DNA barcoding and integrative taxonomy of the Heterolepisma sclerophylla species complex (Zygentoma: Lepismatidae: Heterolepismatinae) and the description of two new species

We present one of the first studies of DNA barcodes (COI sequences) in the basal insect order Zygentoma, and compare the data with nuclear (28S) and mitochondrial (16S) rDNA sequences and morphology for an integrative taxonomic study of the Heterolepisma sclerophylla Smith species group. DNA sequence analyses identified deep divisions between Queensland and New South Wales populations, and among populations in each state. Detailed morphological and morphometric evaluation of the specimens failed, in most cases, to identify unambiguous morphological characters of diagnostic value for each population, possibly due to the interaction of morphological conservatism with high levels of variability resulting from their continued moulting after reaching sexual maturity. Several strong consistent characters were identified to support the description of a southern Queensland population as a new species (Heterolepisma cooloola sp. nov.). The combined molecular and morphological data support the view that the presence of lanceolate scales and the absence of macrochaetae from the anterior margin of the frons are more significant to phylogeny than the arrangement of styli and the shape of the thoracic sternites in Heterolepisma. Specimens from Glen Davis, NSW, while indistinguishable from H. sclerophylla in all other characters examined, were found to possess one fewer pair of abdominal styli in both sexes and are also described as a new species (Heterolepisma coorongooba sp. nov.). Five lineages are recognized within the remaining NSW material but as reliable (non-overlapping) morphological and morphometric differences could not be identified, they are not described here as new species. Heterolepisma sclerophylla sensu stricto is considered to be a complex of morphologically ill-defined species or perhaps subspecies. The silverfish subfamily Heterolepismatinae Mendes, 1991 is poorly understood but quite diverse in spite of a certain superficial uniformity. Twenty-four species have been described with a mainly Gondwanan distribution extending to coastal southern Japan, Vietnam, Angola, Mozambique, Somalia, Zanzibar as well as to many Pacific and Indian Ocean Islands, but Smith (2017) suggested that there may be more than 100 morphospecies in Australia alone. In Australia the genus can be found in habitats ranging from the fringes of tropical rainforest to the dry deserts and even rarely in subalpine regions. Specimens are collected from dry leaf litter, under or within cavities in the bark of dead or living trees, sometimes under stones and occasionally within abandoned termite galleries. Several authors (Wygodzinsky, 2 Records of the Australian Museum (2019) Vol. 71 1967, Irish, 1990 and Mendes, 1991) consider the genus to exhibit many plesiomorphies although Smith (2016b) discusses alternative interpretations. The species Heterolepisma sclerophylla Smith was described from specimens collected at Broulee on the NSW south coast (Smith, 2014). Other localities listed included specimens from Glenbrook, Guerilla Bay, Lower Portland, Megalong Valley, Burralow (near Bilpin), Clandella State Forest, Hawkesbury Heights, Ku-ring-ai, Nattai and Wellington, all in NSW (Fig. 1). The species was distinguished from other Heterolepisma by a combination of features including a glabrous urosternite I, 1+1 single macrochaetae on urosternites II–VIII, styli on urosternites VII–IX in the ♀ and VIII–IX in the ♂, 2+2 combs on urotergite I, posterior combs on all nota, urotergal submedial combs each of two macrochaetae, a small apically-round, triangular prothoracic sternum which is longitudinally somewhat furrowed, and a rounded (rather than truncate) urotergite X. Variations in morphology between individuals (and even from left to right on a single individual) were mentioned in the original description but considered as “normal” given the continuous moulting of silverfish throughout their life, even after attaining sexual maturity. The large and “unlimited” number of moults during the Figure 1. Collection localities. life of silverfish results in considerable variation between specimens and makes morphological comparisons difficult. Material with the same characters listed above has since been collected from a wide range of locations in Queensland, including Carlo Point and Cooloola in South East Queensland, Undara, Chillagoe and near Bamaga at the extreme northern tip of eastern Australia. Material was also obtained from further localities in New South Wales (Fig. 1). The material from Glen Davis was found to resemble H. sclerophylla in most aspects but had one fewer pairs of styli in both sexes. The number of pairs of styli has been considered variable in this genus and some species, notably Heterolepisma stilivarians Silvestri, from Western Australia, have been described as having variable numbers of styli (Silvestri, 1908). All the NSW and QLD material had been identified by the first author as belonging to, or being close to, H. sclerophylla, however preliminary molecular studies by the second author found distances of around 8% in DNA barcode sequences (the 5'-half of the COI gene) between populations from the type locality and the Blue Mountains (Megalong and Lower Portland) and even greater divergence from the Cooloola, southern Queensland, material (circa 14%). These large distances are similar to observed distances between the described species H. highlandi Smith, 2014 G. B. Smith et al.: Silverfish of the Heterolepisma sclerophylla complex 3 and H. buntonorum Smith, 2016, which raised the question of whether H. sclerophylla contains cryptic species, and prompted further study. The fourth author also performed preliminary analyses of the mitochondrial 16S rDNA gene for the population of “sclerophylla” from Cooloola and individual specimens from North Nowra and Bamaga, with similar results. Further sequencing of both DNA barcodes and a nuclear gene, the D9-D10 region of the 28S ribosomal DNA, was conducted on several specimens each from Broulee, Megalong, Wellington and Cooloola/Carlo Point and a few specimens from other named localities. Detailed morphological and morphometric comparisons of several specimens from each locality were conducted. Some characters, such as the presence of lanceolate scales and the number of divisions in the ovipositor cannot be easily determined in alcohol material, so much of the material was dissected and mounted before examination. Materials and methods Abbreviations and definitions Roman numerals are used to indicate abdominal segment number. In addition, the following abbreviations are used: AM: Australian Museum, 1 William St, Sydney 2010 Australia; asl: above sea level (in metres); HW: head width (in millimetres); H+B: head and body length (in millimetres); L/W: length to width (ratio); NSW: New South Wales; PI, PII, PIII: legs of prothorax, mesothorax and metathorax respectively; PCA: principal component analysis; penult: penultimate, referring to second last article of maxillary palp; QLD: Queensland; QM: Queensland Museum, Grey St & Melbourne St, South Brisbane QLD 4101; ult: ultimate (referring to last article). The prefixes pro, meso and meta are affixed to thoracic characters such as sterna, tibia and tarsus. Specimens are currently stored in 80% ethanol/water unless specifically mentioned as being in 100% ethanol or else mounted on slides using Tendeiro medium. The term macrochaetae refers to the larger stronger bristles (always apically bifurcated), setae to smaller thinner bristles (bifurcated or simple), setulae to the very small, usually straight, setae associated with the combs and cilia to the curly thin hairs, often associated with the combs, setal collar or notal margins. Collection of material, locality data and preparation Details of the material examined are included with the description of each species. Locality co-ordinates were obtained using a hand-held Garmin eTrex®10 GPS with a claimed accuracy usually under 5 m. Dissected specimens have been mounted on two slides using Tendeiro medium (one slide with head and thorax, the other with the abdomen). Drawings were made with the aid of an Olympus CX31 binocular microscope fitted with a U-DA drawing attachment. The holotype and a paratype of Heterolepisma cooloola are deposited with the Queensland Museum in Brisbane. Most of the remaining material is deposited with the Australian Museum in Sydney and each slide or specimen has an accession number starting with K (e.g., K.377269). Some material is still held by the first author, refrigerated in 100% ethanol. This material carries the author specimen data base number (e.g., gbs004932) and will eventually also be deposited with the Australian Museum. Sampling, DNA extraction, PCR and DNA sequencing Table 1 lists the 68 specimens subjected to DNA analysis, their collection localities, species identification, and BOLD and GenBank accession numbers for their DNA sequences. Species from three genera of Ctenolepismatinae