Porcellionidae is one of the richest families of Oniscidea globally distributed but we still lack a comprehensive and robust phylogeny of the taxa that are assigned to it. Employing five genetic markers (two mitochondrial and three nuclear) we inferred phylogenetic relationships among the majority of Porcellionidae genera. Phylogenetic analyses conducted via Maximum Likelihood and Bayesian Inference resulted in similar tree topologies. The mtDNA genes cytochrome oxidase I (COI) and 16s rRNA (16s) were used for clade dating using previously published mutation rates. Our results provide evidence against the monophyly of both Porcellionidae and the largest genus of the family Porcellio. These results are compared to previous published work based on morphological evidence. The genera Leptotrichus and Brevurus are not grouped with the rest of Porcellionidae whereas Agnaridae are grouped with part of Porcellionidae. Armadillidium and Schizidium (Armadillidiidae) occupy a basal position on the phylogenetic tree. Even though the African genera Tura and Uramba (distributed in East Africa) are grouped together there is no general geographical pattern in other sub-clades. Additional taxonomic issues that arise in this work such as the assignment of the recently described genus Levantoniscus, arealso discussed. The status of Porcellionidae should be further revised and morphological characters traditionally used in Oniscidea taxonomy should be reconsidered in view of molecular evidence. The origin of the monophyletic clade within Porcellionidae as indicated in the present work is dated back to the Oligocene (~32 mya).

Crinocheta , genetic markers, monophyly, node dating, taxonomic revision, woodlice

The Oniscidea family Porcellionidae is one of the richest in species, with 333 species, belonging to 19 genera, currently assigned to it (Sfenthourakis and Taiti 2015). Family members are unable to conglobate, with the exception of the genus Atlantidium Arcangeli, 1936. There is remarkable morphological variation among Porcellionidae species and genera, especially in head structure, pleotelson, and body shape. Familial assignment of taxa is based mostly on the combination of two character states, namely an antennal flagellum with two articles and the presence of monospiracular, covered lungs on the first two pairs of pleopods (Schmidt 2003). However, certain authors, based on morphological and recent molecular work, suggest that these characters could be symplesiomorphies, as they are not exclusively found in Porcellionidae (Schmalfuss and Ferrara 1978, Schmidt 2003).

Different authors have found Porcellionidae to be closely related with Oniscidae, Trachelipodidae, Cylisticidae, Agnaridae or Armadillidiidae (Michel-Salzat and Bouchon 2000, Mattern 2003, Schmidt 2008, Lins et al. 2017). Furthermore, monophyly of the most species-rich genera, Porcellio Latreille, 1804 and Porcellionides Miers, 1877, has been debated on the basis of both morphology (Vandel 1962, Schmalfuss 1992, 1998) and molecular evidence (Michel-Salzat and Bouchon 2000, Mattern 2003). More specifically, some Porcellionides species appear to be more closely related to the genus Porcellio (Michel-Salzat and Bouchon 2000, Mattern 2003) or even to the genus Cylisticus Schnitzler, 1853 that belongs to another family (Cylisticidae),than to other congeneric species (Michel-Salzat and Bouchon 2000). Hence, also the monophyly of the family has been repeatedly questioned on the basis of both morphological and genetic data (Schmalfuss 1989, Michel-Salzat and Bouchon 2000, Mattern and Schlegel 2001, Schmidt 2003, 2008).

Members of Porcellionidae were originally reported from the circum-Mediterranean region, Atlantic islands, Arabian Peninsula and East Africa. Nowadays they are known from all over the world, being introduced into many regions by human activities (Schmidt 2003). Porcellionidae are considered to be among the isopod species that are better adapted to terrestrial environments, and they can be found in a wide range of habitats, from tropical rainforests to deserts (Schmidt 2003, Medini-Bouaziz et al. 2017).

The present study aims to a more detailed investigation of phylogenetic relationships among genera of Porcellionidae, using two mitochondrial and three nuclear genes that allow estimation of divergence times among extant taxa.

