Distribution and conservation status of threatened endemic amphibians within the Aspromonte mountain region, a hotspot of Mediterranean biodiversity

Amphibian biodiversity loss in recent years has exceeded that of all other groups of vertebrates. In this context, biodiversity hotspots represent priority targets for conservation in amphibian populations. However, little information is available on the distribution and conservation status of amphibian species within most biodiversity hotspots. Here, we characterized the distribution and conservation status of four threatened endemic amphibians ( Bombina pachypus , Salamandra salamandra gigliolii , Salamandrina terdigitata , and Rana italica ) in the Aspromonte Mountain region, a biodiversity hotspot in southern Italy where the conservation status of amphibians is almost unexplored. We conducted an intensive field survey of 507 potential breeding sites spanning over 2,326 km 2 . We found that all four species were widespread in the study area. We observed 337 species occurrences: 63 for S. s. gigliolii , 29 for S. terdigitata , 84 for B. pachypus , and 161 for R. italica . Species distribution analysis revealed that S. s. gigliolii and R. italica populations had an extended and homogenous distribution. Conversely, S. terdigitata showed a dispersed pattern, with long distances among breeding sites, and B. pachypus an aggregated pattern, associated with the availability of suitable artificial habitats. On the other hand, we reported a decrease in B. pachypus occurrence in its natural habitats, which reflects a negative trend of its populations. Overall, our results provide an encouraging framework for the conservation of amphibian populations in this area, but highlight the low coverage of threatened amphibian populations in protected areas, highlighting the need for a reassessment of conservation policies and spatial conservation planning for the Aspromonte region.


Introduction
Amphibians are declining worldwide owing to habitat degradation, pollution, climate change, and emerging diseases (Blaustein and Kiesecker 2002;Collins and Storfer 2003;Stuart et al. 2004;Collins 2010;Blaustein et al. 2011;Catenazzi 2015;Scheele et al. 2019). Nearly half of all amphibian species are experiencing regional or local decline, and approximately one-third are threatened by extinction. In Europe, 23% of amphibian species are included in the threatened categories of the International Union for Conservation of Nature European Red List (Temple and Cox 2009;Sindaco 2016). Although the extinction rate in amphibians exceeds that of any other group of vertebrates (Stuart et al. 2004;McCallum 2007), conservation policies have made little progress in halting or reversing these trends (Grant et al. 2016).
The Italian Peninsula is one of the most important biodiversity hotspots in the Palearctic region and hosts a highly diverse amphibian fauna, comprising about half of all amphibian species in Europe (Sindaco et al. 2006;Temple and Cox 2009;Rondinini et al. 2013), as well as many endemic evolutionary lineages. A generalized decline has been detected in the Italian populations of many amphibians, including some endemic taxa such as the Apennine yellow-bellied toad (Bombina pachypus), the Italian stream frog (Rana italica), the fire salamander (Salamandra salamandra gigliolii) and the spectacled salamanders (Salamandrina terdigitata and S. perspicillata) (Barbieri et al. 2004;Lanza et al. 2007;Andreone et al. 2009). Despite this widespread decline, monitoring schemes aimed at establishing demographic trends and conservation statuses are scattered and mostly limited to a few amphibian populations (e.g. Italica S. 2011). In particular, the conservation status of amphibians in the southernmost part of the Italian Peninsula is severely under-investigated. It is worth noting that this part of the peninsula holds the most of the intraspecific diversity for several species (Canestrelli et al. 2010;Bisconti et al. 2018a;Chiocchio et al. 2019), including most Italian amphibians investigated to date (Canestrelli et al. 2006a(Canestrelli et al. , 2006b(Canestrelli et al. , 2008(Canestrelli et al. , 2012(Canestrelli et al. , 2014Chiocchio et al. 2021). Consequently, a thorough understanding of the conservation status of amphibian populations in southern Italy is a priority for establishing effective conservation plans at both local and regional scales.
In this study, we aimed at characterizing the distribution and conservation status of the threatened amphibians inhabiting the Aspromonte Massif, a mountain region in the southernmost part of the Italian Peninsula (Table 1). The Aspromonte Massif is an extended and heterogeneous mountain region surrounded by the sea and characterized by a high diversity of habitats spanning dry grasslands and shrublands to oak, pine, and beech forests, including a wide range of humid areas that host many amphibian species. Recent studies identified the Aspromonte region as a glacial refuge and hotspot of genetic diversity for many temperate species (Todisco et al. 2010;Canestrelli et al. 2010), including amphibians (Canestrelli et al. 2006a;Iannella et al. 2018). Despite the biogeographic and conservation value of this area, surveys on the distribution and conservation status of amphibians are scarce and discontinuous over space and time (Tripepi et al. 1999;Tripepi and Sperone 2007;Temi and Agristudio 2018). The only exhaustive study carried out in this area to clarify the conservation status of an endangered amphibian, the Apennine Yellow-bellied toad, highlighted the strong discordance between the historical and current distribution of the investigated species (Zampiglia et al. 2019). This evidence, coupled with the scarcity of data for other threatened amphibians, suggests the need for a thorough investigation of the current distribution of amphibians in this area. The objective of this study is to provide an updated and detailed reference framework concerning the distribution, habitat preferences, and conservation status of threatened amphibians in southern Italy. To achieve this, we apply a fine-scale field survey to map the current presence of threatened amphibians in the Aspromonte region. We will focus on four amphibian species endemic to the Italian Peninsula, which have shown local or regional decline because of habitat reduction/alteration or pathogen outbreaks: Bombina pachypus, Rana italica, Salamandrina terdigitata, and Salamandra salamandra gigliolii (see Table 1).

