Where are they from and where are they going? Detecting areas of endemism, distribution patterns and conservation status of the order Spirostreptida in Brazil (Diplopoda, Juliformia)

Millipedes are well-known for their limited dispersal abilities with species restricted to mountains, islands, and patches of forest being important models for formulating and testing biogeographic hypotheses. The order Spirostreptida is composed of nearly 1300 species distributed across the Afrotropical, Oriental, and Neotropical regions. The order is divided into the suborders Cambalidea and Spirostreptidea with the families Pseudonannolenidae (Cambalidea) and Spirostreptidae (Spirostreptidea) occurring in Brazil. To date, there have been no studies on the biogeography of Neotropical Spirostreptida. We employed a multi-approach analysis with Parsimony Analysis of Endemicity, Endemicity Analysis, and Infomap, to detect areas of endemism and patterns of distribution of the order in Brazil. Six areas of endemism are proposed for the 133 Brazilian species: Northern Serra Geral, Southeastern mountain ranges, Cerrado and Atlantic Forest zone, Eastern Cerrado and Serra do Espinhaço, Pantanal, and Southern Amazon and Cerrado zone. Most endemic areas fall within the Atlantic Forest, which has been previously shown to be an important area of endemism for many invertebrate taxa. The richest biomes are the Atlantic Forest with 75 species and the Cerrado with 55 species, while the least rich biomes are the Caatinga with six species and the Pampa with three species. Although the southeastern region of Brazil has the highest species richness, it also has the highest concentration of threatened species. Based on International Union for Conservation of Nature (IUCN) criteria, most Brazilian Spirostreptida are either endangered or critically endangered, with the highest concentration of endangered taxa occurring in the Atlantic Forest biome.


Introduction
The recognition of areas of endemism is essential for the understanding of historical biogeography and for determining conservation priorities of specific regions (Myers et al. 2000;Casagrande and Grosso 2013;Martínez-Hernández et al. 2015). Areas of endemism correspond to regions which share a common history mainly as the result of historical events (e.g. vicariance) or ecological factors acting to delimit the distribution of local species (Nelson and Platnick 1981;Rosen 1988;Hovenkamp 1997;Szumik et al. 2002Szumik et al. , 2012Morrone 2009). Different methods, such as Parsimony Analysis of Endemicity (PAE), Endemicity Analysis (EA), Brooks Parsimony Analysis (BAP), Paralogy-free subtree analysis, and Vicariance Events Analysis, have been used to identify such areas (Rosen 1988;Morrone 1994;Szumik et al. 2002Szumik et al. , 2012Hausdorf and Hennig 2003;Szumik and Goloboff 2004;Aagesen et al. 2013;Edler et al. 2017). Numerous studies have been conducted using these methods to detect areas of endemism in the Neotropical region based on the distribution of insects (Domínguez et al. 2006;Sigrist and Carvalho 2009;Ferrari et al. 2010;Silva and Vaz-de-Mello 2020), spiders (Sigrist and Carvalho 2009;Oliveira et al. 2015), and harvestmen (DaSilva et al. 2017).
Millipedes (class Diplopoda) are known for their low vagility and limited distributions with species often restricted to mountains, islands, and patches of forest (Golovatch and Kime 2009;Enghoff 2015). In some cases, species occur exclusively in a single cave and possess unique morphological adaptations as a consequence of their restriction to subterranean environments (Golovatch and Kime 2009;Iniesta and Ferreira 2015;Shear et al. 2016;Enghoff and Reboleira 2020). Members of the class have been used as important models for formulating and testing biogeographic hypotheses (Simonsen 1992;Enghoff 1993Enghoff , 1995Enghoff , 2015Wesener and VandenSpiegel 2009;Suzuki et al. 2012;Means and Marek 2017;Reip and Wesener 2018). Moreover, millipedes are recognized for their important role as detritivores, playing a crucial role in the decomposition of plant matter and nutrient cycling (Schubart 1942;Crawford 1992;Golovatch and Kime 2009;Suzuki et al. 2013;David 2015;Nsengimana et al. 2018;Potapov et al. 2019). More recently, studies have focused on the relationship between microclimatic factors and the distribution and life cycle of millipedes (David and Vannier 1997;Ott and Van Aarde 2014;Bogyó et al. 2015;Lazorík and Kula 2015), especially in regards to global change (David 2009;Gilgado et al. 2022).