At least four out of five targeted genes were successfully amplified and sequenced for the great majority of available individuals, with final DNA extraction yield over 20 ng/μl and A260/A280 purity rate over 1.5.Since some important samples were old (collected more than two decades ago, mainly from Africa) or ill-preserved for a long time (i.e., in 70% alcohol) we didn’t manage to retrieve sequences from all targeted genes. However, specimens not represented by all gene fragments were also included in the analyses. The final concatenated alignment obtained consisted of 3,841 base pairs (bp). More details about the aligned sequences length, conserved, variable and parsimony-informative sites for each gene are given in Table 2.

Available sequences were separated in different groups at the genus level except for Porcellio species which were treated as different groups due to the alleged non-monophyly of the genus. Between groups p-genetic distances for each gene are given in Suppl. material 1.

The best-fit nucleotide substitution models for each partition/gene selected under the BIC criterion were (for both MRBAYES and BEAST) the HKY+G+X, HKY+G+X, TRNEF+G, TRN+G, TRN+G+X and GTR+G+X for COI, 16s, 18s, 28s and NAK genes, respectively. The selected model under --raxml commandline option at PartionFinder was the GTR+G (-ln =26511.0556641) for all genes.

Maximum Likelihood and Bayesian Inference analyses (implemented both in BEAST and MRBAYES) resulted into phylogenetic trees with similar, well-supported topologies. Given the congruence among the results of the two methods, only the Bayesian tree is presented herein (Figure 1). The ML tree is given in Suppl. material 1: (Figure S1). The separate analysis of different gene markers showed that the concatenated tree topology is mainly determined by nuclear genes. Missing data, and possibly also the depth of the phylogeny, led to largely unresolved trees for mtDNA markers. Nevertheless, these were used mainly to estimate node dates based on published mutation rates. The poor mtDNA-based resolution did not affect the final tree, given that the tree based solely on nuclear genes (see Suppl. material 1) has identical topology.

Figure 1. 

Dated phylogram based on concatenated data set including five genes (COI, 16s, 18s, 28s, NAK), generated using a relaxed lognormal clock in BEAST. BI posterior probabilities (>0.9) and ML bootstrap values (>60) are presented above the nodes. Estimated mean divergence time is given below the nodes only where nodes are statistically supported or the topology was identical between BI, ML and BEAST analyses. Subclades including individuals from more than one species have been collapsed to genus level, since all (except Porcellio) were monophyletic. Abbreviations: P. Porcellionidae, T. Trachelipodidae, A. Agnaridae, R. Armadillidiidae. Numbers in parentheses after each taxon name refer to numbering of taxa in Table 1.

Our results provide evidence against the monophyly of both the family Porcellionidae and the genus Porcellio. Brevurus appears to belong to a supported distant clade, external to that formed by the remaining Porcellionidae+Trachelipodidae+Agnaridae. Leptotrichus is an external branch to Agnaridae + part of Porcellionidae. Monophyly of Agnaridae is supported. Levantoniscus forms the sister clade of all monophyletic Porcellionidae. Finally Armadillidiidae branches early in the tree, not showing any close relationship to Porcellionidae.

The African genera Tura and Uramba are sister taxa sharing a common ancestor at around 22.2 mya (95% HPD 12.1 – 33.5 mya) and are grouped with Agabiformius. On the other hand, Thermocellio, also distributed in Kenya and the neighboring Tanzania, appears to be more closely related to Porcellio laevis, native to Europe and North Africa. Another African/Atlantic genus, Soteriscus, forms a well-supported clade with Lucasius and Mica that are distributed in Africa and on some Mediterranean islands. The Mediterranean genera Acaeroplastes, Caeroplastes, Porcellionides and Proporcellio, together with part of Porcellio, are grouped in the most derived clade that diverged at around 27 mya.

The genus Porcellio as currently perceived is represented in two well-supported separate clades. P. laevis groups with Thermocellio while P. nasutus with Acaeroplastes in a clade also including Caeroplastes.

Genetic distances between Porcellionidae genera (or species in the case of the non-monophyletic Porcellio) varied significantly among genes. The range of variation per gene is: COI: 16.9–50.3 %; 16s: 16.9–36.5 %; 18s:3.6–28.5 %; 28s:0.4–44.2%; NAK: 2.3–9.1%. The p-distances between Trachelipus and Agnara for NAK, and P. laevis and Lucasius for 18s, could be artifacts due to the comparatively shorter sequence length in Agnara and P. laevis, respectively (see Suppl. material 1).

