Desoria calderonis sp. nov ., a new species of alpine cryophilic springtail (Collembola: Isotomidae) from the Apennines (Italy), with phylogenetic and ecological considerations

1,4 Department of Biosciences, Università degli Studi di Milano, Via Celoria 26, 20133 Milano, Italy . 2,3,7,10 Department of Life Sciences, University of Siena, Via A. Moro 2, 53100 Siena, Italy. 5 Research and Museum Collections Offi ce, Climate and Ecology Unit, MUSE-Science Museum, Corso del Lavoro e della Scienza 3, 38122 Trento, Italy . 6 Department of Life, Health & Environmental Science, University of L’Aquila, Piazzale Salvatore Tommasi 1, Coppito, L’Aquila, Italy . 8,9 Department of Environmental Science and Policy, Università degli Studi di Milano, Via Celoria 10, 20133 Milano, Italy .


Introduction
Cryophilic springtails (Hexapoda: Collembola) are cold-and moisture-requiring organisms, whose typical habitat is near or above ice or snow (Deharveng et al. 2008;Fjellberg 2010;Buda et al. 2020;Valle et al. 2020;Jureková et al. 2021). Desoria saltans Nicolet, 1841 (Entomobryomorpha: Isotomidae) is among the best known cryophilic springtails and is commonly known as the "glacier fl ea". This species was already cited by the Italian geologist Stoppani in his early essay "Il Bel Paese" (Stoppani 1876) for its showy, swarming and large assemblages on Alpine glaciers. Nowadays there are indications that "glacier fl eas" include multiple taxa of cryophilic Isotomidae related to the genus Desoria : Desoria Nicolet in Desor, 1841, Gnathisotoma Cassagnau, 1957and Myopia Christiansen & Bellinger, 1980(Najt 1981Fjellberg 2010). These organisms appear to be differentiated among isolated glacial areas (Deharveng 1975;Lauga-Reyrel & Lauga 1995). By comparing different European glacier forelands, Hågvar et al. (2020) observed how the species belonging to this group, and to a few other pioneer genera, are important not only as components of supraglacial communities (see also Gobbi et al. 2021), but also as early colonisers of recently-deglaciated terrains. High diversity of cryophilic Isotomidae was observed in southern Alaska and in the Canadian Rocky Mountains in North America (Fjellberg 2010). Therefore, we can assume high diversity of the genera related to Desoria at similar sites in Europe.
The genus Desoria comprises 101 described species (Bellinger et al. 1996(Bellinger et al. -2021, mostly distributed in the Holarctic, especially at high latitudes (Potapov 2001). Desoria differs from the closely related genus Isotoma Bourlet, 1839 by the number of apical setae on tibiotarsi (11, with the exception of the pjasini -group), and the absence of ventroapical spine-like setae on the manubrium (Potapov 2001;Fjellberg 2007). From the ecological point of view, Desoria also differs from Isotoma by being more frequent at cold and wet sites and thus including many hygrophile and cold-adapted species (Potapov 2001). Species belonging to the genus Desoria overwinter in the adult stage and are often active on snow or ice, on which they feed (Hao et al. 2020) and migrate (Hågvar 2000;Zhang et al. 2017). Among these cold-adapted species, Desoria and taxonomically related genera like Agrenia Börner, 1906, Gnathisotoma , Myopia and Kaylathalia Stevens & D'Haese, 2016(Najt 1981Fjellberg 2010;Stevens & D'Haese 2017) include cryophilic species (Potapov 2001;Fjellberg 2007Fjellberg , 2010. Cryophilic species, in particular, are of great interest because they are particularly threatened by the current global warming and are good candidates as indicators for conservation projects aimed at investigating refugial glaciated areas (Gobbi et al. 2021).
In this work, we present a description of a new species of Desoria found in one of the southernmost European relict glacial areas, the Calderone glacier (Central Apennines, Italy; Grunewald & Scheithauer 2010). This species is thus a good indicator of a "cold-spot" of the local glacial biodiversity (Cauvy-Fraunié & Dangles 2019) in the Mediterranean region. In order to provide a robust taxonomic classifi cation for this new species, we applied both morphological and genetic approaches, to obtain a phylogenetic framework. In addition, we provide remarks on its habitat and microhabitat preferences.

