Elevational patterns of bird species richness on the eastern slope of Mt. Gongga, Sichuan Province, China

In biological systems, biological diversity often displays a rapid turn-over across elevations. This defining feature has made mountains classic systems for studying the spatial variation in diversity. Because patterns of elevational diversity can vary among lineages and mountain systems it remains difficult to extrapolate findings from one montane region to another, or among lineages. In this study, we assessed patterns and drivers of avian diversity along an elevational gradient on the eastern slope of Mt. Gongga, the highest peak in the Hengduan Mountain Range in central China, and a mountain where comprehensive studies of avian diversity are still lacking. We surveyed bird species in eight 400-m elevational bands from 1200 to 4400 m a.s.l. between 2012 and 2017. To test the relationship between bird species richness and environmental factors, we examined the relative importance of seven ecological variables on breeding season distribution patterns: land area (LA), mean daily temperature (MDT), seasonal temperature range (STR), the mid-domain effect (MDE), seasonal precipitation (SP), invertebrate biomass (IB) and enhanced vegetation index (EVI). Climate data were obtained from five local meteorological stations and three temperature/relative humidity smart sensors in 2016. A total of 219 bird species were recorded in the field, of which 204 were recorded during the breeding season (April–August). Species richness curves (calculated separately for total species, large-ranged species, and small-ranged species) were all hump-shaped. Large-ranged species contributed more to the total species richness pattern than small-ranged species. EVI and IB were positively correlated with total species richness and small-ranged species richness. LA and MDT were positively correlated with small-ranged species richness, while STR and SP were negatively correlated with small-ranged species richness. MDE was positively correlated with large-ranged species richness. When we considered the combination of candidate factors using multiple regression models and model-averaging, total species richness and large-ranged species richness were correlated with STR (negative) and MDE (positive), while small-ranged species richness was correlated with STR (negative) and IB (positive). Although no single key factor or suite of factors could explain patterns of diversity, we found that MDE, IB and STR play important but varying roles in shaping the elevational richness patterns of different bird species categories. Model-averaging indicates that small-ranged species appear to be mostly influenced by IB, as opposed to large-ranged species, which exhibit patterns more consistent with the MDE model. Our data also indicate that the species richness varied between seasons, offering a promising direction for future work.


Introduction
The knowledge of the subterranean fauna from Portugal has significantly increased over the last decade, with the description of a high number of obligate subterranean species (tripling their number) and the establishment of new biogeographic patterns (Reboleira 2012). A high number of these species are stygobiont (i.e., confined to groundwater), mostly from wells in the north of the country, where evapotranspiration is higher (Reboleira et al. 2011. They include 62 species of crustaceans, mostly asellids, syncarids and amphipods, and one species of annelid .
In this work we describe the first stygobiont species of Coleoptera from Portugal, a diving beetle of the subtribe Siettitiina (Dytiscidae, Hydroporinae, Hydroporini; type genus: Siettitia Abeille de Perrin, 1904). Siettitiina includes the only known European genera of Dytiscidae which have stygobiont members: Siettitia, with two species in France, Iberoporus Castro & Delgado, 2001, with one species in south Spain, Etruscodytes Mazza et al., 2013, with one Italian species, andGraptodytes Seidlitz, 1887, with the Moroccan G. eremitus Ribera & Faille, 2010 among several epigean members Faille 2010, Nilsson andHájek 2018a). The subtribe also includes some North American stygobiont species, with an uncertain phylogenetic position (Miller et al. 2013, Kanda et al. 2016, Miller and Bergsten 2016, Nilsson and Hájek 2018b. The new species is known from a single female found in a well-studied cave in central Portugal. Despite multiple visits to the same cave no additional specimens have been found, so we describe here the species on the basis of its morphological singularity and of the molecular data that places it unambiguously among the west Mediterranean species of Siettitiina.