Study area
The Aspromonte is a mid-to-high mountain region (maximum altitude 1,957 m asl), located at the southern boundary of the Italian Peninsula ( Fig. 1) and characterized by a high diversity of habitats and singular climatic conditions. The seasonal distribution of rainfall has typical Mediterranean features, with less rainfall in the summer months than in the winter months (Colacino et al. 1997). The mountainous belts are characterized by abundant precipitation during autumn, winter, and spring, and streams and brooks are mainly perennial. Most humid habitats in the hilly and coastal strips show dependence on seasonal precipitations. In mountainous areas, the average winter temperatures are quite low, with minimum temperatures occurring in January and February and frequently dropping below 0 °C. Conversely, in the coastal area, the average summer temperatures exceed 40 °C in July and August. In general, the western and eastern sides are characterized by almost opposite microclimatic features, that is, wetter and cooler on the western side and warmer and drier on the eastern side. Brullo et al. (2001) identified two distinct macro-bioclimates in Aspromonte (divided into different thermotypes): a Mediterranean rain-seasonal oceanic bioclimate between 0 and 1,100 m asl and a temperate oceanic bioclimate between 1,100 and 1,957 m asl.
Climatic and orographic diversification have generated ecological gradients and highly heterogeneous vegetation typologies (Spampinato et al. 2008). In endemism, relict species, and ancient woods, the Aspromonte represents a hotspot of plant and animal biodiversity at the species, genetic, and community levels (Spampinato et al. 2008;Zampiglia et al. 2019;Piovesan et al. 2020). Biodiversity conservation within the Aspromonte region is undertaken by the Aspromonte National Park (ANP), together with 57 Natura 2000 areas, 55 of which are Special Areas of Conservation (SACs) and two are Special Protection Areas (SPAs).