Study area and data collection
Specimens were collected by the fl otation method (Marshall et al. 1994) on the Calderone glacier. The Calderone glacier is located on the Gran Sasso Massif, in the Central Apennines (Italy, Abruzzo; 42°28ʹ16.4″ N, 13°34ʹ01.4″ E). Presently, it is classifi ed as glacieret (total surface < 0.04 km 2 ; Smiraglia & Diolaiuti 2015) and it is almost totally covered by stony debris. We searched for Desoria calderonis sp. nov. in all glacial environments, on the supraglacial debris and on the Little Ice Age (LIA) moraines. In order to better verify its distribution, both fl otation and pitfall trap methods were used for 15 sampling points on the supraglacial debris as well as 6 sampling points on LIA moraines as controls. In order to better characterize the ecology of Desoria calderonis sp. nov., we recorded micrometeorological (temperature and humidity) and soil data of its habitat and of the surrounding habitat where it was not found, in particular: • temperature was recorded on supraglacial debris by 15 and on LIA moraines by 2 dataloggers (iButton 1922) for the period 9 July 2020 -27 July 2021; using this data, we calculated the mean annual temperature, the mean temperature during the snow-free period, the minimum and the maximum temperatures and the duration of the snow-cover; • relative humidity was recorded in both environments by a datalogger (Tinytag Plus) for the same period; with this data we calculated the mean annual value and the mean value during the snow-free period; • soil samples were collected for 15 points on the supraglacial debris and 6 points on LIA moraines. In soil samples we measured the value of carbonate calcium content, organic matter content and pH. All dataloggers were positioned 10 cm under the surface.

Specimen conservation and preparation
Specimens (1 holotype and 15 paratypes), preserved in 90% ethanol at -20°C, were initially cleared by a short immersion in 10% KOH solution and then mounted on slides using lactic acid or Marc André as a preservative solution. Additional specimens (fi ve) were prepared for scanning electron microscopy: they were completely dehydrated in absolute ethanol, before critical point drying in a Balzers Union (FL-9496) apparatus and the subsequent sputter coating with gold in an Edwards Sputter Coater S150B.
Morphological observations were performed with a Leica Laborlux S light microscope and a Quanta400 (FEI) scanning electron microscope.

Molecular analysis
Whole genomic DNA was extracted from 10 specimens, individually, using the Wizard ® SV Genomic DNA Purifi cation System (Promega, Madison, WI, USA). The mitochondrial marker analyzed -cytochrome c oxidase subunit 1, 5P fragment ( cox1 ) -was amplifi ed with a universal primer pair (Folmer et al. 1994). PCRs were prepared in a 25 μL reaction volume containing: 2.5 μL of whole genomic DNA, 1.25 μL of both forward and reverse primers (10 μM), 2.5 μL of MgCl 2 (2.5 mM), 2.5 μL of deoxynucleotides (dNTPs, 10 mM), 5 μL of Green GoTaq Flexi Buffer (Promega, Madison, WI, USA), 0.125 μL of GoTaq ® G2 Flexi DNA Polymerase (Promega, Madison, WI, USA), 5 μL and 9.875 μL of ddH 2 O. Amplifi cations were run on a GeneAmp ® PCR System 2700 (Applied Biosystems, Foster City, CA, USA) thermal cycler with the following conditions for each of the 35 cycles: a denaturation step at 95°C for 1 min, an annealing step at 50°C for 1 min and an elongation step at 60°C for 90 s. An additional initial denaturation step was set at 95°C for 5 min as well as a fi nal extension step at 72°C for 7 min. PCR products were purifi ed with the kit Wizard ® SV Gel and PCR Clean-up System (Promega, Madison, WI, USA) and sequenced on both strands using a DNA Analyzer ABI 3730 at Biofab (Rome, Italy). Sequences were then manually corrected and assembled in Sequencher ver. 4.2.2 (Gene Codes, Ann Arbor, MI, USA).