The order Spirostreptida Brandt, 1833 comprises nearly 1300 described species commonly found in tropical, subtropical, and temperate forests (Shear 2011;Enghoff et al. 2015). The distribution of the order corresponds to the classic Gondwana pattern with members occurring throughout the Afrotropical, Oriental, and Neotropical regions (Hoffman 1980;Jeekel 1985;Enghoff et al. 2015). Historically, a number of biogeographic hypotheses have been proposed regarding dispersion events of spirostreptidan groups towards the southeastern United States and northern Africa (see Jeekel 1985). The order is divided into the suborders Cambalidea and Spirostreptidea, each containing five families (Shear 2011;Enghoff et al. 2015). The families Pseudonannolenidae (Cambalidea) and Spirostreptidae (Spirostreptidea) are the dominant groups in the Neotropical region occurring from the Mexican to the South American transition zones Iniesta et al. 2020a, b). Although Brazil stands out in terms of high levels of species richness and endemism for millipedes, the distribution patterns and areas of endemism of both families are still poorly known (Schubart 1945a(Schubart , 1950Hoffman 1980;Krabbe 1982). In the country, members of Spirostreptida have received little attention concerning their conservation 1 3 status although some taxa, such as Pseudonannolenidae from subterranean habitats, have been categorized according to IUCN Red List criteria (Karam-Gemael et al. 2018.
In the present work, we investigate the biogeographic patterns of Spirostreptida in Brazil to address the following questions: (1) what are the areas of endemism for the order?
(2) What are the species distribution patterns for the group and how are these patterns spatially arranged in regard to the country's terrestrial biomes? (3) What is the conservation status of these species according to IUCN criteria? These questions were addressed using a multi-approach analysis, including distribution patterns and conservation statuses of native species from a biogeographical perspective.

Data collection
The study group comprised 133 species from 21 genera in two families (Pseudonannolenidae and Spirostreptidae) (Fig. 1). Non-native species or species with doubtful records were excluded from the analyses. Distribution data were collected through a series of literature queries and supplemented with data from the following Brazilian institutions ( Data were checked to eliminate repetitive samples, and any samples with imprecise or incongruent locality information were not included in the analyses. All species described by the end of 2020 were included in the dataset. Location information and geographical coordinates were verified in Google Earth Pro (Alphabet, California, USA, v.7.1). All map scales are given in kilometers. For this work, we employed the six Brazilian terrestrial biomes according to Coutinho (2006), viz., Amazon Forest (Amf); Atlantic Forest (Atf); Caatinga (Ca); Cerrado (Ce); Pampa (Pap); and Pantanal (Pat).

Areas of endemism
To address the first question (what are the areas of endemism for the order?), we carried out two analyses of endemicity based on distribution data. For the Parsimony Analysis of Endemicity (PAE) (Rosen 1988;Morrone 1994;Nihei 2006) the entire region was divided into 1600 quadrats plotted on a map with a grid size of 1°. Each species was coded according to its absence (0) or presence (1) in each quadrat. A hypothetical ancestral area (allzero) was added to the matrix to root the tree (Rosen 1988). Quadrats without recorded species were omitted from the analysis. The dataset was analyzed under the parsimony criterion using TNT 1.5 (Goloboff et al. 2008;Goloboff and Catalano 2016). New Technology Search was performed under equal weights with the following parameters: random seed 0, sectorial search in default mode, and tree-fusing with five rounds until the most parsimonious tree was reached 50 times independently (Goloboff 1999).