Field survey
The identification of potential presence sites for all the investigated species was obtained by consulting the available literature (Ruffo and Stoch 2005;Lanza et al. 2007; and reference therein) and analyzing cartographic data. Cartographic data were retrieved from the Italian Military Geographic Institute (Military Geographic Institute) and satellite images to identify potential aquatic habitats. In line with other studies (e.g. Sindaco et al. 2006;Brandmayr et al. 2017), we considered the original standard grid defined by the European Community as a cartographic reference (reference system ETRS89-LAEA Europe -Lambert Azimuthal Equal Area projection), a universal transverse mercator (UTM) grid with cells of 10 × 10 km mesh (N = 45, 2.804 km 2 ) and WGS84 projection.
Field surveys were conducted from spring 2016 to autumn 2021, during the period of activity of the investigated species, focusing on the species' reproductive season, according to the phenology of each species, i.e. from March to October. We applied standard methodologies, such as the visual encounter survey (VES) and calling survey (CS), as reported by Heyer et al. (1990) and Sindaco (2016). We also followed all the relevant guidelines to avoid disease spreading during monitoring activities (Phillott et al. 2010). We examined adults, subadults, larvae, and eggs. Each site was visited three times. The field survey involved both natural habitats and artificial environments (Suppl. material 2: Table S2). Investigation of B. pachypus covered a large part of the Aspromonte Massif (19 UTM grids, for a total amount of 1,822 km 2 ) whereas investigations of the other three species were almost exclusively in the ANP and adjacent territories (18 UTM grids, for a total amount of 913 km 2 ). Each site was georeferenced using a GPS device and subsequently transferred to a geographic information system (GIS) platform (ArcMap 10.7.1, ESRI Inc., CA, USA). As B. pachypus and S. terdigitata are mentioned in Annexes II and IV of the EU Habitat Directive, we added abundance estimates of populations using visual counts for these two species. All procedures were approved by the Italian Ministry of Ecological Transition and the Italian National Institute for Environmental Protection and Research (ISPRA; permit number: 7727, 15-04-2016).

Data analysis
For each presence site, the following environmental information was retrieved: altitude from sea level using the DTM (Digital Terrain Model), with a resolution of 10 m; habitat type according to the CORINE Biotopes (European Commission 1991) and EUNIS codes (APAT 2004); and Horton order of river courses (Horton 1945). The environmental typology surrounding the sites was defined by buffer analysis (see e.g. Plăiaşu et al. 2012;Canessa et al. 2013) with a buffer radius of 100 m (3.14 ha). Each buffer area was superimposed on the land cover map (LC, De Fioravante et al. 2022) to generate an independent set of polygons representing land cover with a resolution of 10 m. The percentage of land cover within each area was calculated. Similarly, altimetry, exposure, and slope were obtained using the DTM layer, with a resolution of 10 m. Data visualization and analysis were performed using ArcMap 10.7.1.
To produce information for conservation management purposes, we estimated the following parameters: the diffusion index (DI, Ragni 2002), the distribution of species within the study area using the minimum convex polygon method (MCP, Mohr 1947), and the pattern of distribution using the average nearest neighbor analysis (ANN, Clark and Evans 1954). DI measures the ratio between the number of cells with the species occurrence out of the total number of cells investigated and spans from 0 (i.e., the species is not present in any cell) to 1 (i.e., the species is present in all the cells investigated). MCP generates a polygon whose angles are the outermost locations of the presence sites. Although MCP tends to overestimate home ranges by including areas not frequently visited by the species (Borger et al. 2006), we employed this method because it allows the identification of the boundaries of the species range in the study area (Burt 1943); the minimum boundary geometry tool was used to obtain MCP. The ANN is a method used to describe the spatial structure of occurrences (Pommerening et al. 2020). It determines the clustering of occurrences in the study area by measuring the distance between each location and its neighboring centroid location and then calculating the average among all the nearest neighbor distances. If the average distance is less than the average for a hypothetical random distribution, the distribution is considered clustered; if the average distance is greater than the hypothetical random distribution, the distribution is considered dispersed (Clark and Evans 1954).
Finally, we estimated species habitat selection in the study area using the Manly standardized habitat selectivity index (α) (Manly et al. 1972(Manly et al. , 2002. This index represents the proportional use of a resource (or habitat) divided by the proportional availability of each habitat. The values of α range from 0 to 1: α = 1/k, where k is the number of available resources, and indicates that the resource is used randomly and in proportion to the abundance in the environment; α > 1/k indicates positive selection of the resource; α < 1/k indicates resource avoidance. Habitats with the highest α were considered key habitats for the species (Desbiez et al. 2009).