Phylogenetic analysis
Given the current uncertainties on the monophyly of the genus Desoria (Stevens et al. 2006;Stevens & D'Haese 2017) and of its closest relatives, a preliminary analysis was conducted to identify Desoria as well as phylogenetically related sequences in order to bypass the assumption of a monophyletic Desoria in the process of taxa selection. All records belonging to the family Isotomidae were downloaded from the BOLD database (Ratnasingham & Hebert 2007) with their metadata (last download 30 June 2021). The 8483 records were fi ltered to retain only those including information for the COI-5P cox1 fragment and having an assigned bin. The longest sequence for each bin was extracted using the R package 'bold' (ver. 1.2.0, written by Scott Chamberlain) and used as representative for the bin in the following analysis. For all records, metadata associated with a specifi c bin were revised to identify the taxonomic attribution(s), if available, of each bin, whereas bins with no taxonomic information below the family level in any sequence were discarded. Records were associated with 728 bins in BOLD, 348 of which included sequences with at least some associated taxonomic information. After the addition of the new species, D. calderonis sp. nov.(mean uncorrected divergence within the species is 0.4%, S.D. 0.2), the total dataset was composed of 349 sequences by 438 aligned positions (1 st and 2 nd positions only). The dataset was aligned with MAFFT (Katoh 2002) in order to calculate both uncorrected p-distances and phylogenetic relationships. These latter were obtained, using 1 st and 2 nd codon position data set, by IQ-TREE (ver. 1.6.12, default settings with 1000 fast bootstrap replicates and the model was optimized using ModelFinder; Nguyen et al. 2015). A reasonably supported node (bootstrap 92) was identifi ed, including all Desoria sequences -with the exception of Desoria trispinata (3 bins, 18 sequences) and Desoria tshernovi (1 bin, 1 sequence, unrelated to 6 D. tshernovi sequences within the node) -as well as sequences from other genera. This subdataset, inclusive of related species and three outgroups ( Cryptopygus terranovus, Parisotoma notabilis and Folsomia quadrioculata , grouping outside the Desoria cluster), accounted for 89 sequences by 438 aligned positions and was reanalyzed as above to investigate the phylogenetic position of the new species in the context of Desoria and related species. Once a fi nal phylogenetic tree had been obtained, records that appeared phylogenetically related to D. calderonis sp. nov. were further revised in the BOLD database metadata or in the original literature to assess the existence of ecological and/or phylogeographic similarities.

Etymology
The epithet of the new species refl ects the name of the site from which holotype and paratypes derive, Calderone glacier (Italy, Abruzzo, Gran Sasso massif).  Table 1). Colour violet-black on abdomen and antennae, lighter on furca and legs, which are brownish (Fig. 1); juveniles are much paler, bluish. Cuticle granulation fi ne and regularly distributed; all dorsal tergites clearly separated from each other. Abd. III and IV of approximately same width.

Measurements
See Table 1.
Within the group, considering the number of common characteristics, D. calderonis sp. nov. appears most similar to D. duodecemoculata -present in Italy, Austria, Spain and France (Potapov 2001) and D. nivalis , present in the Alps (France, Switzerland, Austria) and possibly in eastern Europe (Potapov 2001) (Table 2). Nevertheless, some features allow us to discriminate the new species from these. First, D. duodecemoculata and D. nivalis do not have short, thick and cylindrical, but only hairlike s -setae on Ant I. In addition, the new species differs from D. duodecemoculata by the chaetotaxy of VT and the number of dorsal setae on the dens; it differs from D. nivalis by having Ant II-III violet-black (white in D. nivalis ), a longer PAO and by the number of dorsal setae on the dens.
Both D. duodecemoculata and D. nivalis belong to the nivalis -complex, a group of European mountain species included in the violacea -group. Desoria nivalis , in particular, is known to live near snow fi elds and other cold sites in high mountains (Handschin 1924;Franz & Serrl-Butschek 1954), while the ecology and the taxonomy of the other members of the complex need to be revised (Potapov 2001). Because of these similarities in morphology and ecology, we could ascribe our species to this nivaliscomplex, even if we reported marked differences , in particular the presence in D. calderonis sp. nov. of the sensory fi eld on Ant. I.

Phylogenetic context
The phylogenetic tree (Fig. 6), with a log-likelihood of 3077.45, is characterized by good support at recent nodes but low support at deeper nodes. It appears subdivided into three major clusters, two dominated by Isotoma and Isotomurus , respectively, and one by Desoria , with records representing 14 different Desoria species as well as others incompletely identifi ed as Desoria sp. This latter cluster also included scattered sequences from the following genera:

Discussion
Our analysis supports the idea that Desoria is a polyphyletic group, as already observed by Stevens & D'Haese (2017). In particular, Desoria clustered with Vertagopus , Pseudisotoma , Isotoma , Skadisotoma , Proisotoma , Isotomurus , Agrenia , Metisotoma , Axelsonia , Chionobora and Kaylathalia. Most of the Desoria records came from areas characterized by a markedly cold climate and here cluster with other strictly cold-adapted and cryophilic organisms. In particular, Agrenia Börner, 1906 is a Holarctic genus living in damp and cold habitats such as banks of cold-water streams and the shores of lakes in tundra and mountain zones (Fjellberg 1988), on or near snow (Fjellberg 1976(Fjellberg , 1986(Fjellberg , 1994 and polar deserts (Chernov et al. 1977). Hågvar (2010) reported A. bidenticulata , Desoria olivacea and D. infuscata as the most pioneer species of the proglacial succession along Midtdalsbreen glacier (Norway). Skadisotoma Greenslade & Fjellberg , 2015 is a mountain endemic Australian genus linked to snow patches (Greenslade & Fjellberg 2015) and taxonomically related to Desoria . Chionobora Greenslade & Potapov , 2015 is an endemic hygrophilous Tasmanian genus, living around the lakes of the Central Plateau, the largest area of high ground in Tasmania. Kaylathalia is an Antarctic isotomid and was regarded as belonging to Desoria until Stevens & D'Haese (2017). Isotoma and Isotomurus are present in this cluster with a few sparse sequences identifi ed at the genus level, but most of the sequences belonging to these genera were grouped in two alternative well defi ned groups outside the Desoria cluster. Thus, we hypothesize that the sparse Isotoma and Isotomurus sequences clustering with Desoria could be phylogenetically misplaced or not properly identifi ed specimens. Therefore, we suppose that the genera more strictly related to Desoria are Vertagopus, Skadisotoma , Agrenia, Metisotoma, Axelsonia, Chionobora and Kaylathalia.
The position of the new species in a cluster composed entirely of Desoria sequences (with the exception of a single sequence of Vertagopus ) supports its morphological identifi cation as belonging to the genus. Desoria calderonis sp. nov. is dissimilar (17.8% genetic variability) to all other sequences present in the BOLD database. Morphologically, D. calderonis sp. nov. is more similar to D. nivalis and D. duodecemoculata of the mountain and cold-adapted nivalis -complex ( violacea -group). However, this affi nity was not testable phylogenetically with the available dataset, since no sequences from this complex are present.
In terms of subgeneric relationships, D. calderonis sp. nov. appears to be related to species belonging to both the fennica-( D. tigrina : bin BOLD:ACS3918; D. germanica and D. intermedia : bin BOLD:AAI9461) and violacea -group ( D. blufusata : bin BOLD:ACT9239 and D. violacea : BOLD:AEA8472), leaving the question of its morphological assignment to the violacea -group untestable. In general, our phylogenetic results do not support the morphological groups reported by Potapov (2001) as natural assemblages.
From a methodological standpoint, the Desoria phylogeny presented here is diffi cult to interpret due to the possibility that some key groups are polyphyletic, the lack of sequence data for crucial elements and the uncertainty in the attribution of some sequences. In fact, building robust phylogenetic trees would require the combination of data from a large number of genes, integrating nuclear and mitochondrial Fig. 6. Phylogenetic tree of Desoria calderonis sp. nov . and related species, on the basis of the cox1 gene. Names include the BOLD bin number, as well as the taxonomic attribution and number of sequences included in the bin. Genera were abbreviated where unambiguous within the bin. When records of the same bin had multiple taxonomic attributions, the one at the lowest level was retained if all were compatible. Alternatively, all were listed separately. Bootstrap support is indicated if > 80. ♠: olivaceagroup; ♣: fennica -group; ♥: violacea -group of Desoria .
information, and the cox1 barcoding fragment is suboptimal in terms of resolution, especially at deeper nodes. Nevertheless, a great advantage of cox1 is the availability of a large number of sequences in the BOLD database, including sequences from rare or diffi cult to sample species. Thus, the use of this gene allows for preliminary considerations on the phylogenetic context of the new species even in the absence of a consolidated taxonomy of the group, which would require more complete phylogenetic analyses.
Desoria calderonis sp. nov. was described for the Calderone glacier, a relict, isolated glacier of the Apennines, a peripheral mountain chain without other existing glaciers. This is in line with the hypothesis of an undescribed fragmented glacial springtail biodiversity in refugial areas. It is noteworth that, these glacial areas are highly threatened by climate change (Grunewald & Scheithauer 2010). We do not know the fate of the cold adapted and cryophilic collembolan fauna in these areas, but extinction seems to be a likely scenario (Greenslade & Fjellberg 2015). This underlines the importance of studying these unique environments and preserving their biodiversity in order to know it before its defi nitive disappearance.