The same dataset was analyzed in an Endemicity Analysis (EA) under NDM using the program VNDM ver. 3.0 (Szumik and Goloboff 2004) with grid sizes of 1°, 2°, and 3°. Approaches with different grid sizes were conducted to check the effect of this parameter on patterns of endemism based on distribution data (Aagesen et al. 2009;Escalante et al. 2010Escalante et al. , 2013. The searches were conducted to retain areas with scores equal to or above 2.0 and containing at least two endemic species. The searches were repeated 100 times, keeping overlapping areas only if 90% of the species in each area were unique. Consensus areas  (Brölemann, 1902), Paraná state; e Spirostreptidae sp., Paraná state; f Spirostreptidae sp., Rio de Janeiro state. Photographs by L.F.M. Iniesta 1 3 of endemism using a cut-off of 100% similarity and strict consensus were computed (Szumik et al. 2002;Szumik and Goloboff 2004). As PAE and EA have different premises and parameters, we selected the final areas of endemism based on the outputs of both analyses, with the limits of the areas of endemism (totally or partially overlapping) drawn based on the clades and quadrats recovered in PAE and EA.
In addition to PAE and EA, analyses of bioregions were conducted using Infomap Bioregions (Vilhena and Antonelli 2015;Edler et al. 2017). This analysis is a networkbased method which provides bipartite networks of species using a clustering algorithm based on geographical data, so that grid cells merge into different clusters, and thus, into distinct bioregions. We performed the analysis with the following parameters: cell size (minimum = 1°, maximum cell size = 4°), cell capacity (minimum = 10, maximum = 100), and grid resolution of 1°. We used the following cluster parameters: number of trials = 1, cluster cost = 1.00, and with false weight on abundance.

Patterns of richness and distribution
To answer the second question (what are the species distribution patterns and how are these patterns spatially arranged in regard to the country's terrestrial biome?), species richness was calculated separately for all terrestrial Brazilian biomes. To examine the patterns of Spirostreptida distribution, we performed an analysis to check the extent of the range of each genus with a grid size of 1° (for more details see Lennon et al. 2004;Colli-Silva and Pirani 2019), with three different distribution categories defined according to the number of grids containing at least one record: narrow (1-2 grids), medium (3-9 grids), and widespread (10 or more grids). The analysis was conducted at the generic level due to the paucity of species-level taxonomic treatments for most genera within the order. To avoid any distribution bias (e.g. undersampled areas, disjunct areas), the distribution categories were suggested only by the number of grids with occurrence points. In this sense, the distance between grids (if more than one is recovered) was not considered here to determine patterns of distribution. The turnover of species proposed by Whittaker (1960) was calculated in order to measure the rate at which species assemblages change by grid. All these analyses were performed with DIVA-GIS ver. 7.5.0. (Hijmans et al. 2001).
In addition, we used the "Betapart" package in R (Baselga and Orme 2012) to assess the beta-diversity of all taxa. This analysis computed the total dissimilarity using Sørensen index, turnover, and nestedness components according to the distribution of the order across all biomes (Baselga 2010(Baselga , 2012.

Conservation status
The third question (what is the conservation status of these species according to IUCN criteria?) was addressed via the R package "ConR v. 1.2.2" (Dauby et al. 2017) computing the extent of occurrence (EOO) and area of occupancy (AOO) of all species. This package generated automated conservation assessments (AA) according to IUCN Red List categories: Least Concern, Near Threatened, Vulnerable, Endangered, and Critically Endangered. These analyses are focused on providing tools for conservation policies based on a biogeographical approach. In addition, we assessed all occurrences and distribution categories for each genus throughout the Brazilian biomes. Importantly, this procedure provides only a preliminary assessment of the conservation status of the group, considering their distribution and number of records, and not fragmentation of habitats, decline or fluctuations of populations, for instance. Data were plotted using the R package "ggplot2" (Wickham 2016).