were used as outgroups: Ctenolepisma longicaudata Escherich, Qantelsella louisae Smith, Acrotelsella erniei Smith and Acrotelsella parvelar Smith. DNA extractions performed at the AM (for DNA barcode/ COI and 28S rDNA sequences) used either the Bio Basic EZ-10 96 well plate Genomic DNA Isolation Kit (Astral Scientific, Taren Point, NSW) or the Bioline Isolate II Genomic DNA Kit (Bioline, Eveleigh, NSW) following the manufacturers’ protocols, with exceptions noted below. DNA extractions performed at Marist College, New York (for 16S rDNA sequences) used a Qiagen DNeasy Blood and Tissue Kit (Qiagen, Germantown, MD). About five whole specimens from each key locality (Broulee, Megalong, Wellington and Cooloola/Carlo Point), collected directly into 100% ethanol, were soaked in DNA extraction buffer containing proteinase-K at room temperature for three hours. The remaining cuticle was returned to 100% ethanol and later dissected in 80% ethanol and mounted on to slides using Tendeiro medium (usually head and thorax on one slide, abdomen on the other). DNA for specimens from most other localities was extracted from legs which had been removed from the specimen shortly after collection and stored in 100% ethanol at 4°C. The rest of the specimen was kept in 80% ethanol at room temperature. Polymerase Chain Reaction (PCR) amplification of the DNA barcode region of the mitochondrial COI gene used the primers and followed the method of Mitchell (2015). For the 28S rDNA D9-D10 region, we used one forward (28S_8fm) and two reverse PCR primers (28S_10rm and 28S_11rm), which were simply 5'-M13-tailed versions of Machida and Knowlton’s (2012) primers [28S] #8, [28S] #10_RC and [28S] #11_RC, respectively. PCR conditions for both genes followed those reported in Mitchell (2015) for COI. PCR products were purified using ExoSAP and sequenced in both directions using ABI Big Dye Terminator v.3.1 chemistry by Macrogen Inc. (Seoul, South Korea). The 16S rDNA gene was amplified using primers 16Sar and 16Sb (Edgecombe et al. 2002). Amplification, PCR purification, sequencing, and chromatogram analyses were carried out as in Espinasa and Cappuccio (2008). One 16S sequence was downloaded from GenBank for Tricholepidion gertschi to serve as an outgroup. DNA sequence assembly and phylogenetic analysis Forward and reverse direction sequence trace files were assembled using Geneious v. 9.1.7 (Kearse et al., 2012) and consensus sequences were aligned using Muscle (Edgar, 2004) and adjusted by eye. DNA sequences, sequence trace 4 Records of the Australian Museum (2019) Vol. 71 files, and specimen collection data were uploaded to BOLD (Ratnasingham and Hebert, 2007) and GenBank. Two 16S sequences (samples K.377738 and K.377751) were uploaded to BOLD, but were not accepted by GenBank as they are less than 200 bp in length. Four data sets were constructed: one for each of the three genes alone, COI, 28S, and 16S, and a concatenated gene data set. The COI alignment was trimmed from the 3'-end to 559 nucleotides to minimize missing data before phylogenetic analysis. The concatenated gene data set comprised only samples for which both 28S and COI sequences has been obtained. For outgroups C. longicaudata and Acrotelsella spp. the 28S and COI sequences were generated from different specimens (and for Acrotelsella from different species) thus the sequences are labelled with “Chimera” rather than with a Museum accession number in Fig. 4, although Table 1 lists the accession numbers of the specimens used for each gene. We did not include 16S sequences in the concatenated data set because there was only one specimen for which we had both 16S and other sequence data. FABOX v. 1.4.1 (Villesen, 2007) was used to edit sequence names. MEGA v.7.0.26 (Kumar et al., 2016) was used to calculate genetic distances. For phylogenetic analysis, nucleotide substitution models, and data partitions (codon positions) for COI, were tested using PartitionFinder2 (Lanfear et al., 2012) on the CIPRES computing platform (Miller et al., 2010), with the most appropriate model selected using the Bayesian Information Criterion. For 16S the best model was GTR+G, and for 28S the best model was K80+G. For COI the data had two partitions, codon positions 1 and 2 (best model TRNEF+I+G) and codon position 3 (best model HKY+G). For the concatenated data set PartitionFinder selected the same two partitions and models as COI, with 28S included in the partition with COI codon positions 1 and 2. Exploratory phylogenetic analyses were performed in Fasttree 2 (Price et al., 2010) while final analyses were performed using Maximum Likelihood (ML) as implemented in RAxML 7.2.8 (Stamatakis, 2006) and Bayesian Inference (BI) in MrBayes v.3.2.6 (Ronquist et al., 2012) using the plugins available in Geneious. RAxML analyses used the ML search convergence criterion and performed 1,000 fast bootstrap replicates. MrBayes analyses used four heated chains with chain temperature = 0.2, a chain length of 2,000,000 generations, subsampling every 1,000 generations, and a burnin of 500 samples (25%). Convergence was assessed by examining the average standard deviation of splits frequencies, with values below 0.01 indicating convergence. Convergence was reached in two million generations for the single gene analyses and in one million generations for the concatenated genes data set. Morphological and morphometric evaluation Specimens were measured in alcohol according to the methodology in Smith (2013). Due to the continuous moulting, most measurements were compared as ratios, either length to width, or compared to head width. While some immature specimens were measured, especially when fewer “adult” specimens were available, the measurement data is intentionally biased towards the largest available sexually mature individuals. Consistent measurement data is difficult to generate, not only due to the large number of instars but it is also difficult to align the part to be measured completely horizontal. Some specimens are bent over and cannot be flattened for accurate measurement. Being softbodied the specimen can be distorted due to damage or preservation, with the various segments either contracted into each other or distended hence the unreliability of H+B, thorax and abdomen lengths. In addition, the nota can sometimes have lost their curvature and hence appear wider than expected probably because the specimen was partially squashed during collection. Ranges of measurement data are often quite wide and it is therefore difficult to find parameters where there is no overlap in the range. Some measurements, e.g., L/W of inner processes of coxites IX, L/W of thoracic sterna and L/W of urotergite X, can only be made using dissected material. The number of divisions in each of the ovipositor valves must also be made with dissected material. The following measurement ratios were compared: thorax length/HW, thorax length/H+B, thorax length/width (widest part of mesonotum), abdomen length/H+B, width/HW of all nota, thorax length/abdomen length, H+B/HW, L/W of pedicel, L/W of the scape, the length of the ultimate article of the maxillary palp relative to its width, to head width and the length of the penultimate article, the length of both pedicel and scape relative to each other and to head width, length of tarsi of all legs/HW, length of tarsus /tibia of all legs, L/W of ultimate article of maxillary palp, length of ultimate/ penultimate articles of maxillary palp, L/W ultimate article of labial palp, length of pedicel/scape, L/W of thoracic sterna, their width relative to head width, their length relative to each other, the width of the gap between the combs on the metathoracic sternite relative to the average length of the combs, the length of all tarsi and tibiae relative to both their width and HW, the length of the tibia and tarsi of PI relative to PIII, the L/W of urotergite X and the inner and outer processes of coxites IX (the latter independently for males and females), length stylus VII/IX, length of stylus VIII/ IX and the length of the ovipositor relative to head width. Morphological characters compared include the presence and distribution of lanceolate scales, the chaetotaxy of the frons, clypeus and labrum, the relative size and chaetotaxy of the pedicel and scape, the chaetotaxy of the outer face of the mandibles, the length of molar region and the chaetotaxy at the end of molar region, the number of teeth on lacinia and its length relative to galea, the number of lamellae and setae on the lacinia, the chaetotaxy of the articles of the maxillary palp, the relative length of the most distal articles and the presence, type and location of the branched papillae on the ultimate article of both the males and females, the chaetotaxy of the submentum and mentum, the chaetotaxy of the labial palp and the shape of the last article and the arrangement of its papillae, the density of anterior collar of the pronotum, location of trichobothrial areas and their association with macrochaetae, the lateral and posterior chaetotaxy of all nota, the chaetotaxy of the presternum, the shape, size and chaetotaxy