Taxon sampling, DNA extraction and sequencing
For the phylogenetic placement of the new species we used the datasets of Ribera and Faille (2010) and Abellán et al. (2013), with the inclusion of additional sequences (mostly nuclear genes) and taxa (Table 1). Most notably is the inclusion of Siettitia avenionensis Guignot, 1925, the second oldest described stygobiont water beetle worldwide. Partial sequences of the genes COI and 18S were obtained from a larva preserved in 70% ethanol, collected in 1989 (Table 1). Other attempts to extract and sequence different larvae from the same locality collected in 1984 and 1992 (Ph. Richoux leg.) proved unsuccessful. Extractions of single specimens were non-destructive, using a standard phenol-chloroform method or the DNeasy Tissue Kit (Qiagen GmbH, Hilden, Germany). Vouchers and DNA samples are kept in the collections of the Museo Nacional de Ciencias Naturales, Madrid (MNCN), the Institute of Evolutionary Biology, Barcelona (IBE) and the Natural History Museum of Denmark (NHMD). Table 1. Material used in the molecular phylogeny of the Siettitia group of genera, with locality, collector, and EMBL accession numbers. Newly obtained sequences are in bold typeface. Nomenclature follows Nilsson and Hájek (2018a Examples of most species of Palaearctic Siettitiina were included, including all stygobiont or interstitial species with the exception of Graptodytes aurasius Jeannel, 1907 (Algeria), Siettitia balsetensis Abeille de Perrin, 1904 (France) and Etruscodytes nethuns Mazza et al., 2013 (Italy). Trees were rooted in the split between Graptodytes+Metaporus Guignot, 1945 and the rest of Siettitiina, based on previous phylogenetic results ( Ribera et al. 2008, Abellán et al. 2013. Fragments of five genes in five sequencing reactions were sequenced, three mitochondrial (1) 5' end of cytochrome c oxidase subunit 1 (COI-5, "barcode" fragment of Hebert et al. 2003); (2) 3' end of cytochrome c oxidase subunit 1 (COI-3); (3) 5' end of 16S RNA plus the Leucine tRNA plus 5' end of NADH dehydrogenase subunit I (16S); and two nuclear fragments (4) an internal fragment of the small ribosomal unit, 18S RNA (18S) and (5) an internal fragment of Histone 3 (H3). Details on primers used are provided in Table 2. Sequences were assembled and edited with Geneious v6.0.6 (Kearse et al. 2012); new sequences (111) have been submitted to the EMBL database with accession numbers LS999692-LS999802 (Table 1).

Phylogenetic analyses
Edited sequences were aligned using the online version of MAFFT 7 with the G-INS-I algorithm (Katoh and Toh 2008).
BEAST 1.8 (Drummond and Rambaut 2007) was used for Bayesian phylogenetic analyses, using a molecular-clock approach for estimating divergence times. We applied a partition by genes with uncorrelated lognormal relaxed clocks to estimate substitution rates and a Yule speciation process as the tree prior, using GTR+I+G and HKY+I+G evolutionary models. We calibrated the tree using rates estimated in Andújar et al. (2012)   H3aF (5') ATGGCTCGTACCAAGCAGACRCG Colgan et al. (1998) H3aR (3') ATATCCTTRGGCATRATRGTGAC the burn-in fraction with Tracer v1.6 (Drummond and Rambaut 2007). We also used a fast Maximum Likelihood (ML) heuristic algorithm in RAxML-HPC2 (Stamatakis 2006) in the CIPRES Science Gateway (Miller et al. 2010), using the same partition scheme as in BEAST with a GTR+G evolutionary model independently estimated for each partition and assessing node support with 100 pseudoreplicates with a rapid bootstrapping algorithm (Stamatakis et al. 2008).

Results
The two BEAST analyses (GTR and HKY evolutionary models) resulted in identical topologies and very similar branch lengths, although convergence for GTR evolutionary models was poor for some genes (nad1, 18S), so we present here only the results of the HKY models (Fig. 1). The topology was also almost identical to that obtained with RAxML (Fig. 1). We obtained a well-supported, well-resolved phylogeny of Siettitiina (Fig. 1). In agreement with previous results we recovered three clades, Graptodytes+Metaporus, Stictonectes Brinck, 1943+ Porhydrus Guignot, 1945, and the Siettitia group of genera as here defined, including Siettitia, Rhithrodytes, and Iberoporus (plus most likely Etruscodytes, see Discussion). The new species was placed as sister to Iberoporus cermenius Castro & Delgado, 2001 with strong bootstrap support (BS = 73%), although in the Bayesian analyses the support was lower (posterior probability, pp = 0.73). Both species were in turn sister to Rhithrodytes argaensis Bilton & Fery, 1996plus R. agnus Foster, 1992 in a very well supported clade (BS = 94; pp = 0.97), and then to Siettitia (Fig. 1). All other sampled species of Rhithrodytes were placed as sister to this clade, rendering the genus paraphyletic. In order to preserve the monophyly of Rhithrodytes we thus transfer the two species to the genus Iberoporus, Iberoporus agnus (Foster, 1992) comb. n. and Iberoporus argaensis (Bilton & Fery, 1996), comb. n.
According to our calibration, the separation between the new species and Iberoporus cermenius was dated at ca. 10 Ma (95% HPD 13.4-6.9 Ma), with a similar age for the split from I. agnus + I. argaensis ]), during the Tortonian (Fig. 1).

