Echiniscus virginicus, Riggin, 1962

Gąsiorek, Piotr, Vončina, Katarzyna, Kristensen, Reinhardt Møbjerg & Michalczyk, Łukasz, 2021, Fig. 7 in Fig. 4. A in Saphonecrus globosus Schweger and Tang 2015, Zoological Studies (Zool. Stud.) 60 (70), pp. 1-45 : 3-4

The Echiniscus virginicus complex

A dataset of ITS-1, ITS-2 and COI from Gąsiorek et al. (2020) was used. The final alignment length was 1725 bp. Using PartitionFinder v.2.1.1 ( Lanfear et al. 2017) under the Bayesian information criterion (BIC), the best scheme of partitioning and substitution models for posterior phylogenetic analysis were chosen. The analysis was run to test all possible models implemented in MrBayes. As COI is a protein-coding gene, before partitioning, we divided our alignments of this marker into three data blocks constituting three separate codon positions. GTR+G was inferred to be the best-fit model for the first coding site of COI and a joined ITS-1+ITS-2 partition, GTR+I – for the second coding site of COI, and HKY+G – for the third coding site of COI. Bayesian inference (BI) marginal posterior probabilities were calculated using MrBayes v.3.2 ( Ronquist and Huelsenbeck 2003). Random starting trees were used and the analysis was run for ten million generations, sampling the Markov chain every thousand generations. An average standard deviation of split frequencies of <0.01 was used as a guide to ensure the two independent analyses had converged. The program Tracer v.1.6 ( Rambaut et al. 2014) was then used to ensure that Markov chains had reached stationarity and to determine the correct ‘burn-in’ for the analysis, which was the first 10% of generations. The ESS values were greater than 200 and a consensus tree was obtained after summarizing the resulting topologies and discarding the ‘burn-in’. All final consensus trees were visualised by FigTree v.1.4.3, available from http://tree.bio.ed.ac.uk/ software/figtree. The parameters and programmes were identical in the latter datasets if not specified otherwise.

Hypechiniscus phylogeny

A dataset of 18S rRNA, 28S rRNA and ITS-1 from Gąsiorek et al. (2021a) was used. The final alignment length was 2363 bp. PartitionFinder indicated the following models for predefined partitions: TRN+I+G (18S rRNA), GTR+G (28S rRNA) and TVM+I (ITS- 1). These models were used in BI reconstructions in MrBayes. ModelFinder ( Kalyaanamoorthy et al. 2017) was used to choose the best-fit models for Maximum Likelihood (ML) analyses—K2P+I+G4 (18S rRNA), TVMe+G4 (28S rRNA) and K3Pu+F+I (ITS-1)— according to the Bayesian information criterion. W-IQ-TREE was used for ML reconstruction ( Nguyen et al. 2015; Trifinopoulos et al. 2016). One thousand ultrafast bootstrap (UFBoot) replicates were applied to provide support values for branches ( Hoang et al. 2018).

Nebularmis phylogeny and biogeography

A dataset of 18S rRNA, 28S rRNA, ITS-1 and ITS-2 from Gąsiorek et al. (2021b) was used. The final alignment length was 2825 bp. PartitionFinder indicated GTR+I+G for two separate partitions (18S rRNA + 28S rRNA and ITS-1 + ITS-2). The original concatenated matrix was analysed using BEAST ( Drummond and Rambaut 2007). Four combinations of clock and tree priors were chosen and run in parallel, analogously to the analyses from Gąsiorek et al. (2021b): (a) a random local clock ( Drummond and Suchard 2010) with the coalescent tree prior, (b) a random local clock with speciation: Yule process as the tree prior, (c) a strict clock ( Ferreira and Suchard 2008) with the coalescent tree prior, and (d) a strict clock with speciation: Yule process as the tree prior. Tree searches were run for 10 million generations, sampling the tree every 1000 steps. The trees were summarized with TREEANNOTATOR software (distributed with BEAST), with the first 1000 trees removed. Tracer was then used to check the stationarity of Markov chains and determine the ‘burn-in’.

Consensus trees constructed from all datasets shared identical topologies (consistent with the variant b from Gąsiorek et al. 2021b). Consequently, the first 9000 trees were removed from the set of trees b, and the remaining 1000 trees were used in independent statistical dispersal-vicariance analyses (S-DIVA; Ronquist 1997; Yu et al. 2015), implemented in RASP ( Yu et al. 2020), with phylogenetic uncertainty considered in the calculations. Nebularmis records were assigned to the zoogeographic realms ( Ficetola et al. 2017) and records of N. reticulatus ( Murray, 1905) outside the Palaearctic were discarded as unreliable ( Gąsiorek et al. 2019c 2021b). The maximum number of areas at a node was set to 3.