Scorpionidae, LATREILLE, 1802

Prendini, Lorenzo & Loria, Stephanie F., 2020, Systematic Revision Of The Asian Forest Scorpions (Heterometrinae Simon, 1879), Revised Suprageneric Classification Of Scorpionidae Latreille, 1802, And Revalidation Of Rugodentidae Bastawade Et Al., 2005, Bulletin of the American Museum of Natural History 2020 (442), pp. 1-480 : 19-22

( Scorpionidae : Heterometrinae Simon, 1879 , stat. nov.)

several species or represented only by type mate- DNA SEQUENCING: Two loci in the nuclear rial, which could not be dissected. genome, the 18S rDNA (18S) and 28S rDNA (28S),

Morphological terminology follows previous and three loci in the mitochondrial genome, 12S papers on Scorpionidae Latreille, 1802 , by the rDNA (12S), 16S rDNA (16S), and cytochrome c first author (e.g., Prendini, 2000b, 2001a; Pren- oxidase subunit I (COI) were Sanger dideoxy dini et al., 2003; Tahir and Prendini, 2014; Pren- sequenced using an ABI Prism 3730 XL DNA dini, 2016). The terminology applied in the new Sequencer (Perkin-Elmer, Melville, NY) at the trichobothrial interpretation of R. keralaensis, AMNH Sackler Institute of Comparative Genomproposed here, follows Vachon (1974). ics. Double-stranded sequences were edited and

A morphological data matrix, comprising 186 assembled into consensus sequences using morphological characters (appendices 1, 2) Sequencher ver. 5.4.6 (Gene Codes Corporation, scored for 132 terminals (including four out- Ann Arbor, MI), and deposited in GenBank groups) and incorporating characters from pre- (https://www.ncbi.nlm.nih.gov/genbank) (appenviously published matrices by Lamoral (1979), dix 3). Sequences of the 18S, 28S, and COI loci Couzijn (1981), Prendini (2000a), and Prendini were length invariant, respectively comprising 1761 et al. (2003), was prepared using WinClada ver. nucleotide base pairs (bp), 514 bp, and 1078 bp in 1.00.08 ( Nixon, 2002), and deposited in Morpho- all terminals, except Nebo hierichonticus (Simon, bank (http://morphobank.org/permalink/ 1872), which was 515 bp for 28S. Sequences of the P3719). Characters scored in the matrix included 12S locus varied from 328–335 bp with a mode of a variety of traits capturing the morphological 333 bp and the 16S locus from 481–487 bp with a diversity within Scorpionidae , most pertaining to mode of 481 bp. DNA sequences were unavailable the adult stage: total length, 1 (0.5%); chelicerae, for 10 species, for which samples could not be 4 (2%); carapace, 30 (16%); pedipalps, 61 obtained; these were represented in the analyses by (32.5%); coxosternum, 1 (0.5%); legs, 33 (18%); morphological characters only. genital operculum, 1 (0.5%); hemispermato- DNA SEQUENCE ALIGNMENT: DNA sequence phore, 3 (1.5%); pectines, 5 (3%); tergites, 5 (3%); alignment was performed with MAFFT ver. 7.429 sternites, 4 (2%); metasoma, 29 (16%); telson, 5 by auto-aligning with the “leavegappyregion” (3%); ecology and behavior, 3 (1.5%). The break- option selected ( Katoh et al., 2002; Katoh and down of morphological characters by character Standley, 2013). The aligned sequences were consystem is as follows: surface macrosculpture, 70 catenated to produce a matrix of 4188 bp. The con- (38%); macrosetae, 32 (17%); color patterns, 19 catenated alignment contained 925 variable sites (10%); shape and size, 29 (16%); surface topog- and 769 parsimony-informative sites. Considered raphy, 13 (7%); trichobothria, 9 (5%); genitalia, as a percentage of the alignment length of each 4 (2%); ungues, 3 (2%); sutures, 3 (2%); denti- locus, the numbers of variable positions and parsition, 1 (1%). Forty-seven (25%) multistate char- mony-informative sites were highest for 12S (variacters were treated as unordered/nonadditive able, 56%; parsimony informative, 48%), followed ( Fitch, 1971) to avoid a priori character state by 16S (49%, 40%), COI (40%, 36%), 28S (6%, 2%) transformations. and 18S (2%, 0.3%). As expected for a protein-cod-

ing gene, the third codon position of the COI was the most informative, containing 77% and 82% of the variable and parsimony-informative sites in the locus, followed by the first codon position, containing 19% and 15%, respectively.

PHYLOGENETIC ANALYSIS: The morphological character matrix and the concatenated dataset of aligned nuclear and mitochondrial DNA sequences (hereafter, “the molecular dataset”) were analyzed simultaneously using maximum likelihood (ML), Bayesian inference (BI), and parsimony with equal weighting and implied weighting ( Goloboff, 1993), applying five values for the concavity constant, k, ranging from strong to mild (k = 1, 3, 10, 60, 100). The morphological and molecular datasets were also analyzed separately using parsimony with equal weighting and implied weighting with k = 1, 3, 10, 60, 100, and the molecular dataset was analyzed separately using ML and BI.

Parsimony analyses were performed using TNT ver. 1.1 ( Goloboff et al., 2003, 2008), with a script from Dimitrov et al. (2012, 2013) modified by Santibáñez-López et al. (2014a), which includes tree drifting, mixed sectorial search, and tree fusing for the tree search. Gaps were treated as missing data and uninformative characters deactivated. A jackknife analysis was performed to evaluate nodal support using another script modified from Dimitrov et al. (2012). A bootstrap analysis was performed for comparison with 1000 pseudoreplicates.

The ML analysis was conducted using RAxML- HPC ver. 8.2.12 on the CIPRES gateway (https:// www.phylo.org; Stamatakis, 2006, 2014). The dataset was partitioned by loci, a MULTIGAMMA model was implemented, and an ascertainment bias applied to the Mk model ( Lewis, 2001) for the morphological partition. A rapid bootstrap analysis with 1000 replicates ( Stamatakis, 2014) was used to search for the ML tree.

Twenty-four nucleotide substitution models were tested for each locus using JModeltest ver. 2.1.6 ( Guindon and Gascuel, 2003; Darriba et al., 2012) on the CIPRES gateway. Models were selected using the Akaike information criterion (AIC) and a gamma distribution assumed for the morphological dataset ( Lewis, 2001). Using these models, BI was conducted in MrBayes v. 3.2.7 on the CIPRES gateway ( Ronquist and Huelsenbeck, 2003) with unlinked parameters for all partitions. The analysis was terminated after 25 million generations when the standard deviation of the split frequencies was below 0.01. The preferred hypothesis for the phylogeny of Heterometrinae , with unambiguous morphological synapomorphies optimized, is presented in fig. 10. A more detailed explanation of the methods and results of the phylogenetic analyses is provided by Loria and Prendini (in press).