of all thoracic sterna, the scale covering and chaetotaxy of PI and PIII, the chaetotaxy of the urotergites and urosternites (position of combs as well as the number of macrochaetae, marginal setae, setulae and cilia), the shape and chaetotaxy of urotergite X, the number and size of the styli, the number of divisions of the ovipositor, the size, position and chaetotaxy of the paramera and the distribution of pigment. G. B. Smith et al.: Silverfish of the Heterolepisma sclerophylla complex 5 Table 1. Museum, BOLD and GenBank accession numbers for all sequences obtained. BOLD Numbers GenBank Accession Numbers Museum ID Species Extra Info State Site Sample ID Process ID COI 28S 16S K.261126 H. cooloola sp. nov. paratype QLD Cooloola gbs004890 ZYI042-17 MF040956 — — K.261128 H. cooloola sp. nov. paratype QLD Cooloola gbs004891 ZYI043-17 — KY951403 — K.261130 H. cooloola sp. nov. paratype QLD Cooloola gbs004892 ZYI044-17 MF040955 KY951402 — K.261132 H. cooloola sp. nov. paratype QLD Cooloola gbs004894 ZYI045-17 MF040954 — — K.261134 H. cooloola sp. nov. paratype QLD Cooloola gbs004895 ZYI046-17 MF040953 — — K.261136 H. cooloola sp. nov. — QLD Carlo Point gbs004868 ZYI038-17 MF040951 KY951399 — K.261138 H. cooloola sp. nov. — QLD Carlo Point gbs004873 ZYI039-17 — KY951406 — K.261140 H. cooloola sp. nov. — QLD Carlo Point gbs004877 ZYI040-17 — KY951405 — K.261180 H. cooloola sp. nov. — QLD Carlo Point gbs004867 ZYI037-17 — N/A [<200 nt] — K.261182 H. cooloola sp. nov. — QLD Carlo Point gbs004879 ZYI041-17 MK185704 KY951404 KY951372 K.261186 H. cooloola sp. nov. paratype QLD Cooloola gbs003331 ZYI059-17 — — KY951367 K.377746 H. cooloola sp. nov. — QLD Cooloola gbs003324 ZYI060-17 — — KY951368 K.377748 H. cooloola sp. nov. — QLD Cooloola gbs003353 ZYI016-17 MF040952 KY951400 — K.377752 H. cooloola sp. nov. — QLD Carlo Point gbs004871 ZYI061-17 — — KY951371 K.377753 H. cooloola sp. nov. — QLD Carlo Point gbs004883 ZYI062-17 — — KY951369 K.377754 H. cooloola sp. nov. — QLD Cooloola gbs004893 ZYI063-17 — — KY951370 QM207012 H. cooloola sp. nov. paratype QLD Cooloola gbs003330 ZYI015-17 — KY951401 — K.261204 H. coorongooba sp. nov. paratype NSW Glen Davis gbs004629 ZYI024-17 MF040960 KY951410 — K.261208 H. coorongooba sp. nov. holotype NSW Glen Davis gbs004626 ZYI021-17 MF040957 KY951407 — K.261210 H. coorongooba sp. nov. paratype NSW Glen Davis gbs004627 ZYI022-17 MF040958 KY951408 — K.261212 H. coorongooba sp. nov. paratype NSW Glen Davis gbs004628 ZYI023-17 MF040959 KY951409 — K.260992 H. sclerophylla paratype NSW Broulee gbs001024 ZYI001-17 MF040943 KY951392 — K.260994 H. sclerophylla — NSW Guerilla Bay gbs001032 ZYI002-17 MF040942 — — K.261084 H. sclerophylla — NSW Nattai gbs003094 ZYI014-17 — KY951386 — K.261142 H. sclerophylla — NSW Broulee gbs004969 ZYI048-17 MF040946 KY951396 — K.261143 H. sclerophylla — NSW Broulee gbs004985 ZYI049-17 — KY951375 — K.261145 H. sclerophylla — NSW Broulee gbs004986 ZYI050-17 MF040945 KY951395 — K.261150 H. sclerophylla — NSW Broulee gbs004987 ZYI051-17 MF040944 KY951394 — K.261152 H. sclerophylla — NSW Broulee gbs004989 ZYI052-17 MK185703 KY951393 — K.261158 H. sclerophylla — NSW Wellington gbs004856 ZYI036-17 MF040947 — — K.261160 H. sclerophylla — NSW Wellington gbs004853 ZYI034-17 MF040925 KY951377 — K.261162 H. sclerophylla — NSW Wellington gbs004850 ZYI032-17 MF040926 KY951379 — K.261164 H. sclerophylla — NSW Wellington gbs004851 ZYI033-17 — KY951378 — K.261166 H. sclerophylla — NSW Megalong gbs004820 ZYI027-17 MF040930 KY951382 — K.261168 H. sclerophylla — NSW Megalong gbs004823 ZYI028-17 MF040929 — — K.261170 H. sclerophylla — NSW Megalong gbs004825 ZYI029-17 MF040928 — — K.261172 H. sclerophylla — NSW Megalong gbs004826 ZYI030-17 — KY951381 — K.261174 H. sclerophylla — NSW Megalong gbs004831 ZYI031-17 MF040927 KY951380 — K.261184 H. sclerophylla — NSW Jibbon gbs004812 ZYI026-17 MF040931 — — K.261194 H. sclerophylla — NSW Lower Portland gbs001166 ZYI005-17 MF040940 — — K.261196 H. sclerophylla — NSW North Nowra gbs004241 ZYI054-17 — — KY951366 K.261200 H. sclerophylla — NSW Bucketty gbs004299 ZYI020-17 MF040932 KY951383 — K.261234 H. sclerophylla — NSW North Nowra gbs004253 ZYI019-17 MF040933 KY951384 — K.261240 H. sclerophylla — NSW Nattai gbs003090 ZYI013-17 — KY951387 — K.261246 H. sclerophylla — NSW Wellington gbs004855 ZYI035-17 — KY951376 — K.377577 H. sclerophylla — NSW Lower Portland gbs001178 ZYI006-17 MF040939 — — K.377580 H. sclerophylla — NSW Glenbrook gbs001125 ZYI004-17 MF040941 KY951391 — K.377581 H. sclerophylla — NSW Glenbrook gbs003060 ZYI012-17 MF040935 KY951388 — K.377583 H. sclerophylla — NSW Megalong gbs001409 ZYI009-17 MF040938 — — K.377592 H. sclerophylla — NSW Burralow gbs001730 ZYI010-17 MF040937 KY951390 — K.377593 H. sclerophylla — NSW Burralow gbs001733 ZYI011-17 MF040936 KY951389 — K.377738 H. sclerophylla — NSW Megalong gbs004841 ZYI057-17 — — N/A [<200 nt] K.377751 H. sclerophylla — NSW Megalong gbs004839 ZYI056-17 — — N/A [<200 nt] K.377772 H. sclerophylla — NSW Bents Basin gbs004782 ZYI025-17 — MK185708 — K.377749 H. sp. nr. cooloola — QLD Undara gbs003828 ZYI018-17 MF040934 KY951385 — K.377755 H. sp. nr. cooloola — QLD Chillagoe gbs001266 ZYI007-17 MF040950 — — K.377756 H. sp. nr. cooloola — QLD Chillagoe gbs001277 ZYI008-17 MF040949 — — K.377757 H. sp. nr. cooloola — QLD Bamaga gbs003755 ZYI058-17 — — KY951364 K.377759 H. sp. nr. cooloola — QLD Punsand gbs003750 ZYI017-17 MF040948 KY951397 — N/A H. sp. nr. cooloola — QLD Mingella gbs001322 ZYI053-17 — KY951398 — K.261244 H. buntonorum — TAS Knocklofty gbs004929 ZYI047-17 MF040923 KY951373 — K.377604 H. highlandi paratype NSW Wee Jasper gbs001119 ZYI003-17 MF040924 KY951374 — N/A H. sp. nr. highlandi — NSW Red Hill gbs004492 ZYI055-17 — — KY951365 gbs001275 Ctenolepisma longicaudata outgroup QLD Chillagoe gbs001275 ZYI064-18 MK185702 — — K.377675 Ctenolepisma longicaudata outgroup TAS Hobart gbs001836 ZYI065-18 — MK185707 — K.377609 Acrotelsella erniei outgroup, holotype NT Mt Sonder gbs001438 ZYI066-18 MK185701 — — K.261103 Acrotelsella parlevar outgroup, holotype TAS Travellers Rest gbs004624 ZYI067-18 — MK185706 — QM228755 Qantelsella louisae outgroup, holotype QLD Bladensburg gbs003917 ZYI068-18 MK185705 MK185709 — N/A Tricholepidion gertschi outgroup — — — — — — AY191994 6 Records of the Australian Museum (2019) Vol. 71 There is a large amount of variability in almost all these characters. Sometimes the variability is between individuals and other times from the left to the right side of the same specimen. The number and thickness of macrochaetae seems to be greater on larger specimens and the pigmentation seems to be stronger. Some populations were found to have more individuals with a certain character, e.g., a glabrous posterior medial region on the mesosternum but not in all cases and unless a character proved consistent for all individuals examined within one lineage versus another then it was concluded that the character did not qualify as diagnostic. Many or even most macrochaetae have been lost on dissected specimens and chaetotaxy comparisons are based largely on insertion points. It is possible that significant differences in the size and form of the bristles exist that could not be seen when comparing the slide material. For example, the apex of the more lateral macrochaeta of the posterior combs of the metanotum in a specimen from North Nowra (K.261196) was very thin and tapering, almost trichobothria-like, while that of a specimen from Lower Portland (K.261050) was delicately bifurcate and another from Broulee (K.261218) was thicker at the base than K.261196 but also simple apically. More specimens, in better conditions (perhaps collected live and held until after the following moult) may provide additional useful characters. To explore the morphometric data, we carried out two Principal Components Analyses (PCA). In the first analysis (PCA1), we excluded juvenile specimens as well as specimens and morphological characters for which the data was incomplete. This resulted in a dataset consisting of 40 specimens (from seven lineages) and 36 morphological characters. The second analysis (PCA2) was performed on 21 adult female specimens, and included 15 female-only characters relating to the ovipositor morphology. A separate analysis was conducted on the gap between the combs versus the average length of the combs using only the available slide mounted material. PCA was conducted in R v3.3.3, (R Core Team, 2017), and visualized using the R package “ggbiplot” (Vu, 2011). Data was log transformed before analysis, means were set to zero and variance was scaled between 0 and 1. As the dataset for the metathoracic combs was found to be non-normally distributed (Shapiro-Wilk test, W-statistic = 0.79283, p = 4.061 x 10-7), and without homogeneity of variance (Bartlett’s test, Bartlett’s K2 = 16.255, df = 6, p = 0.01245), the Kruskal-Wallis test was used. The Kruskal-Wallis test returned a significant result (χ2= 35.609, df = 6, p-value = 3.284 x 10-6). The post-hoc Dunn Test without corrections (Dunn, 1964) was conducted in R using the package dunn. test v1.3.5 (Dinno, 2017).