Results
From 2016 to 2021, we visited 507 sites in 27 UTM cells (Fig. 1), corresponding to 46.66% of the study area (mean of 24.14 sites per cell). The investigation involved most of the ANP (245 sites, 48.2%) and 20 of the 55 SACs in the study area. The investigated species were observed in 337 of the 507 visited sites (66.46%), of which 220 were within the ANP; species occurrences were detected in 16 SACs (Suppl. material 2: Table S6). The distribution of species occurrences are summarized in Table 2 and Fig. 2.

Fire salamander
Salamandra salamandra gigliolii was observed at 63 sites, mostly within the ANP ( Fig. 2A, Suppl. material 1: Table S1). This species is mainly distributed along the western side of the mountain massif. The altitude and habitat distributions are summarized in Fig. 3 and Suppl. material 2: Tables S2-S5. The altitudinal distribution spanned from 718 to 1,774 m asl, with most observations occurring between 1,300 and 1,400 m asl (39%). The species was found almost exclusively in forest environments, such as beech forests (LC code 21115, 67.49%) and oro-  Mediterranean mountain pine forests (LC code 21122, 19.54%). Permanent streams and brooks represented the most frequented aquatic habitats (81%), with a few interesting observations in peat bogs (5%). The most frequent river courses belonged to Horton's orders 1 (38%) and 2 (35%). This species was found in seven SACs (Suppl. material 2: Table S6). ANN analysis (Table 2) showed a random distribution pattern (NNR: 0.986; zscore: -0.197; p-value: 0.843). The Manly Index (Suppl. material 2: Table S7) suggests a species positive selection for altitude range between 1,200 and 1,500 m asl, slope classes of 0-15° and 15-30°, fresh sides affected by the Tyrrhenian humid currents (i.e., N, NW, and W exposure), beech forest and mountain pine forest habitats (LC codes 21115 and 21122, respectively).

Spectacled salamander
Salamandrina terdigitata was observed at 29 sites, most of which were within the ANP (Fig. 2B, Suppl. material 1: Table S1). The altitude and habitat distributions are summarized in Fig. 4 and Suppl. material 2: Tables S2-S5. The altitudinal distribution was from 122 to 1,573 m asl, with most observations below 700 m asl. The species mainly lives in forest environments, such as oak forests and broad-leaved woods (LC code 21111; 25.52%), beech woods (LC code 21115; 16.78%), and olive groves (LC code 21132; 13.96%). The species was mainly found in permanent streams and brooks (61%) and temporary streams and brooks (32%). The most frequent river course was Horton's order 4 (29%) and 3 (26%). This species was found at four SAC sites (Suppl. material 2: Table S6).
ANN analysis (Table 2) showed a dispersed distribution pattern (NNR: 1.284; z-score: 2.828; p = 0.004). The Manly Index (Suppl. material 2: Table S8) suggests a positive selection for altitudes of 100-200 m asl, 700-800 m, and 1,500 m above sea level, as well as a positive selection for slopes below 15° on cool and humid sides (i.e., W, NW, and E exposure). Interestingly, despite the extensive use of oak forests (LC code 21111) and beech forests (LC code 21115), the Manley index showed species positive selection for less available habitats, that is, orchards (LC code 21131) and chestnut (LC code 21114).
The number of observed individuals at each site during each visit spanned from 1 to approximately 1000: 1 to 7 adult individuals were observed at 10 sites; up to 50 individuals (larvae and/or adults) were observed at 15 sites; more than 50 individuals were observed at four sites, of which, at one we counted approximately 100 individuals (mainly larvae) and at another approximately 1000 larvae (see Suppl. material 1: Table S1).
The number of observed individuals spanned from 1 to 120 adult individuals: up to 50 individuals were counted at 71 sites and more than 50 were observed at 12 sites, of which 5 showed more than 100 individuals. It is worth noting that B. pachypus was not found at 14 sites of historical presence, which have been checked at least three times each year.

