Curitiba-PR, Brazil).

Acronym. Park Name Establishme nt Location Distance to City Centre Park Area Forest Area
Figure 1.Study locations in Curitiba (PR, Brazil) with park boundaries in red and MMOF areas in lime green.

Table 1 .
Description of the studied urban parks with MMOF remnants (Curitiba-PR, Brazil).

Table 2 .
Diversity indices for the tree component of MMOF remnants in the urban parks (Curitiba-PR, Brazil).

Table 3 .
Phytosociological parameters of key tree species in the urban parks (Curitiba-PR, Brazil).

Table 5 .
Vertical structure of the MMOF remnants in urban parks (Curitiba-PR, Brazil).

Table 6 .
Phytosociological studies of the tree composition of MOF remnants in the states PR, SC, and RS (Brazil).

Table A1 .
Botanical families and tree species found in MMOF remnants of four urban parks in Curitiba (PR, Brazil).