The species Heterolepisma sclerophylla Smith was described from specimens collected at Broulee on the NSW south coast (Smith, 2014). Other localities listed included specimens from Glenbrook, Guerilla Bay, Lower Portland, Megalong Valley, Burralow (near Bilpin), Clandella State Forest, Hawkesbury Heights, Ku-ring-ai, Nattai and Wellington, all in NSW (Fig. 1). The species was distinguished from other Heterolepisma by a combination of features including a glabrous urosternite I, 1+1 single macrochaetae on urosternites II-VIII, styli on urosternites VII-IX in the ♀ and VIII-IX in the ♂, 2+2 combs on urotergite I, posterior combs on all nota, urotergal submedial combs each of two macrochaetae, a small apically-round, triangular prothoracic sternum which is longitudinally somewhat furrowed, and a rounded (rather than truncate) urotergite X. Variations in morphology between individuals (and even from left to right on a single individual) were mentioned in the original description but considered as "normal" given the continuous moulting of silverfish throughout their life, even after attaining sexual maturity. The large and "unlimited" number of moults during the life of silverfish results in considerable variation between specimens and makes morphological comparisons difficult.
Material with the same characters listed above has since been collected from a wide range of locations in Queensland, including Carlo Point and Cooloola in South East Queensland, Undara, Chillagoe and near Bamaga at the extreme northern tip of eastern Australia. Material was also obtained from further localities in New South Wales (Fig.  1). The material from Glen Davis was found to resemble H. sclerophylla in most aspects but had one fewer pairs of styli in both sexes. The number of pairs of styli has been considered variable in this genus and some species, notably Heterolepisma stilivarians Silvestri, from Western Australia, have been described as having variable numbers of styli (Silvestri, 1908). All the NSW and QLD material had been identified by the first author as belonging to, or being close to, H. sclerophylla, however preliminary molecular studies by the second author found distances of around 8% in DNA barcode sequences (the 5'-half of the COI gene) between populations from the type locality and the Blue Mountains (Megalong and Lower Portland) and even greater divergence from the Cooloola, southern Queensland, material (circa 14%). These large distances are similar to observed distances between the described species H. highlandi Smith, 2014 and H. buntonorum Smith, 2016, which raised the question of whether H. sclerophylla contains cryptic species, and prompted further study. The fourth author also performed preliminary analyses of the mitochondrial 16S rDNA gene for the population of "sclerophylla" from Cooloola and individual specimens from North Nowra and Bamaga, with similar results.
Further sequencing of both DNA barcodes and a nuclear gene, the D9-D10 region of the 28S ribosomal DNA, was conducted on several specimens each from Broulee, Megalong, Wellington and Cooloola/Carlo Point and a few specimens from other named localities. Detailed morphological and morphometric comparisons of several specimens from each locality were conducted. Some characters, such as the presence of lanceolate scales and the number of divisions in the ovipositor cannot be easily determined in alcohol material, so much of the material was dissected and mounted before examination.