It is worth noticing also that minimum and maximum distances are not exhibited by the same taxa for all genes. More specifically, highest / lowest genetic divergence is found between the following groups: Tura - Porcellio nasutus / Soteriscus – Leptotrichus (16s), Porcellio laevis – Lucasius / Proporcellio - Porcellionides (COI), Agabiformius - Porcellio nasutus / Caeroplastes - Acaeroplastes (18s), Brevurus - Thermocellio / Porcellio laevis - Thermocelio (28s) and Uramba – Brevurus / Proporcellio - Porcellionides (NAK). The allegedly congeneric Porcellio species never exhibit a minimum genetic distance.

This is the first comprehensive study aiming to resolve phylogenetic relationships among Porcellionidae genera using a multi-locus approach, thus increasing reliability of results. Our findings undermine the monophyly of both the family Porcellionidae and the genus Porcellio, in line with suggestions by previous authors (Schmalfuss 1989, Mattern 2003, Michel-Salzat and Bouchon 2000, Schmidt 2003, 2008).

The extremely high genetic distances, which reached up to 50.3 in mtDNA and 44.2 in nDNA, are confirming the vast divergence among taxa within Porcellionidae. Observed inconsistencies of group distances among different genes highlight the usefulness of the multi-locus approach followed herein for a reliable phylogenetic reconstruction of the taxa examined.

In view of the herein estimated phylogeny, a monophyletic Porcellionidae should exclude Brevurus and Leptotrichus. Moreover, the supposedly subtle morphological differences between Leptotrichus and Agabiformius that had led to a presumed sister-group relationship between these genera, are misleading, since they are found to be very distant (Schmalfuss 2000, Verhoeff 1908). Brevurus has been proposed as a possible synonym of Porcellium Dahl, 1916 (a genus of Trachelipodidae) (Khisametidonova and Schmalfuss 2012), an hypothesis that cannot be evaluated in view of our results.

The genus Levantoniscus, tentatively assigned to Trachelipodidae (Cardoso et al. 2015), has been found to be closer to the monophyletic subgroup of Porcellionidae. Given that the genus appears as the sister clade of all remaining monophyletic Porcellionidae, we cannot propose the assignment of this taxon into the same family, given that no known morphological characters can be used as synapomorphies of such a taxon. The characters considered as autapomorphies of Levantoniscus by Cardoso et al. (2015) could as well define a separate new family. A more inclusive phylogeny is required before we can decide on its familial status, given also the lack of robust synapomorphies defining Trachelipodidae, a family in need of a sound revision.

As indicated by the tree topology, Porcellionidae is more closely related to Trachelipodidae and Agnaridae rather than Armadillidiidae. A similar result has been found by Lins et al. (2017), even though these authors had included only two species in two genera (Porcellio and Porcellionides) of Porcellionidae in their analysis. It is evident that morphological characters traditionally used in Oniscidea systematics, such as the structure of pleopodal lungs, the number of flagellar segments and the head structure, do not seem to provide adequate evidence that support a robust taxonomy, at least not in all cases.

In conclusion, the monophyly of Porcellionidae as currently perceived cannot be supported by molecular evidence. Of course, we still need to identify phenotypic synapomorphies defining the family, since the characters used so far cannot be considered as valid. In addition, the genus Porcellio needs to be revised, as it appears to be polyphyletic, comprising of at least two separate groups.

The monophyletic subgroup of Porcellionidae seems to have an African origin, diverging at the end of the Palaeogene (Oligocene) and then differentiating further during the Miocene. Based on the cladochronology estimated herein, more basal cladogenetic events, leading to the branching of other related families, happened in the Eocene. This chronology is compatible with the very old (Mesozoic) origin of Oniscidea suggested by Broly et al. (2013, 2015).

We are grateful to Prof. Nikos Poulakakis and Emmanouela Karameta for their support with data processing.