Diagnosis.
A blind and depigmented species of Iberoporus, larger and wider than the other subterranean species of the genus, with a cordiform pronotum without lateral stria, less prominent constriction between pronotum and elytra and with a more transverse pronotum. Appendages longer and more slender, especially antennae and pro-and mesotibiae. Male unknown.
Description. Body length 2.8 mm, maximum width 1.1 mm. Habitus: Body elongate, strongly parallel-sided (including pronotum and head) (Fig. 2), flattened in lateral view (Fig. 3a); in dorsal view lateral outline with a slight discontinuity between posterior angles of pronotum and base of elytra. Body and appendages uniformly pale orange (cuticle appears translucent after DNA extraction due to digestion of soft tissue).
Head (Fig. 2): Wide, anterior margin almost perfectly semicircular, deeply encased in pronotum, with two lateral dark scars in place of eyes; surface smooth, with very sparse small shallow punctures, surface weakly micro-reticulated, stronger on margins, glabrous. Antennae with ovoid pedicel, distal antennomeres conical, more elongate.
Pronotum (Figs 2, 3): Cordiform, margins sinuated, anterior part slightly wider than head, posterior part narrower than head and base of elytra; anterior margin more or less straight (except angles), angles strongly acute; posterior margin sinuated, angles acute; sides without rim, anterior margin with transverse depression with irregular row of large punctures; posterior margin with some sparse large punctures very loosely forming a row. Pronotum without sublateral stria on each side, with only a slight depression and very irregular row of larger punctures. Surface smooth, with fine shallow punctures denser on disk, with very fine microreticulation, stronger near margins, cells not contiguous; centre of disc with small longitudinal rectangular mark. Pronotum with long lateral sensorial setae (Fig. 3b).
Elytra (Figs 2, 3): almost parallel-sided on basal 2/3, apical third regularly acuminate. Sides of elytra with weak rim, not visible from above. In lateral view margin of elytra almost straight, only very weakly ascending to humeral angle in anterior quarter; epipleuron not visible until shoulders. Surface with same structure as on pronotum, with very sparse larger punctures; larger punctures forming very loose and irregular lines on elytra; more distinct near to suture and on disk. With long sensorial setae on margins (Fig. 3b). Without traces of hind wings.
Ventral surface (Fig. 4): Uniformly pale, colour similar to dorsal surface. Prosternal process lanceolate, apex acuminate; not reaching anteromedial metaventral process. Epipleuron becoming narrower short before mid-length, without oblique carina near shoulder. Metepisternum more or less triangular in shape. Metacoxal lines obsolete; joint hind margin of metacoxal processes incised; lobes of processes rounded.
Etymology. From "Πλούτων" (Ploutōn), the ruler of the underworld in the Greek mythology. Name in apposition.
Notes on the habitat. Soprador do Carvalho is a cave with approximately 4 km of horizontal development (Fig. 7). It is the largest cave of the so-called Dueça Speleologi- cal System, located in the north-eastern part of the Sicó karst area in central Portugal (Neves et al. 2005). The subterranean stream feeds the spring of the Dueça River, a contributor to the Mondego River. The substrate of the river is mostly composed of   clasts and gravel, with large clay deposits on the margins. The specimen was found in the bottom of a clay pound connected to the margin of the subterranean stream. Other invertebrate stygobionts are found in this stream, such as a new species of the asellid genus Proasellus and of the amphipod genus Pseudoniphargus, and unidentified copepods (Reboleira 2012). In the terrestrial compartment of the cave, several cave-adapted species are known: the pseudoscorpion Occidenchthonius duecensis Zaragoza & Reboleira, 2018; the millipede Scutogona minor Enghoff & Reboleira, 2013; the woodlice Trichoniscoides sicoensis Reboleira & Taiti, 2015 (which has an amphibian behaviour and can be collected inside the stream totally submerged) and Porcellio cavernicolus Vandel, 1946;and the dipluran Podocampa cf. fragiloides Silvestri, 1932(Enghoff and Reboleira 2013, Reboleira et al. 2015, Zaragoza and Reboleira 2018. Over recent years, the cave is being explored for tourism. This may represent a major threat, as tourists constantly trample the bottom of the subterranean stream where the new species was found. Remarks. Iberoporus pluto sp. n. is most similar in its external morphology to I. cermenius. Both share a similar shape of the head, a cordiform pronotum without lateral stria, and similar general appearance (Figs 2, 5a). In the absence of males of I. pluto sp. n. (and in addition to the genetic differences), both species can be easily separated by the body shape, larger and wider in I. pluto sp. n., and with a less prominent constriction between pronotum and elytra (clearly visible in I. cermenius) and with a more transverse pronotum. The appendages of I. pluto sp. n. are also longer and more slender, especially the antennae and the pro-and mesotibiae (Figs 2, 5a). Iberoporus cermenius has also well-defined parasutural rows on the elytra formed by large punctures, which are absent in I. pluto sp. n.