Abbreviations and definitions
Roman numerals are used to indicate abdominal segment number. In addition, the following abbreviations are used: AM: Australian Museum, 1 William St, Sydney 2010 Australia; asl: above sea level (in metres); HW: head width (in millimetres); H+B: head and body length (in millimetres); L/W: length to width (ratio); NSW: New South Wales; PI, PII, PIII: legs of prothorax, mesothorax and metathorax respectively; PCA: principal component analysis; penult: penultimate, referring to second last article of maxillary palp; QLD: Queensland; QM: Queensland Museum, Grey St & Melbourne St, South Brisbane QLD 4101; ult: ultimate (referring to last article). The prefixes pro, meso and meta are affixed to thoracic characters such as sterna, tibia and tarsus.
Specimens are currently stored in 80% ethanol/water unless specifically mentioned as being in 100% ethanol or else mounted on slides using Tendeiro medium. The term macrochaetae refers to the larger stronger bristles (always apically bifurcated), setae to smaller thinner bristles (bifurcated or simple), setulae to the very small, usually straight, setae associated with the combs and cilia to the curly thin hairs, often associated with the combs, setal collar or notal margins.

Collection of material, locality data and preparation
Details of the material examined are included with the description of each species. Locality co-ordinates were obtained using a hand-held Garmin eTrex®10 GPS with a claimed accuracy usually under 5 m. Dissected specimens have been mounted on two slides using Tendeiro medium (one slide with head and thorax, the other with the abdomen). Drawings were made with the aid of an Olympus CX31 binocular microscope fitted with a U-DA drawing attachment.
The holotype and a paratype of Heterolepisma cooloola are deposited with the Queensland Museum in Brisbane. Most of the remaining material is deposited with the Australian Museum in Sydney and each slide or specimen has an accession number starting with K (e.g., K.377269).
Some material is still held by the first author, refrigerated in 100% ethanol. This material carries the author specimen data base number (e.g., gbs004932) and will eventually also be deposited with the Australian Museum. Table 1 lists the 68 specimens subjected to DNA analysis, their collection localities, species identification, and BOLD and GenBank accession numbers for their DNA sequences. Species from three genera of Ctenolepismatinae were used as outgroups: Ctenolepisma longicaudata Escherich, Qantelsella louisae Smith, Acrotelsella erniei Smith and Acrotelsella parvelar Smith. DNA extractions performed at the AM (for DNA barcode/ COI and 28S rDNA sequences) used either the Bio Basic EZ-10 96 well plate Genomic DNA Isolation Kit (Astral Scientific, Taren Point, NSW) or the Bioline Isolate II Genomic DNA Kit (Bioline, Eveleigh, NSW) following the manufacturers' protocols, with exceptions noted below. DNA extractions performed at Marist College, New York (for 16S rDNA sequences) used a Qiagen DNeasy Blood and Tissue Kit (Qiagen, Germantown, MD).

Sampling, DNA extraction, PCR and DNA sequencing
About five whole specimens from each key locality (Broulee, Megalong, Wellington and Cooloola/Carlo Point), collected directly into 100% ethanol, were soaked in DNA extraction buffer containing proteinase-K at room temperature for three hours. The remaining cuticle was returned to 100% ethanol and later dissected in 80% ethanol and mounted on to slides using Tendeiro medium (usually head and thorax on one slide, abdomen on the other).
DNA for specimens from most other localities was extracted from legs which had been removed from the specimen shortly after collection and stored in 100% ethanol at 4°C. The rest of the specimen was kept in 80% ethanol at room temperature.
Polymerase Chain Reaction (PCR) amplification of the DNA barcode region of the mitochondrial COI gene used the primers and followed the method of Mitchell (2015). For the 28S rDNA D9-D10 region, we used one forward (28S_8fm) and two reverse PCR primers (28S_10rm and 28S_11rm), which were simply 5'-M13-tailed versions of Machida and Knowlton's (2012) primers [28S] #8, [28S] #10_RC and [28S] #11_RC, respectively. PCR conditions for both genes followed those reported in Mitchell (2015) for COI. PCR products were purified using ExoSAP and sequenced in both directions using ABI Big Dye Terminator v.3.1 chemistry by Macrogen Inc. (Seoul, South Korea).
The 16S rDNA gene was amplified using primers 16Sar and 16Sb (Edgecombe et al. 2002). Amplification, PCR purification, sequencing, and chromatogram analyses were carried out as in Espinasa and Cappuccio (2008). One 16S sequence was downloaded from GenBank for Tricholepidion gertschi to serve as an outgroup.

DNA sequence assembly and phylogenetic analysis
Forward and reverse direction sequence trace files were assembled using Geneious v. 9.1.7 (Kearse et al., 2012) and consensus sequences were aligned using Muscle (Edgar, 2004) and adjusted by eye. DNA sequences, sequence trace files, and specimen collection data were uploaded to BOLD (Ratnasingham and Hebert, 2007) and GenBank. Two 16S sequences (samples K.377738 and K.377751) were uploaded to BOLD, but were not accepted by GenBank as they are less than 200 bp in length.
Four data sets were constructed: one for each of the three genes alone, COI, 28S, and 16S, and a concatenated gene data set. The COI alignment was trimmed from the 3'-end to 559 nucleotides to minimize missing data before phylogenetic analysis. The concatenated gene data set comprised only samples for which both 28S and COI sequences has been obtained. For outgroups C. longicaudata and Acrotelsella spp. the 28S and COI sequences were generated from different specimens (and for Acrotelsella from different species) thus the sequences are labelled with "Chimera" rather than with a Museum accession number in Fig. 4, although Table 1 lists the accession numbers of the specimens used for each gene. We did not include 16S sequences in the concatenated data set because there was only one specimen for which we had both 16S and other sequence data.
FABOX v. 1.4.1 (Villesen, 2007) was used to edit sequence names. MEGA v.7.0.26 (Kumar et al., 2016) was used to calculate genetic distances. For phylogenetic analysis, nucleotide substitution models, and data partitions (codon positions) for COI, were tested using PartitionFinder2 (Lanfear et al., 2012) on the CIPRES computing platform (Miller et al., 2010), with the most appropriate model selected using the Bayesian Information Criterion. For 16S the best model was GTR+G, and for 28S the best model was K80+G. For COI the data had two partitions, codon positions 1 and 2 (best model TRNEF+I+G) and codon position 3 (best model HKY+G). For the concatenated data set PartitionFinder selected the same two partitions and models as COI, with 28S included in the partition with COI codon positions 1 and 2.
Exploratory phylogenetic analyses were performed in Fasttree 2 (Price et al., 2010) while final analyses were performed using Maximum Likelihood (ML) as implemented in RAxML 7.2.8 (Stamatakis, 2006) and Bayesian Inference (BI) in MrBayes v.3.2.6 (Ronquist et al., 2012) using the plugins available in Geneious. RAxML analyses used the ML search convergence criterion and performed 1,000 fast bootstrap replicates. MrBayes analyses used four heated chains with chain temperature = 0.2, a chain length of 2,000,000 generations, subsampling every 1,000 generations, and a burnin of 500 samples (25%). Convergence was assessed by examining the average standard deviation of splits frequencies, with values below 0.01 indicating convergence. Convergence was reached in two million generations for the single gene analyses and in one million generations for the concatenated genes data set.

