A new species of rake-legged mite, Caeculus cassiopeiae (Prostigmata, Caeculidae), from Canada and a systematic analysis of its genus.

Abstract The genus Caeculus Dufour (Prostigmata, Caeculidae) contains 19 previously described species, most of which are found in North America, and for which no comprehensive phylogenetic treatment exists. Here, one new species from Alberta, Canada, is described: Caeculus cassiopeiae Bernard & Lumley, sp. nov., and another caeculid known to be present in Canada is documented. The new species is characterized within the genus with a character state matrix, from which an updated key is produced. A systematic analysis of all 20 species based on morphological and geographical distribution traits obtained from literature represents the first phylogenetic review of the genus.


Introduction
The Caeculidae contains 108 previously described species of large (750-3000 µm) prostigmatic mites in seven genera distributed worldwide, with 19 species in the genus Caeculus Dufour (Walter et al. 2009;Taylor et al. 2013;Mangová et al. 2014;Ott and Ott 2014;Taylor 2014;Fuangarworn and Butcher 2015;Rivas et al. 2016;Per et al. 2017;Porta et al. 2019). Employing the spiniform setae on legs I for which they are called rake-legged mites, caeculids are ambush predators of small arthropods including collembolans (Otto 1993). They do so camouflaged against rocky or sandy substrates in arid environments (Coineau 1974). Other than morphological reviews by Vitzthum (1933), Grandjean (1944), and Coineau (1974), a dichotomous key by Franz (1952), and original species descriptions, Caeculidae have not been widely collected or studied. Hence, a phylogenetic assessment of Caeculus has not yet been attempted.
In July 2014, we collected two specimens in yellow pan traps in Medicine Hat, Alberta, Canada. They represent a new species, Caeculus cassiopeiae sp. nov., described below. This record increases the number of known Canadian caeculids to two species. Evert E. Lindquist (Canadian National Collection, Ottawa, Ontario, Canada) collected the other species in Alberta's Writing-on-Stone Provincial Park in 1978, and the same species in Alberta's Waterton Lakes National Park in 1980, which are the only previously known collections from Canada and are deposited in the CNCI. He identified this species only to Caeculus and did not publish it, but herein we identify it as C. cremnicolus Enns. The only published record of caeculoids in Canada is by Lindquist et al. (1979), who listed at least one unidentified caeculid species in southern inland British Columbia based on personal communication with Valin G. Marshall (Canadian Forest Service, Victoria, British Columbia, Canada), but without collection or further identification information. Although Lindquist et al. (1979) estimated the potential discovery of 2 additional undetected species based on records in the United States, the updated catalogue of Canadian Acari still lists only one previously recorded caeculid species (Beaulieu et al. 2019).
We constructed a character state matrix to compare C. cassiopeiae sp. nov. to the descriptions of all other known species of Caeculus. In the absence of molecular data, we used the matrix as a phenotypic platform for a phylogenetic analysis of the genus, which illuminates the congeners most closely related to the new species, and provides a springboard for further assessment of the genus.