Results
We compiled 545 occurrence points of Brazilian Spirostreptida. A total of 61 species were recovered with a single record, while Pseudonannolene robsoni Iniesta & Ferreira, 2014 was recovered with the highest number of occurrences (26 points), followed by Pseudonannolene tricolor Brölemann, 1902 and Plusioporus setiger (Brölemann, 1902) with 24 occurrence points each. The PAE, EA, and Infomap analyses, carried out to detect endemic areas for Spirostreptida in Brazil, reflected the same pattern of biogeographical units: Pantanal, southern Amazon rainforest, eastern Cerrado, and central Atlantic Forest (mainly the mountain ranges of Serra da Mantiqueira and Serra do Mar) (Figs. 2, 3). A summary containing all endemic areas recovered convergently in all analyses is presented in Fig. 8.
The PAE resulted in 196 trees, with 197 steps, consistency of 68, and retention of 62. Four areas were recovered with at least one quadrat containing exclusive species: area 1 (blue)-two grids in the Amazon River and Marajó Bay; area 2 (yellow)-six grids in the southern Amazon Rainforest, southern Pantanal, and western Cerrado (Araguaia River); area 3 (red)-one grid in the southern Caatinga; area 4 (green)-six grids in the central Atlantic Forest (Mantiqueira and Mar mountain ranges) (Fig. 2). The values obtained in this analysis with species recovered by grid are presented in Supplementary material 1.
The EA recovered seven consensus areas with endemicity indices (EI) ranging from 2.0 to 5.19 (grid size of 1°), six consensus areas with EI ranging from 2.0 to 6.98 (grid size of 2°), and eight consensus areas with EI ranging from 2.04 to 9.94 (grid size of 3°). The endemic areas recovered were the Atlantic Forest and southern Cerrado (in green), and the Pantanal, eastern Cerrado, and southern Amazon Rainforest (in yellow) (Fig. 3a). The number of consensus areas by each grid size and number of endemic species are presented in Supplementary material 2.
The Infomap analysis resulted in eight bioregions (Fig. 3b), with the number of species records by cell varying in relation to each bioregion. Bioregions 1 and 3-8 correspond to the Atlantic Forest and (partially) the Cerrado, with the most common species being Pseudonannolene robsoni, Pseudonannolene microzoporus Mauriès, 1987, and Plusioporus setiger. Bioregion 2 corresponds to the Pantanal, eastern Cerrado, and southern Amazon Rainforest, with Trichogonostreptus (Oreastreptus) mattogrossensis (Silvestri, 1902) being the most common species. Sixty-one species of Pseudonannolenidae and Spirostreptidae were represented by a single record across all bioregions. A summary table for all bioregions identified in this analysis is presented in Supplementary material 3.
The regions with the highest species richness of Pseudonannolenidae and Spirostreptidae are located primarily in the Atlantic Forest, with a maximum of 11 species by grid across both families (Fig. 4a). A similar trend is observed for the turnover of species, in which for both families the highest values of turnover (values above of 3.05 and 7.33) were recovered in the Atlantic Forest (Fig. 4b). The Atlantic Forest contains at least six exclusive genera, followed by the Amazon Forest with four genera, and the Pantanal with a single genus. The Cerrado, Caatinga, and Pampa did not contain any exclusive genera. The richest biomes are the Atlantic Forest with 75 species (46 Pseudonannolenidae and 29 Spirostreptidae) and the Cerrado with 55 species (35 Pseudonannolenidae and 20 Spirostreptidae), while the least rich biomes are the Caatinga with six 1 3 species (three Pseudonannolenidae and three Spirostreptidae) and the Pampa with three species (one Pseudonannolenidae and two Spirostreptidae) (Fig. 5a). Eleven genera occur only in a single biome. Of the 21 genera, six are classified as widespread (occurring in at least 10 grids), seven as having a medium distribution (between 3 and 9 grids),  and eight as having a narrow distribution (less than or equal to 2 grids) (Fig. 5b). The values of Sørensen dissimilarity, nestedness of species assemblages, and spatial turnover are 0.912, 0.128, and 0.784, respectively, with the highest values found between the Amazon Forest-Caatinga, Amazon Forest-Pampa, Caatinga-Pantanal, Caatinga-Pampa, and Cerrado-Pampa (Fig. 6).