Discussion
We obtained for the first time a phylogeny of Siettitiina including a species of its type genus, Siettitia. Despite the incomplete data, there is strong support for the existence of a clade including Siettitia, Iberoporus, and Rhithrodytes, what we call the Siettitia group of genera. Our results also clearly demonstrate the parayphyly of Rhithrodytes, and the need to transfer two of the species to maintain its monophyly. The relationships between Rhithrodytes and the other three European stygobiont genera of Siettitiina (Siettitia, Iberoporus, and Etruscodytes), although widely recognised, had not been clearly established. Originally, the genus Rhithrodytes was erected for a group of species of Graptodytes (the group IV of Zimmermann 1919, or the group "crux" of Guignot 1947) with a curved apex of the median lobe of the aedeagus, a lateral stria running the whole length of the pronotum (Bameul 1989) and (as recognised later), a transverse carina in the epipleura (Fery 2013). With the exception of the epipleural carina, the rest of the characters are shared with the subterranean genus Siettitia, which has been for long recognised to be closely related to some of the species included in Rhithrodytes (e.g., R. bimaculatus (Dufour, 1852); Régimbart 1905, Zimmermann 1932 (Table 3).
Subsequent to the description of Rhithrodytes two genera were described each for a single European stygobiont species: Iberoporus and Etruscodytes. Iberoporus cermenius shares the structure of the male genitalia with Rhithrodytes and Siettitia, but it is in particular very similar to that of I. agnus and I. argaensis. These two species (formerly in Rhithrodytes) have a more straight median lobe and a different shape of the apex of the parameres (Bilton andFery 1996, Fery 2016).
The body shape of I. agnus and I. argaensis has also some similarities to the species of Iberoporus, parallel-sided and elongated (Figs 5b, c; see figs 12-19 in Fery 2016). Iberoporus cermenius shares with Siettitia the structure of the metacoxal processes, something that could be related to the subterranean habitat and a poor swimming ability (Castro and Delgado 2001).
Etruscodytes, described from a male and a female, also shares with Rhithrodytes and Siettitia the general structure of the aedeagus (note that the tip of the aedeagus in the figure of Mazza et al. 2013 seems to be broken) and the long lateral striae of the pronotum (Table 3), but nevertheless was described in a separate genus due to some morphological peculiarities (Mazza et al. 2013). Thus, according to the description by Mazza et al. (2013), the species would have (1) head wide and "subsquare" (regularly rounded in Siettitia and Rhithrodytes; although more similar to that of Iberoporus); (2) presence of short and flattened setae on pronotum and elytra; (3) prosternal process contacting anteromedial process of metaventrite (also in Rhithrodytes, not in Siettitia and Iberoporus, Table 3); (4) anteromedial process of metaventrite rounded (pointed in Siettitia according to Mazza et al. 2013); (5) ventrites II and III not fused (fused in Siettitia and I. cermenius, not in I. pluto sp. n. or Rhithrodytes); (6) elytra not completely fused (fused in Siettitia, not in Iberoporus and Rhithrodytes). Some of these characters seem to be clear autapomorphies related to the subterranean life (fusion of elytra or ventrites, particularly shaped setae, lack of lateral striae on the pronotum, lack of carina on the epipleuron), and others are of uncertain interpretation. Thus, the structure of the prosternal process is sometimes difficult to appreciate, but there do not seem to be fundamental differences between the species (note that in fig. 7 in Mazza et al. 2013 the prosternal process seems to fit below the anteromedial process of the metaventrite, which is likely an artefact), being the differences consequence of the different position of the mesocoxa (contiguous or not) and ultimately the width of the body, which in turn may depend on the habitat and ecology of the species. More data, especially molecular sequences of Etruscodytes and Siettitia, and the likely discovery of other subterranean taxa would contribute to the understanding of the evolution of this western Mediterranean lineage.