Italian stream frog
Rana italica was observed at 161 sites, most of which were within the ANP (Fig. 2D, Suppl. material 1: Table S1). The altitude and habitat distributions are summarized in Fig. 6. The altitudinal distribution spanned from 118 to 1,822 m asl, with a peak of observations around 1,300 m asl (12%). The species has been frequently observed in beech forests (code LC 21115, 32.23%) and oak and evergreen broad-leaved tree forests (code LC 21111, 22.91%). The species has been mostly observed in permanent and temporary streams and river courses of 1-4 Horton's code. These species were observed in 14 SACs (Suppl. material 2: Table S6).

Discussion
Results from this survey update the information on distribution and conservation status of four threatened amphibian species inhabiting the Italian Peninsula. We focused on the southernmost portion of the Italian Peninsula, as it is an acknowledged hotspot of biodiversity within the Mediterranean region and one of the most unexplored regions of Europe in terms of amphibian ecology and conservation. We collected data on a large number of new and previously unknown breeding sites for the four investigated species, which depicts an encouraging frame on the conservation status of the populations in this area.
Overall, we found that these four species were widespread in the study area. The inspection of the distribution maps (Fig. 2) outlined R. italica as the most abundant species and S. terdigitata as the least abundant species. S. terdigitata also showed the  most fragmented distribution, as supported by ANN analysis with a highly dispersed distribution pattern. These distribution patterns are in line with those shown by these species on the Italian Peninsula in previous studies (Lanza et al. 2007). By contrast, S. s. gigliolii and B. pachypus were more widespread in the Aspromonte Massif than in the northern regions of the Italian Peninsula (Lanza et al. 2007), despite their nonuniform distributions in the study area. Indeed, S. s. gigliolii was mainly distributed on the western side of the mountain massif, with only a few observations on the eastern side. This distribution reflects the needs of the species for cooler microclimate conditions, which are more common on the western and northwestern mountain sides in this region. Interestingly, the distribution map of S. s. gigliolii occurrences ( Fig. 2A) highlighted strong contiguity of populations, as supported by ANN analysis, which excluded either clustered or dispersed patterns of distribution. Conversely, B. pachypus was more abundant in the southwestern part of the Aspromonte region, despite many observations having been localized to the eastern side of the ANP. Furthermore, inspection of the B. pachypus distribution (Fig. 2C) showed a strongly clustered distribution, as confirmed by ANN analysis. This pattern of distribution could reflect the patchy availability of the artificial environments colonized by B. pachypus, that is, irrigation tanks, drinking troughs, and artificial ponds, which currently constitute the speciespreferred habitats (Fig. 5).
The habitat use of the four species in the Aspromonte Massif (Figs 3-6) agrees with available information on species ecology, although there were some relevant differences. S. s. gigliolii showed a marked preference for a higher altitudinal range in Aspromonte (1,200-1,300 m asl) compared to the rest of the peninsula (where the species shows a preference for altitudes around 800-1,000 m) (Lanza et al. 2007). This difference is even more marked in R. italica, which is commonly more frequent below 800 m asl (Lanza et al. 2007) but showed a peak frequency distribution around 1,300 m asl in Aspromonte. In contrast, B. pachypus showed a wider distribution at lower altitudes than the rest of the peninsula. Moreover, we found a population at 1,613 m asl, which, along with observations reported in the Pollino Massif (around 1,600 m asl: Barbieri et al. 2004;Talarico et al. 2004;Romano et al. 2012), is the highest record for B. pachypus. We also reported a wider range of habitats for S. terdigitata, suggesting a wider ecological niche of populations in Aspromonte than in the northernmost populations. Finally, our results emphasize the importance of preserving forests, which comprised the most frequented habitats of S. terdigitata, S. s. gigliolii, and R. italica.
The comparison of our data with those of previously published studies (Tripepi and Sperone 2007) highlights a general increase in species presence, especially for B. pachypus and S. terdigitata. This increase can be attributed to the wider geographic coverage of this survey, compared to previous surveys, rather than a real increase in species distribution. It is worth noting that the unavailability of exact locations from previous surveys does not allow for direct comparison and inferences on population trends. However, we visited 26 known historical sites for B. pachypus (see Zampiglia et al. 2019) and in 14 of them (54%), the presence of the species was not confirmed. A negative trend was also reported by local people (shepherds, farmers, woodcutters, and walkers), who confirmed that B. pachypus is less widespread now compared to the early 1990s when it was ubiquitous and abundant in hilly and mountain streams and ponds, as well as in artificial aquatic environments used for agriculture and local pastoralism. Our data suggest a change in species habitat suitability, as we only observed B. pachypus in five natural waterbodies (6% of the observations), with the majority of observations (53) occurring in artificial habitats: irrigated tanks (35), drinking troughs (15), and artificial ponds (3). Interestingly, the importance of artificial habitats, especially drinking troughs, in preserving amphibian populations has been recently raised for other Italian species in different regions (see Buono et al. 2019). This pattern necessitates the possible integration of amphibian conservation with rural activities in a sustainable development scenario. Finally, although our data outline a good conservation status of B. pachypus populations in Aspromonte compared to that in central and northern Italy, we report a negative trend in population that has to be accounted for by conservation stakeholders. This evidence, coupled with the expected reduction in habitat availability in the next 50 years due to climate change (Zampiglia et al. 2019) and the scarcity of populations enclosed in protected areas (Fig. 2C), makes clear the need for rapid and effective conservation actions for B. pachypus in the Aspromonte region.