Morphological and morphometric evaluation
Specimens were measured in alcohol according to the methodology in . Due to the continuous moulting, most measurements were compared as ratios, either length to width, or compared to head width. While some immature specimens were measured, especially when fewer "adult" specimens were available, the measurement data is intentionally biased towards the largest available sexually mature individuals. Consistent measurement data is difficult to generate, not only due to the large number of instars but it is also difficult to align the part to be measured completely horizontal. Some specimens are bent over and cannot be flattened for accurate measurement. Being softbodied the specimen can be distorted due to damage or preservation, with the various segments either contracted into each other or distended hence the unreliability of H+B, thorax and abdomen lengths. In addition, the nota can sometimes have lost their curvature and hence appear wider than expected probably because the specimen was partially squashed during collection. Ranges of measurement data are often quite wide and it is therefore difficult to find parameters where there is no overlap in the range. Some measurements, e.g., L/W of inner processes of coxites IX, L/W of thoracic sterna and L/W of urotergite X, can only be made using dissected material. The number of divisions in each of the ovipositor valves must also be made with dissected material.
The following measurement ratios were compared: thorax length/HW, thorax length/H+B, thorax length/width (widest part of mesonotum), abdomen length/H+B, width/HW of all nota, thorax length/abdomen length, H+B/HW, L/W of pedicel, L/W of the scape, the length of the ultimate article of the maxillary palp relative to its width, to head width and the length of the penultimate article, the length of both pedicel and scape relative to each other and to head width, length of tarsi of all legs/HW, length of tarsus /tibia of all legs, L/W of ultimate article of maxillary palp, length of ultimate/ penultimate articles of maxillary palp, L/W ultimate article of labial palp, length of pedicel/scape, L/W of thoracic sterna, their width relative to head width, their length relative to each other, the width of the gap between the combs on the metathoracic sternite relative to the average length of the combs, the length of all tarsi and tibiae relative to both their width and HW, the length of the tibia and tarsi of PI relative to PIII, the L/W of urotergite X and the inner and outer processes of coxites IX (the latter independently for males and females), length stylus VII/IX, length of stylus VIII/ IX and the length of the ovipositor relative to head width.
Morphological characters compared include the presence and distribution of lanceolate scales, the chaetotaxy of the frons, clypeus and labrum, the relative size and chaetotaxy of the pedicel and scape, the chaetotaxy of the outer face of the mandibles, the length of molar region and the chaetotaxy at the end of molar region, the number of teeth on lacinia and its length relative to galea, the number of lamellae and setae on the lacinia, the chaetotaxy of the articles of the maxillary palp, the relative length of the most distal articles and the presence, type and location of the branched papillae on the ultimate article of both the males and females, the chaetotaxy of the submentum and mentum, the chaetotaxy of the labial palp and the shape of the last article and the arrangement of its papillae, the density of anterior collar of the pronotum, location of trichobothrial areas and their association with macrochaetae, the lateral and posterior chaetotaxy of all nota, the chaetotaxy of the presternum, the shape, size and chaetotaxy of all thoracic sterna, the scale covering and chaetotaxy of PI and PIII, the chaetotaxy of the urotergites and urosternites (position of combs as well as the number of macrochaetae, marginal setae, setulae and cilia), the shape and chaetotaxy of urotergite X, the number and size of the styli, the number of divisions of the ovipositor, the size, position and chaetotaxy of the paramera and the distribution of pigment. There is a large amount of variability in almost all these characters. Sometimes the variability is between individuals and other times from the left to the right side of the same specimen. The number and thickness of macrochaetae seems to be greater on larger specimens and the pigmentation seems to be stronger. Some populations were found to have more individuals with a certain character, e.g., a glabrous posterior medial region on the mesosternum but not in all cases and unless a character proved consistent for all individuals examined within one lineage versus another then it was concluded that the character did not qualify as diagnostic. Many or even most macrochaetae have been lost on dissected specimens and chaetotaxy comparisons are based largely on insertion points. It is possible that significant differences in the size and form of the bristles exist that could not be seen when comparing the slide material. For example, the apex of the more lateral macrochaeta of the posterior combs of the metanotum in a specimen from North Nowra (K.261196) was very thin and tapering, almost trichobothria-like, while that of a specimen from Lower Portland (K.261050) was delicately bifurcate and another from Broulee (K.261218) was thicker at the base than K.261196 but also simple apically. More specimens, in better conditions (perhaps collected live and held until after the following moult) may provide additional useful characters.
To explore the morphometric data, we carried out two Principal Components Analyses (PCA). In the first analysis (PCA1), we excluded juvenile specimens as well as specimens and morphological characters for which the data was incomplete. This resulted in a dataset consisting of 40 specimens (from seven lineages) and 36 morphological characters. The second analysis (PCA2) was performed on 21 adult female specimens, and included 15 female-only characters relating to the ovipositor morphology. A separate analysis was conducted on the gap between the combs versus the average length of the combs using only the available slide mounted material.

Molecular data
Forty-two 28S sequences were obtained, with 32 being derived from primer pair 28S_8fm/28S_11rm and comprising 442 bp aligned, while the remaining 10 sequences were obtained with primer pair 28S_8fm/28S_10rm and comprised 230 bp aligned. For 28S the mean base compositions for A, C, G and T were 25.6%, 23.6%, 27.8% and 23.0%, respectively, or 51.4% GC.
Twelve 16S sequences were obtained and the alignment had a length of 503 bp. Figures 2-4 show the ML trees obtained for the mitochondrial genes (COI and 16S rRNA), for the nuclear gene 28S rRNA, and for the concatenated 28S and COI genes, respectively. BI Posterior Probabilities (PP) are shown above branches if greater than 0.90 and ML bootstrap percentages (BP) are shown below branches if greater than 70%, while paratypes are indicated with an asterisk and holotypes with double asterisks.
All BI analyses converged. ML and BI analyses gave almost identical results, the only differences being that that placement of some taxa (e.g., sample K.377759 from Punsand, see below, and K.377604 H. highlandi) were strongly supported by BI but had no support (BP <50) under ML.
For COI the BI tree ( Fig. 2) did not recover the ingroup (the H. sclerophylla group) as monophyletic. Instead, the H. sclerophylla group was split into Queensland and NSW lineages, although the single Punsand (Cape York) specimen was placed as sister-group to the NSW lineage. Heterolepisma highlandi is placed between the two H. sclerophylla group lineages while H. buntonorum is basal to them, and the two taxa are separated by an uncorrected distance of 10.8%.
Within the Queensland lineage, the eight samples from Cooloola and Carlo Point in South East Queensland formed a tightly clustered clade with 94% bootstrap support, although each sequence was unique. Seven of eight sequences showed 0.4-1.8% uncorrected distances among them, while the eighth sequence was somewhat separated at a distance of 1.8-4.0% from the others. Specimens from Chillagoe (n = 2), Undara (n = 1) and Punsand (n = 1) showed inter-population distances of 4.0-9.4%.
Within the NSW group six distinct clades were apparent, each with low intra-clade distances compared with inter-group distances, and with very strong bootstrap support (>90%) in all but a single case, where bootstrap support was still strong (74%). Clade 1 comprised a single sample from North Nowra, approximately 110 km north of the type locality, Broulee, on the NSW south coast. Clade 2 comprised five samples from Broulee, one of which is a paratype, one from the nearby locality of Guerilla Bay, as well as one specimen from Jibbon, another coastal locality 217 km north of Broulee. The maximum intra-clade DNA distance was 5.3%, which was the same as the nearest neighbour distance for a specimen from Lower Portland. Clade 3 comprised four specimens from Glen Davis with maximum intra-clade distance of 0.8% and a minimum nearest neighbour distance of 7.2%. Clade 4 comprised three samples from Wellington with a maximum intra-clade distance of 1.3% and a minimum nearest neighbour distance of 5.4%. Clade 5 comprised five samples from Burralow, Glenbrook and Bucketty, with a maximum intra-clade distance of 4.9% (for the specimen from Bucketty) and a minimum nearest neighbour distance of 4.0%. Clade 6 comprised seven samples, five from Megalong Valley and two from Lower Portland, with a maximum intra-clade distance of 4.0% and a minimum nearest neighbour distance of 3.6%.   The BI tree for the limited 16S data (Fig. 2) also showed the southern Queensland population to be distinct from the far North Queensland sample and separation of these taxa from the two NSW populations sampled, North Nowra and Megalong.
The 28S rDNA data showed a broadly similar BI tree topology ( Fig. 3) with 6.4% distance between H. buntonorum and H. highlandi, a deep split between all Queensland (including Punsand) and NSW populations, deep splits among Queensland populations with the Cooloola/Carlo Point population distinct (nearest neighbour distance 10.8%). However, much less variation was observed among NSW populations. Considering the six lineages defined by the DNA barcode data, very little intra-lineage DNA variation was observed, except for the sample from Bucketty, which was 0.5% different from Glenbrook, Burralow and Nattai samples. This lineage could be distinguished from the Megalong population (0.5% distance). The North Nowra sample was 0.9-1.8% distant from other NSW H. sclerophylla group samples, however it was placed within a lineage containing both Broulee and Wellington samples. 28S sequences from the latter two localities are identical, except that specimen K.261164 from Wellington is longer at 442 nt versus 230 nt because the longer amplicon failed to amplify in most of these specimens.
The concatenated COI and 28S dataset yields a tree in which deeper relationships in the ingroup follow the 28S topology and shallower relationships, e.g., with the H.      (4) 3.54-4.24 3.62 (7) 3.14-4.06 Tibia PI/PIII

