Material and methods
Study sitenext section
The new species is known only from phreatic waters found in two caves located at Cerro de Santiago, a Cambrian limestone outcrop within the municipality of Cazalla de la Sierra (Seville, southern Spain). The area is next to Riera de Benalija, a tributary brook to the Guadalquivir River that flows into El Pintado reservoir. The cave lakes are hydrographically connected to the reservoir so that their level follows the annual oscillation of the reservoir’s water table. Specimens were found crawling in high numbers on the bottom of the cave lakes and were readily attracted by bait. Some specimens were observed venturing in open water, but most remained associated to the substratum. Direct observations did not suggest the species carried an interstitial life. Crustaceans found accompanying the new species at the time of sampling include Metahadzia uncispina Notenboom, 1988; Salentinella seviliensis Platvoet, 1987; Stenasellus escolai Magniez, 1977; Microcharon marinus Chappuis and Delamare, 1954; Megacyclops brachypus Kiefer, 1954; Diacyclops bicuspidatus odessanus (Shmankevich, 1875); Cypria ophtalmica (Jurine, 1820) and a not yet determined harpacticoid copepod. In addition, the bathynellid syncarid Hexabathynella sevillaensis Camacho, 2005, was recently described from the same caves (Camacho, 2005).
Specimens were collected directly with a hand-held plankton net in the cave lakes and fixed in situ in 95% ethanol. Once in the laboratory, they were dissected in lactic acid under the stereomicroscope, and appendages figured with a Leica DM2500 microscope equipped with Nomarski differential interference contrast and a drawing tube. Body measurements were derived from the sum of the maximum dorsal dimensions of body somites and exclude telson length. Type material is deposited in the invertebrate collection of Naturalis Biodiversity Center, Leiden (RMNH). Abbreviations and nomenclature used in morphological descriptions are as follows: A1 (antennule); A2 (antenna); G1-G2 (gnathopods I and II, respectively); P3-P7 (pereiopods III to VII, respectively); and U1-U3 (uropods I to III, respectively).
DNA isolation, PCR amplification and sequencing
Genomic DNA was purified from 13 specimens corresponding to nine different Haploginglymus taxa, viz. the new species described herein, the aberrant H. morenoi, and seven not yet formally described species coming from several Iberian locations (v. Table 1 & Fig. 1). In addition, sequences from 14 specimens of Niphargus from four different populations from the Iberian western edge of the Pyrenees (Basque Country), as well as eight specimens belonging to six different Pseudoniphargus species from the Iberian Peninsula, Portugal and the Canary Islands were also obtained and included in the dataset as potentially close outgroups. DNA extraction was performed using the DNeasy Tissue kit (Qiagen, West Sussex, UK) following the manufacturer’s protocol. Elutions were done in 100 μL volume and 1 μL was used in PCR reactions. Three different molecular markers were selected for the study, namely: a partial sequence of the mitochondrial Cytochrome c Oxidase subunit 1 gene (cox1; primers LCO1490 and HCO2189; Folmer et al., 1994), a partial sequence of the nuclear ribosomal 28S (LSU; primers 28S lev2; Verovnik et al., 2005 Zakšek et al., 2007), and a fragment of the nuclear Histone 3 gene (H3; primers H3aF and H3aR; Colgan et al., 1998). PCR conditions included 0.2 μM of each primer and 3.5 mM MgCl2 using a standard protocol of 35 cycles, with annealing temperature ranging from 50 to 45 °C (60s) depending on the sample. Denaturation (94 °C) and elongation (72 °C) lasted 30 and 60s, respectively. PCR products were inspected by electrophoresis in 1% agarose gel and purified using MSB Spin PCRapace (Invitek, Berlin, Germany). Sanger sequencing was performed with the same primers using the BigDye Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, CA, USA). Sequences were edited and contigs were assembled using CodonCode Aligner (CodonCode Corporation, Dedham, MA, USA), and deposited at GenBank under the accession numbers referred in Table S1.
The molecular dataset was complemented with sequences retrieved from GenBank corresponding to a selection of 134 Niphargus species/populations covering the entire geographical distribution and morphological disparity of the genus, plus four of the nine additional recognized niphargid genera (v. Horton and Lowry, 2013), viz. Carinurella Sket, 1971, Pontoniphargus Dancău, 1968, Niphargobates Sket, 1981 and Niphargellus Schellenberg, 1938. We completed the dataset with a series of more distant outgroups in the families Gammaridae (6 genera), Gammaracanthidae (1), Eulimnogammaridae (1), Acanthogammaridae (1), Pallaseidae (1), Pontogammaridae (1), Anisogammaridae (1) and Crangonyctidae (3). We rooted the tree with a phoxocephalid and an urothoid, in accord with the most recent hypothesis on amphipod supra-family-level relationships (Lowry and Myers, 2013). Data on all sequences and taxa used are shown in Table S1.
Multiple sequence alignment was performed using MAFFT 7 online version (http://mafft.cbrc.jp/alignment/server/; Katoh and Standley, 2013), using the default FFT-NS-1 algorithm for the cox1 and H3 datasets, and the Q-INS-i option for the LSU sequences since it considers the secondary structure of the RNA sequences. In order to evaluate the effect on the topology of the indels in the LSU alignment, an additional LSU matrix was generated by removing ambiguously aligned regions with Gblocks v. 0.91b (Talavera and Castresana, 2007), and specifying a minimum length of two positions for a block and selecting only positions with a gap in less than 50% of the sequences if they are within an appropriate block. Nucleotide substitution saturation was assessed with the Xia’s test (Xia et al., 2003) implemented in DAMBE v. 5.2.64 (Xia, 2013). Since evidence of substitution saturation was detected on third coding positions of the cox1 alignment (see Results), we also performed the phylogenetic analyses by excluding these characters from the dataset. Genetic divergence among Haploginglymus taxa was estimated through the calculation of the uncorrected pairwise cox1 distances (p-distance) in MEGA7 (Kumar et al., 2016). In order to test the phylogenetic congruence among the three molecular markers, single-gene trees were inferred and the resulting topologies were compared. Optimal partitioning strategy and evolutionary models for each alignment were assessed with PartitionFinder (Lanfear et al., 2012) under the Bayesian Information Criterion (BIC), whereas the phylogenetic inference was conducted using IQTREE multicore v. 1.3.12 (Nguyen et al., 2015) performing 1000 ultrafast bootstrap approximation replicates. Finally, a phylogenetic analysis of the three molecular markers combined was independently carried out under both a maximum likelihood and a Bayesian framework using IQTREE and MrBayes 3.2 (Ronquist et al., 2012), respectively. For the latter, two independent analyses consisting of four chains each were run for 5·107 generations specifying a sampling frequency every 1000 generations, and setting up a burn-in fraction of 35%. MCMC convergence and effective sample size (ESS) estimates were checked with TRACER v. 1.6 (Rambaut et al., 2013).