Regarding the conservation assessments according to the IUCN Red List categories, most Brazilian Spirosteptida are classified as either endangered or critically endangered (Fig. 7a). The highest concentration of endangered species based on the proportion and number of species by grid, and number of records by grid, is in the Atlantic Forest, Distribution categories of the genera specifically the same region that was recovered as endemic (i.e. Cerrado and Atlantic Forest; Fig. 7b).

Endemic areas and distribution patterns of Spirostreptida
The spirostreptidan fauna of Brazil is largely unknown primarily due to the scarcity of trained taxonomists in the country, with new records often turning out to represent undescribed taxa (Hoffman et al. 1996(Hoffman et al. , 2002. The areas of endemism detected in this study were recovered at different levels of geographical congruence, and as such, we delimited them on the basis of their contiguous grids and number of shared species. Six endemic areas are herein proposed for the Brazilian Spirostreptida: Northern Serra Geral (NSG), Southeastern mountain ranges (Smm), Cerrado and Atlantic Forest zone (CAF), Eastern Cerrado and Serra Espinhaço (ECS), Pantanal (Pt), and Southern Amazon and Cerrado zone (SAC) (Table 1, Figs. 8, 9). In relation to the proportion and number of threatened species, and number of records by grid. LC least concern, NT near threatened, VU vulnerable, EN endangered, and CR critically endangered The zone may be viewed as an ecotone between the southeastern Cerrado and the western Atlantic Forest. The region is characterized by a variety of climates, with annual mean temperature ranging from 18 to 26 °C and annual rainfall ranging from 1000 to 1900 mm. Elevation ranges from 250 to 3000 m a.s.l. The area is composed of a shifting mosaic of habitats, with both biomes occurring in well-drained areas primarily along the Grande, Paranaíba, and Tietê rivers, located between strips of gallery, deciduous, and semideciduous seasonal forests Most of these endemic areas are contained within the Atlantic Forest biome (Fig. 10a-e), which has been previously shown to be an area of endemism for many invertebrate and vertebrate species (Silva et al. 2004;Sigrist and Carvalho 2009;Bispo and Lecci 2011;Mendes and Sebastiani 2012;Trevine et al. 2014;DaSilva et al. 2017;Tonetti and Cavarzere 2017). Overall, two regions in the biome are highlighted in relation to their geological aspects and endemic species assemblages, the Atlantic coastal region and its continental islands (e.g. Alcatrazes, Anchieta, Ilhabela, Moela, and Queimada Grande) and the mountain ranges in the eastern part of the region.
Historically, the Atlantic coastal region has received considerable attention due to its unique millipede fauna (Brölemann 1909;Schubart 1945aSchubart , 1949Schubart , 1950, and more recently for the common occurrence of non-native species (Bouzan et al. 2018;Iniesta et al. 2020aIniesta et al. , b, 2021Iniesta et al. , 2022a. Most of the continental islands in the biome were connected to the continent by a land bridge during the recession of seawater at the time of the Last Glacial Maximum (~ 85,000 to 15,000 years ago), with many vertebrate and invertebrate populations The zone may be viewed as an ecotone between the Cerrado and the southern Amazon. The annual mean temperature ranges from 24 to > 26 °C, and the annual rainfall ranges from 1900 to 3100 mm. The elevation ranges from 200 to 800 m a.s.l. The heterogenous vegetation is represented by a mosaic of habitats including rainforests, transitional, and semi deciduous forests See Fonseca (1985), Lorenzi (1992), Coutinho (2006), Colombo and Joly (2010), and Alvares et al. (2013) for details of the biomes and ecoregions remaining isolated from each other and the continent (Martin et al. 1986;Fleming et al. 1998). Schubart (1949, p. 239) suggested this scenario for some millipede species when comparing their wide variation in body size and patterns of coloration between continental and insular populations. The presence of continental islands, large rivers (e.g. Tietê and Paraná), and mountain ranges (e.g. Serra do Mar, Serra de Paranapiacaba, and Serra da Mantiqueira) is a significant factor affecting the distributional patterns and endemism of Spirostreptida in the Atlantic Forest, specifically following the climatic changes during the Pleistocene (see Cabanne et al. 2016). Studies have shown that some mountain ranges in the Atlantic Forest (e.g. the Smm area) correspond to routes of species diversification, leading to higher rates of endemism as compared to surrounding areas (Grazziotin et al. 2006;Bragagnolo et al. 2015;DaSilva et al. 2015;Oliveira et al. 2015;Lago-Barcia et al. 2020).