Morphology
Morphological examination identified several strong characters differentiating the southern Queensland population from those in NSW including the absence of medial anterior chaetotaxy of the frons, the presence of lanceolate scales on the clypeus and legs, a macrochaeta mediad of the anterior trichobothrial area on the pronotum,  Table 4. Loading matrix for principal components 1-8, based on 36 morphological characters in PCA1. Only principal components accounting for more than 5% of the variation are included. Strongest correlations are highlighted in the darkest grey.
a reduction in size of the medioposterior macrochaeta of the submedial dorsal combs and an elongated tapered urotergite X and it is described below as new. While the available molecular data is limited, what is available for the three genes found specimens from several localities in far northern Queensland (Bamaga, Punsand Bay, Chillagoe, Undara and Mingela) to be related to but distinct from the southern Queensland group. Preliminary morphological examination of some of this northern material confirms it shares most of the traits that distinguished the southern Queensland material from the NSW specimens, but from the degree of genetic difference it is likely that these northern Queensland specimens could represent new species. The investigation of this lies outside the scope of the current work. The group from Glen Davis was easily separated from all other groups as both males and females had one fewer pair of styli. No other consistent morphological differences could be found. The Glen Davis material is described below as a new species.
Consistent morphological differences between the remaining five New South Wales groups were not found although some over-lapping differences were noted. Suspected morphological differences and difficulties in interpretation are discussed under each lineage. In the absence of consistent, quantifiable differences, we have chosen to keep them within the H. sclerophylla species group at present, recognizing them as distinct lineages. They may indeed represent separate species, however this would require a lot more molecular and morphological work including further samples from additional localities.

Morphometrics
A summary of all measurement data is found in Table 2.

PCA1.
Our PCA reduced the 36 variables to eight principal components (PCs) that each accounted for at least 5% of the variation (Tables 3 and 4). Together they explained 76.00% of the variation.
Discrimination between lineages is most clear when plotting PC1 against PC2 (Fig. 5). Here, the Cooloola group was the only lineage that could be distinguished from the others along the PC1 axis, which is most strongly correlated with prothorax width/head width, thorax length/thorax width, and PI tarsus/PI tibia. Lineages other than the Cooloola group were not clearly identified.

PCA2.
Our PCA reduced the 15 variables to two PCs that each accounted for at least 5% of the variation: PC1, (73.51%) and PC2, (7.81%). Together they explained 81.32% of the variation. Discrimination between lineages was not very clear, but the Broulee specimens were shifted relative to the other specimens along the PC2 axis, which is most strongly correlated with ultimate maxillary palp article length, abdomen length, PI tibia length and ovipositor length. This may indicate that these characters are most useful in delineating this lineage from others.
Significant differences were found for the gap between the metathoracic combs between the Cooloola group and all other groups, with the only other pair showing significant differences being Wellington and Glenbrook (Table 6).

Heterolepisma highlandi Smith 2014
Heterolepisma highlandi Smith, 2014: 16. Type material (paratype). 1  Comments. The Broulee group has a comparatively short ovipositor both relative to head width  and in the number of divisions (32-37). Most other lineages have ovipositors in mature specimens exceeding twice the head width and with 35-41 divisions. It also seems to have lower levels of pigment overall and the density and strength of the macrochaetae on the head and notal collar is less than for other lineages (Fig. 7). The Nowra lineage has a similarly short ovipositor but appears to have much darker pigment on the labial and maxillary palps and to a lesser extent on the legs. It has denser chaetotaxy along the sides of the head near the eyes (2-4 rows wide versus 2-3 rows wide).
The ovipositor length was reported in Smith (2014) as up to 2.09 times head width. This was based on the result from a specimen (originally recorded as K.377564 in ethanol but now replaced by K.261219 as slide) and was quite different to other specimens measured (range 1.49-1.79). This specimen was re-measured and the measurement is here corrected to 1.95 times HW. This is still much longer than other specimens from Broulee but the specimen only has 35-37 divisions which is within the normal range for the Broulee specimens (32-37). Each individual division is longer than seen with other specimens of this lineage.
The lineage has been collected from three localities along the central and southern coasts of New South Wales. All localities listed have sandy soils, but this may represent a sampling bias rather than be a true indication of habitat. In the type locality it is quite common among dry Eucalypt leaves caught between the fronds of cycads (Macrozamia sp. Zamiaceae). At Guerilla Bay it was collected within abandoned termite galleries on a standing tree and at Jibbon it was collected from dry leaf litter protected from rain beneath a fallen tree or in a large hollow at the base of the tree. It was occasionally found hiding within the bark of certain spongy-barked Eucalyptus or Corymbia trees using pyrethrum sprays, where it was usually collected together with a more abundant (but not yet described) species related to Heterolepisma highlandi Smith.
One Broulee specimen (K.261145) was unusual in that one of its maxillary palps was shorter than usual and had several plumose sensilla along its length rather than the usual three, as found on the other palp where they were more evenly spaced along the article rather than in the distal half. In all specimens examined the plumose sensilla on the palps of the males are much larger (2-3 times larger diameter) and more elaborate than those on the female.