Materials and methods
On 27 July 2014 in Medicine Hat, Alberta, Canada, we collected two female caeculid specimens in yellow pan traps filled with soapy water (Marshall et al. 1994), which for two days were placed on both the arid SW slope of a coulee (glacially formed from sandstone and clay) and on the adjacent plain dominated by non-native crested wheat grass (Agropyron cristatum (L.) Gaertner) and alfalfa (Medicago sativa L.). We revisited the same locality on June 26, 2017 and using a paintbrush collected additional specimens, which were found only on open, exposed soil surfaces of the arid SW slope during the hottest part of the day (13:45-15:30, ≥ 32 °C). The soil surfaces had a thin hardened crust, possibly created by drying after rainfall.
According to the Köppen-Geiger climate classification system, Medicine Hat is a cold semi-arid steppe (BSk) (Peel et al. 2007) with a mean annual precipitation of 322.6 mm and a mean annual temperature of 6.1 °C (NCDIA 2017). Based on geological maps of the region (Berg and McPherson 2005) and the presence of well-rounded gravel and fine-grained sediment, the surficial geology is consistent with Quaternary alluvium. Medicine Hat furthermore has brown chernozem soil by the Canadian classification (Fuller 2010;NSBD 2017), synonymous with ustic mollisol in the USDA soil taxonomy (Haynes 1998).
After photographing the specimens collected in 2014 with a K2 DistaMax longdistance microscope (Infinity Photo-Optical, Boulder, Colorado, USA), we stored one in 95% ethyl alcohol (EtOH) and cleared the other in 85% lactic acid (Thermo Fisher Scientific, Waltham, Massachusetts, USA) and dissected it before mounting on a slide in a solution of 1.66 g polyvinyl alcohol, 10 mL lactic acid, 1 mL glycerol, and 10 mL distilled water (produced by Bioquip, Rancho Dominguez, California, USA) for analysis under dissection and compound microscopes, both of which contributed to creating the free-hand illustrations. With the 2017 specimens, we kept two alive for observation and stored the remainder in 95% EtOH.
In describing the idiosomal morphology of the new species, we followed the terminology outlined by Coineau (1974), which is based on the model of idiosomal divisions by Grandjean (1969) and other aspects of caeculid morphology described by Grandjean (1944). Notation for setae follows Coineau (1964Coineau ( , 1967aCoineau ( , b, 1969Coineau ( , 1974. As described by Coineau (1964), the eponymous characteristic of rake-legged mites is their elongated thickened spiniform leg setae, which are known as "rake" setae and are labelled as such herein.
To compare our specimens to other congeners using established criteria, we mined morphological and geographical distribution data from all known publications on Caeculus to construct a standard categorical character state matrix of the female of 23 taxa in Mesquite version 3.2 (Maddison and Maddison 2017), including all 20 species of Caeculus and three species of Neocaeculus Coineau (Table 1). We incorporated additional chaetotactic data into the matrix for C. echinipes Dufour from Jacot (1936), and for both it and C. americanus Banks from Coineau (1974). The resulting matrix includes mostly characters that are clearly described and/or illustrated in the species descriptions. Missing data are denoted with a "?" and uncertainty between states is characterized with a "/", following the notation used by Maddison and Maddison (2017). In rare circumstances we inferred characters that were consistently mentioned in the descriptions. For instance, because Mulaik (1945) noted tarsal bothridial setae for some species but not for others, we reasoned that he would mention the trait if present; thus if illustrations or text did not include a trait that had been described in other species by the same author, we treated the trait as absent. The matrix did not include information on ecology or internal anatomy because this was lacking in publications.
We conducted a parsimony analysis of these phenotypes with PAUP* version 4.0β10 (Swofford 2000), which involved a heuristic search with a tree-bisectionreconstruction branch-swapping algorithm for 5000 replicates. If the maximum branch length was zero, we set branches to collapse. The setation of N. imperfectus Taylor, Gunawardene & Kinnear resembles that described by Coineau (1974) as the holotrichous setal complement of Caeculidae, so we designated this species as the Table 1. Standard categorical character state matrix for Caeculus females, as well as three outgroup taxa, used for cladistic analysis. Polymorphism is denoted by "&", uncertainty by "/", missing data by "?", and inapplicable data by "-". " †" signifies a size character not included in phylogenetic assessment.    Table 2. Character states of Caeculus species (females) used for morphological comparison and cladistic analysis. Row number aligns to column number in Table 1. A setal pair refers to two setae in symmetry on either side of the mid-sagittal plane. An excluded seta is denoted by "excl." Size character is excluded from analysis.
Male and immatures unknown. Etymology. The b setal arrangement on the centrodorsal opisthosoma resembles the five-star constellation named for Cassiopeia, the vain wife of King Cepheus in Greek mythology. The constellation is also known as the "Celestial M" given its orientation to the horizon when it ascends in the night sky on its arc around Polaris, and is known as the "Celestial W" as it sets. Cassiopeia's rise is most observable in the evenings during northern autumns. The genitive epithet abides by Articles 11.9.1.3 and 31.1.2 of the International Code of Zoological Nomenclature, and hence means "Cassiopeia's rake-legged mite." Ethology. We observed individuals both walking and motionless with legs I positioned in a raptorial manner above the soil surface. A captive individual consumed a prostigmatan we collected from the same exposed soil.   .

Caeculus cremnicolus Enns, 1958
Diagnosis. Distinguished by its unfused mediodorsal opisthosomal sclerites, by its dark sclerites in adult mites, and by three barbed clavate b setae on the centrodorsal opisthosoma arranged in a triangle.

Phylogenetic analysis
The character state matrix contains 51 characters (Table 2), of which 27 are binary and the remainder occur in multiple states. The first 50 characters contributed to the phylogenetic analysis, and the last is a body length character that we retained only to  improve the utility of the matrix as an identification tool but not used in the phylogeny. Among the characters used for phylogenetics, 88% were parsimony-informative. PAUP* assessed 5.659 × 10 9 arrangements in 5000 replicates for the cladistic analysis, generating a 50% majority-rule consensus tree from 722 retained trees (Fig. 6). The parsimony score of the best tree was 199.
The phylogeny reveals three morphological clades, termed A, B, and C (Fig. 6). Caeculus calechius Mulaik is basal within the genus; this taxon and a few other species fall out independently from the clades. Clade A characterizes C. dorotheae Mulaik as sister to C. janetae Higgins & Mulaik (100%), and C. gertschi Mulaik sister to C. hypopachus Mulaik (100%). Despite lower branch support for the B clade, there is high support for C. kerrulius Mulaik being sister to C. echinipes (100%). Sister to clade B is the strongly supported C clade, which represents C. lewisi McDaniel & Boe as sister to C. cassiopeiae sp. nov. (100%), and this pair is sister to C. valverdius Mulaik (100%).