Historical processes, such as climatic events during the last ~ 2 M.Y., along with corresponding alterations of the phytophysiognomy of tropical forests due to fluctuations in wet and dry climates have been demonstrated to have influenced the spatial distributions of birds (Cabanne et al. 2016), beetles (Silva and Vaz-de-Mello 2020), and harvestmen  (Haffer 1969;Ledru et al. 2005;Silveira et al. 2019;Rodríguez-Zorro et al. 2022). Similarly, Walker et al. (2009) revealed that the broad distribution of some millipede species of Narceus Rafinesque, 1820 (Juliformia, Spirobolida) in the Appalachian Mountains is a result of a complex evolutionary history with multiple refugia in southeastern North America and population expansions in the north during the Pleistocene. Decker (2016) suggested that paradoxosomatid species (Polydesmida, Paradoxosomatidae) in Australia have high genetic variability as a consequence of multiple Pleistocene refugia in the southeastern Australian mainland. In fact, historical climatic events in the Atlantic Forest may partially explain the pattern of distribution of Spirostreptida species, especially considering the relevance and influence of environmental conditions on the distribution of low vagility millipede taxa (David 2009;Gilgado et al. 2022).
The highest species richness is primarily found in the states of São Paulo and Rio de Janeiro (Figs. 4a, b, 5a). This is not unexpected, as most of our knowledge regarding Brazil's millipede fauna is a result of extensive studies by the European naturalists O. Schubart and H. W. Brölemann at the beginning of the nineteenth century, which resulted in the description of a large number of species and distributional data from these states (Brölemann 1904(Brölemann , 1909Hoffman 1980;Bouzan et al. 2018;Iniesta et al. 2021). Since these early studies, however, few others have focused on the taxonomy of Brazilian Spirostreptida (see Krabbe 1982). The endemic pattern of Spirostreptida in the Cerrado (areas CAF, ECS, and SAC) may be best explained by the physiognomically heterogeneous vegetation of the biome (Lorenzi 1992) (Fig. 10f). Overall, these areas present a composition typical of a semihumid tropical climate (Alvares et al. 2013) with a shifting mosaic of relatively welldrained gallery forest habitats along streams (e.g. São Francisco and Paraná rivers). Due to the low vagility of millipedes, it is plausible that many species have very limited distributions in specific areas across the biome (e.g. eastern portion of the Cerrado in CAF, or western portion of the Cerrado in SAC) resulting in a distinctive faunal composition. The expansions and contractions of woodlands under cool and moist conditions in the Cerrado during the Last Glacial Maximum may have been influential in the diversification of species (Oliveira et al. 2020), although this has not been well documented for millipedes.
Most studies in the Pantanal have revealed a high species richness of invertebrates influenced by a local mosaic of forests and savannas in extensive floodplains (Marques et al. 2016;Martins et al. 2021). The composition of Spirostreptida millipedes in the Pantanal shows high dissimilarity and high turnover of species as compared to other Brazilian terrestrial biomes (Fig. 6). The spatial distribution of the order is largely concentrated in the northern part of the region and in the Cáceres and Poconé subregions (Fig. 4a, b) where extensive faunal surveys have been conducted during the last two decades (Golovatch et al. 2005;Battirola 2009Battirola , 2017Santos-Silva 2019;Iniesta et al. 2022b). Most Spirostreptidae found in the Pantanal migrate along the surface following the flood line during high water periods (Battirola 2017). The same activity has been observed in some Central Amazonian millipede species (Adis 1981(Adis , 1992 in areas close to the Araguaia and Amazon rivers (Schubart 1947a). Most populations found in these areas are composed of large-bodied individuals with relatively high mobility (Santos-Silva 2019) as compared to species from other endemic areas (e.g. Smm), although it still presents a high endemicity of species.