Conclusions
This study provides further evidence of the vital role of the Aspromonte Massif in biodiversity conservation. In this area, all investigated species were found to be more widespread and relatively more abundant than in the central and northern parts of the Italian Peninsula (see, e.g. Vanni and Nistri 2006). The better conservation status of the populations in this area can be attributed to two main reasons. First, during the Anthropocene, human activity has been less intensive in this region than in the rest of Italy. Because of this, Aspromonte contains several patches of ancient forests which act as "biodiversity tanks" in the face of anthropogenic habitat degradation. Second, Aspromonte harbors a hotspot of genetic diversity for all four investigated species (Canestrelli et al. 2006a(Canestrelli et al. , 2006b(Canestrelli et al. , 2008Mattoccia et al. 2011;Bisconti et al. 2018b;Iannella et al. 2018;Zampiglia et al. 2019). Because genetic diversity provides populations with the potential to adapt to environmental changes, hotspots of genetic diversity represent invaluable resources for species to adapt to global changes (Hampe and Petit 2005;Zachos and Habel 2011). Therefore, preserving populations within the Aspromonte region preserves populations with the highest potential to survive extinction threats that can be used for effective, genetically informed, translocation, and repopulation programs. Nevertheless, even though most of the populations were within the ANP boundaries, our data showed only a few occurrences within the Natura 2000 network. In particular, almost all observed populations of B. pachypus, one of the most endangered vertebrates in Italy, were outside protected areas. These results highlight the need for rapid implementation of new distribution data for these species in future conservation policies concerning the Aspromonte region. In this respect, it is worth mentioning that Aspromonte underwent several destructive fires over the last 10 years, many of which were considered human-induced. In particular, during the summer of 2021, after our field survey was complete, a large fire swept through 17.733 ha of the Aspromonte Massif, destroying 3,200 ha of forest, of which approximately 25 ha were ancient forests, with severe impacts on biodiversity. This fire event involved 34 of 337 sites in this study (10%), of which 18 were B. pachypus breeding sites (20% of B. pachypus breeding sites in this study). This event stresses the need to implement management methods that reduce the spread of fires and, thus, their impact on biodiversity, avoiding to set the hotspot on fire.

Supplementary material 2
Tables S1-S10 Authors: Giuseppe Martino, Andrea Chiocchio, Antonino Siclari, Daniele Canestrelli Data type: Docx file Explanation note: Table S2. Habitat distribution for each studied species (RF -Relative Frequency). The correspondence to the habitats of Annex 1 of the Habitats Directive is specified (HD). Table S3. Land cover distribution in study area and for each studied species. Table S4. Altitude distribution for each studied species (RF -Relative Frequency). Table S5. Horton order distribution for each studied species (RF -Relative Frequency).