Taxonomy
Although Caeculus cassiopeiae sp. nov. is morphologically most similar to C. lewisi McDaniel & Boe and C. valverdius Mulaik, several noteworthy differences exist (Table 1). The dichotomous key below lists six traits that distinguish the new species from C. lewisi. Additionally, following Franz (1952) and McDaniel and Boe (1990), the new species keys to C. valverdius, yet several traits separate them as well: (i) C. valverdius bears six b setae on the centrodorsal opisthosomal sclerite, but five b setae are arranged in an M-shape in the new species; (ii) each laterodorsal opisthosomal sclerite has a single c seta in C. valverdius whereas the new species has two c setae; (iii) the posterior opisthosomal sclerite has five e setae in C. valverdius whereas the new species has seven; (iv) C. valverdius has six setae on each progenital valve whereas the new species has seven; (v) the adanal sclerites each have three setae in C. valverdius and two in the new species; (vi) genu I bears a single posteroventral rake seta in the new species whereas C. valverdius has four; and (vii) C. valverdius has two anterolateral setae on the proximal half of trochanter III, but the new species has three that are more evenly distributed along the length.

Phylogenetic analysis
A few apomorphies denote the relationships within clade A; C. dorotheae and C. janetae both have six anteroventral rake setae on tibia I, which is a unique character. Caeculus gertschi and C. hypopachus each bear four rake setae in that location, a character shared with some members of clade B. Of the species in clade A, most are not recorded as having tarsal bothridial setae bt, which occur in most other Caeculus as well as in the outgroups, indicating possible plesiomorphy for the rest of the genus. One exception is C. gertschi (Mulaik 1945). Caeculus pettiti Nevin and C. mariae Higgins & Mulaik also lack this trait, possibly resulting from homoplasy.
Aside from C. krantzi Coineau, the members of clade B exhibit dark dorsal idiosomal sclerites in adults, although this trait may be homoplastic as it recurs sporadically in the other clades and some taxa not in clades.
All members of the C clade possess five anteroventral rake setae on tibia I, except C. tipus Mulaik, which has four. However, McDaniel and Boe (1990) also describe a male and a nymph C. lewisi with four rake setae in this position. The species of clade C furthermore have four posteroventral rake setae on tibia I, although there are two such rake setae on C. americanus, basal in the clade. Rake seta number is regarded as more stable than other traits (Coineau 1974), so the above traits may be autapomorphies for the clade. Another potential apomorphy for clade C is two or three pairs of aspidosomal Pa setae, although Mulaik (1945) also describes C. dorotheae with three pairs of Pa. All other congeners and all outgroup taxa have a single pair of Pa setae, which likely describes the ancestral state for the genus. Additionally, clade C species all have four to six centrodorsal opisthosomal b setae, also present in C. crossleyi Hagan and C. pettiti, so this may represent synapomorphy. The scarcity of autapomorphies in the topology likely reflects the variable nature of many traits in Caeculidae (Grandjean 1944;Coineau 1974).  Lakes National Park has substrate that is classed as dystric brunisol/cryept inceptisol (NSBD 2017). These disparate abiotic conditions indicate that C. cremnicolus is a habitat generalist. As the localities for C. cremnicolus and C. cassiopeiae sp. nov. are 126 km apart and in the same climate, their ranges could potentially overlap if C. cremnicolus can inhabit the soil at Medicine Hat. These species are nevertheless in separate clades (Fig. 6), signifying multiple introductions of the family into Canada. Such a pattern of introductions may be attributable to postglacial changes in biome distributions, which enabled fragmented xeric populations to expand northward from arid refugia after the last glacial maximum 18,000 years ago (Riddle and Hafner 2006; Graham et al. 2013). This possibility is intriguing considering that at the time Medicine Hat was near the southern limit of the Laurentide Ice Sheet, and by 14,000 years ago it was the first area in Canada freed of the ice as it became semi-arid grassland (Dyke 2005). Likewise, Thaler et al. (1993) described a new locality of Microcaeculus austriacus in Austria that was ~440 km from its known locations in the easternmost extent of the Alps, and alluded to postglacial warming as a possible explanation for the scattered distribution.

Key to adult
Our phylogenetic analysis suggests that the common ancestor of Caeculus inhabited southwestern North America, based on the known locations for C. calechius and Clade A (the most basal clade). The other clades also contain representatives from the North American southwest as well as those from other areas, and there is a diversity of locations represented by the taxa that are not in clades. Caeculus echinipes is the only species of Caeculus described to date from Europe. The results suggest that the ancestor of C. echinipes spread to Europe from North America. Further work to include molecular data would be helpful to clarify the weaker branch support shown in the topology.

Conclusions
Our description of Caeculus cassiopeiae sp. nov. elevates the number of known Canadian caeculids to two. Based on a maximum parsimony analysis of a character state matrix for all members of the genus, it is most closely related to C. lewisi. The phylogeny also suggests multiple introductions of caeculids into Canada, and that the origin of the genus is the American southwest. Further collection of caeculids in North America is required to determine to what extent ranges overlap, examine ecology or additional morphological traits (e.g., minute sensory structures) that were not described in previously published taxonomic accounts, and to enable genetic analysis. Population studies can further describe the degree of intraspecific variation and clarify species boundaries. Our phylogeny provides the first analysis of the genus, which can be useful for future systematic studies that integrate taxonomy and genetics to develop a better understanding of the genus Caeculus and its family.