Most species recovered in the SAC endemic area are concentrated in the southern Amazon and in the northern Cerrado, a transitional zone partly composed of open ombrophilous forest with sparse vegetation and characterized by long periods of drought. The low species richness of the Amazon forest, as compared to the transitional area to the Cerrado, is likely a result of the limited number of identified species and taxonomic studies focused on the millipede fauna of the region (Brölemann 1904;Hoffman et al. 1996Hoffman et al. , 2002.
In general, little is known regarding the ecological drivers affecting the distributional patterns of Spirostreptida at small and medium scales. Nonetheless, average temperature and humidity have been recognized as important bioclimatic variables in determining the occurrence of millipede species around the world (David 2015;Gilgado et al. 2022;Iniesta et al. 2022a). In Brazil, some reports have been made for spirostreptidans with a low ecological demand regarding their occurrence and abundance in old growth forests or in poly/monocultures of Araucaria angustifolia (Bertol.) Kuntze (Araucariaceae), Eucalyptus L'Hér. (Myrtacea), Coffea L. (Rubiacea), Zea mays L., Citrus reticulata Blanco (Rutaceae), Nicotiana L. (Solanaceae), and beans (Fabacea) depending only on relative humidity and availability of organic resources (Schubart 1942(Schubart , 1945b(Schubart , 1949(Schubart , 1955. In addition, in the country some synanthropic species found in close proximity to residential areas are known for their occasional population outbreaks (Schubart 1944(Schubart , 1945b(Schubart , 1947b(Schubart , 1958.

Conservation status of Spirostreptida
Most Brazilian Spirostreptida occur in areas that have not been designated as conservation units or do not have conservation policies in place. To date, five spirostreptidan species are classified as either endangered or critically endangered according to the most recent list of threatened species provided by Brazilian authorities (ordinance MMA no. 148, June, 2022) (Fig. 9). All of these taxa occur in subterranean environments and represent troglomorphic species with specialized morphological traits, such as Pseudonannolene ambuatinga Iniesta & Ferreira, 2013, P. lundi Iniesta & Ferreira, 2015, and P. spelaea Iniesta & Ferreira, 2013 (Pseudonannolenidae) (see Sket 2008).
In Brazil, since the publication of the Federal Decree no. 6640/2008 and the Normative Instruction MMA 02/2017, subterranean environments are protected by law according to geological and biological attributes (e.g. presence of troglomorphic species), so that only caves of maximum relevance are fully protected (Brasil 2008(Brasil , 2017. As a result, most of the Brazilian cave fauna has been extensively studied in the last two decades, resulting in the description of species, habitats, and biological attributes. With the exception of P. lundi, which occurs in a single, relatively well-preserved, rarely visited cave, the remaining species occur in areas which are at least partially impacted by anthropogenic activities. The caves where P. ambuatinga occurs are the focus of an intense investment in mining operations of carbonate rock to produce cement, lime, and soil correctives, directly affecting the subterranean environment and cave communities (Iniesta and Ferreira 2013a). Other impacts observed in the surrounding areas include deforestation, agropastoral practices, and construction in close proximity to drainages and/or cave entrances (Cavalcanti et al. 2012). Pseudonannolene spelaea is regarded as an Amazon Rainforest relict, occurring only in caves of the Serra do Carajás located in the region of the Floresta Nacional de Carajás (FLONA-Carajás), an Amazonian landscape composed of large plateaus of ferruginous outcrops (Iniesta and Ferreira 2013b). This region has experienced intensive exploration of its vast iron ore deposits, which has influenced the local economy and increased environmental pressures (Palheta et al. 2017). Other endemic invertebrates, such as beetles, crickets, flies, centipedes, spiders, and terrestrial isopods, are also found in these ferruginous outcrops, reinforcing the urgency of additional studies to understand the impacts of mining operations on species distribution and conservation (Iniesta et al. 2012;Brescovit et al. 2018;Campos-Filho et al. 2020;Chagas-Jr and Bichuette 2018;Bouzan et al. 2019;Asenjo et al. 2019;Junta et al. 2020;Teodoro et al. 2021). Although no apparent morphological features indicate the restriction of P. imbirensis to caves, all known specimens have been collected from caves or cave entrances (Fontanetti 1996;Gallo and Bichuette 2017;Bichuette et al. 2019). Populations of P. imbirensis from the Angélica, São Bernardo II, and São Vicente II caves were considered troglobitic, while those from the Terra Ronca II cave were considered troglophilic by Gallo and Bichuette (2017). Among the five spirostreptidan species cited in the national list of threatened fauna, P. tocaiensis is a typical troglophilic species known only from its type locality, Toca cave (Itirapina, São Paulo state), which has a fairly well-preserved surrounding area (Fontanetti 1996;Freitas et al. 2004). Importantly, the most recent presidential decree (decree n°10.935, January 12, 2022) is currently allowing the destruction of even the most conservation relevant caves in Brazil, representing a serious threat to all Brazilian subterranean biodiversity (see Ferreira et al. 2022).
Although the southeastern region has the highest spirostreptidan species richness in Brazil (Figs. 4a, 5a), it also has the highest concentration of threatened species (Fig. 7b). The distribution of species in the Atlantic Forest has been severely altered by accelerated deforestation and subsequent fragmentation effects, medium to large-scale farming activities (e.g. coffee, orange, and sugar cane), and by indiscriminate urban expansion and industrialization since the beginning of the 20th Century (see Fonseca 1985;Lembi et al. 2020). Only approximately 10% of the original area of the Atlantic Forest (once covering 16% of Brazil) remains, represented by small forest patches and regenerating areas (Colombo and Joly 2010;Joly et al. 2014;Rezende et al. 2018). A similar situation can be seen in the Pantanal and the Cerrado, in which a number of endemic areas were recovered during this work. Studies have shown that there is a high potential for loss of biodiversity in these regions through deforestation due to extensive cattle farming, agricultural activities, wildfire, hunting, and unregulated tourism (Ratter et al.

3
1997; Durigan et al. 2007;Alho et al. 2019;Berlinck et al. 2022). It should be noted that our analysis of Brazil's endangered millipedes, based on the number of records and distributional patterns, should be treated as preliminary. Other types of data, such as fragmentation of habitats (e.g. extension of deforestation), reduction of AOO/EOO, and population fluctuation, should be considered for a more comprehensive approach. This is the first study focused on exploring the distribution of Spirostreptida in the Neotropics based on analytical methods, with six areas of endemism proposed for the 133 Brazilian Spirostreptida species. Most endemic areas fall within the Atlantic Forest biome, an extremely biodiverse Neotropical area harboring an abundance of invertebrate and vertebrate taxa. Although the southeastern region of Brazil (including the Cerrado and Atlantic Forest endemic areas) has the highest recorded Spirostreptida species richness, it also has the highest concentration of threatened species. Our study demonstrates that species assemblages of Spirostreptida are largely restricted to small and medium scale areas (e.g. islands, mountain ranges, and caves) susceptible to human impacts. Moreover, these areas are severely threatened by the lack of comprehensive conservation strategies, especially those containing caves, which are continually threatened by mining and development operations. Increased taxonomic efforts coupled with indepth exploration of biogeographic patterns may reveal detailed local patterns in some regions, such as in the Amazon and the Caatinga, which are largely unknown regarding Spirostreptida.