North Nowra Lineage
Material examined. 1♀  Comments. The North Nowra lineage appears closest to the Broulee lineage in terms of ovipositor length. In the available slide material it stands out for having very darkly pigmented articles on the labial and maxillary palps (especially the penultimate and preceding articles). The lateral nota and more distal leg articles are also more heavily pigmented. Some caution needs to be exercised here because pigment seems to diminish in both alcohol and slide mounted material over time and there is quite a bit of difference in pigment between the two specimens examined, especially in PI.
Pigment is also generally weak in juvenile specimens making it a difficult character to use.
The Nowra lineage also appears to have denser macrochaetae around the sides of the head, towards the eyes and more on the face of the mandible (around 60-70 vs about 50 in the Broulee material) however the Glenbrook strain also appears to have a similar number of macrochaetae on the mandibles. Counting macrochaetae on the mandibles is also fairly subjective; when does one decide an insertion is that of a macrochaeta or a seta? The meso-and metathoracic sterna may also be more elongate (L/W 1.17 versus 1.00-1.13 and 0.77-0.82 versus 0.71-0.78 respectively).
On the tibia of PI most H. sclerophylla sensu lato have three dorsal carrot-shaped macrochaetae whose position along the margin can be quite variable and three carrotshaped macrochaetae near the ventral margin with a stronger seta proximally that sometimes approaches carrot-shaped. Of the three PI legs available of the Nowra lineage, there only appears to be two macrochaetae on the dorsal surface and four quite solid macrochaetae on the ventral margin. Insertion points on the dorsal surface can be hard to see and the number of specimens insufficient. Urotergite IX may have more strongly developed and numerous setae in the infralateral corners. Urotergite X is also on average longer than the Broulee lineage (L/W 0.57-0.73 vs 0.49-0.59) and the inner processes of coxites IX in the male may be shorter (L/W 0.86-1.14 versus 1.26-1.44.
It may well be possible to describe this lineage as a morphologically distinct species using these characters. However, given the small number of North Nowra specimens examined, from just a single location and the limited molecular data, we have taken a conservative approach and will consider this as a lineage of H. sclerophylla until more data suggest otherwise.

Glenbrook Lineage
Material examined. 1♂ (HW 1.00) (AMS K.377580 in ethanol) Comments. This lineage, as well as the next two lineages, have a longer ovipositor than that of the more coastal Broulee and Nowra lineages although there is some overlap both in length relative to HW (1.58-2.39 versus 1.49-1.95 HW) and also in the number of divisions (35-40 versus 32-37). No correlation could be found between number of divisions and HW in clearly adult specimens including one specimen, from Nattai which was particularly large (HW 1.50 H+B over 11mm) but had only 36 divisions in the ovipositor. This suggests that divisions are not added with increasing size.
Apart from the ovipositor length, no reliable diagnostic character has been identified. The Glenbrook, Nattai and Burrawang specimens also seemed to have more macrochaetae on the external face of the mandibles (around 60 versus around 50) as well as more setae in the lateral group near the molar area (12-16 versus around 11) but this was not the case with the Bucketty specimens, which also differed in generally having more pigment in the palps as well as at least the tarsi and tibia of PII and PIII.
All of the limited number of specimens examined from Glenbrook, Nattai and Burrawang differed from most specimens from other populations in having just three macrochaetae located on the posterior bulge of the femur of PI, instead of the usual four. Specimens from Bucketty however had the more usual four macrochaetae and two of the Broulee specimens also had only three macrochaetae so this character does not appear to be sufficiently reliable. The chaetotaxy of the legs, while having a basic pattern, differed quite considerably between specimens, in terms of the number and size of stout macrochaetae and their position.
Specimens from all four localities sometimes had five macrochaetae in the lateral combs of one or two urotergites whereas other lineages had a maximum of four, but most urotergites had the usual 3-4 macrochaetae.
The posterior margin of urotergite X in all dissected Bucketty specimens also has a wider glabrous area medially than seen in all other dissected and mounted H. sclerophylla sensu lato specimens.
All specimens of this lineage were collected in leaf litter which had gathered in places protected by rain e.g., beneath the trunk of a fallen tree that was not in contact with the ground.

Comments.
No morphological characters could be found to separate this lineage from the previous. It also has an ovipositor which appears to be longer than the coastal lineages being 1.96-2.45 HW with 36-41 divisions vs 1.49-1.95 HW and with 32-37 divisions for the Broulee lineage. The macrochaetae on the margins of the head and the anterior margin on the pronotum are somewhat denser and thicker than on Broulee specimens, similar to that seen in the Glenbrook lineage and the species from Glen Davis (described below).
The lineage has been collected from the Blue Mountains west of Sydney and their foothills, where sandy soils and rocky sandstone outcrops predominate. It is found in the same types of microhabitat, i.e. Eucalyptus and Casuarina leaf litter lying in places protected from rain or where the leaf litter accumulates but dries out quickly e.g., in the forks of trees or on exposed rocks. One specimen was taken from bark using a pyrethrum spray. Comments. This lineage is similar in appearance to the Megalong and Glenbrook lineages with perhaps an even longer ovipositor (2.65 times HW versus up to 2.45) however data only exist for a single large female so the range may overlap with more material. It has the same number of divisions, but each was more elongate. Elongation of the divisions is probably quite variable between individuals as observed when examining a range of specimens of the Cooloola species described below. The Wellington lineage appears to be narrower in the body but there are insufficient larger specimens available and this parameter needs to be verified as it may represent a sampling bias. The gap between the combs of the metathoracic sternum was also comparatively small relative to the combs (1.55-2.75) but only statistically significantly different from the Glenbrook population (2.16-3.47) with the degree of overlap being unhelpful.
The Wellington habitat is more inland and drier than that of the other two lineages but the specimens were collected in the same typical microhabitat, i.e. dry Callitris, Eucalyptus and Casuarina leaf litter protected from rain, in rocky country with sandy soils.
Heterolepisma coorongooba sp. nov. Diagnosis. This species is very similar to Heterolepisma sclerophylla differing in having one fewer pairs of styli in both the male and female (i.e. IX only in ♂, VIII and IX in the ♀). Compared to the lineage from the type locality it also has a longer ovipositor and thicker and perhaps more densely packed chaetotaxy along the pronotal collar and margins of the head.

Description
Appearance: Medium to large silverfish, scale covering in life uniform or slightly mottled grey with brown antennae, terminal filaments brown with lighter annuli around larger macrochaetae resulting in distinctly banded appearance (Fig. 6).
Body length: H+B up to 9.1 mm (♀) 7.0 mm (♂); maximum HW 1.30 mm; thorax: length up to 2.5 mm (or 0.27-0.36 H+B); width up to 2.05 mm, usually slightly widest at the mesonotum; antennae damaged in all specimens, maximum preserved length of antenna 4.4 mm (or 0.56 H+B); terminal filaments damaged in all specimens, maximum preserved length of cercus 2.8 mm (or 0.31 H+B); maximum preserved length of median dorsal appendage 3.4 mm (or 0.42 H+B). Body neither elongate nor broad with thorax slightly wider than abdominal segment I, the following abdominal segments about the same width until the fourth or fifth after which the abdomen tapers posteriorly.
Pigmentation: Pigment brown in alcohol preserved specimens. Present around eyes and to a lesser extent behind the peri-antennal group of macrochaetae; pedicel and scape very lightly pigmented distally, rest of flagellum uniformly lightly pigmented; all articles of maxillary palp with pigment especially in the four most distal articles although much less in the ultimate article; labium with pigment on the three most distal articles, being strongest on the penultimate, especially distally. Pigment present in anterior corner and along margins of all nota. Legs not heavily pigmented, with light pigment along outer margin of the coxa and the trochanter, apically on the outer femur and strongest on the tibia especially on the dorsal surface, present on basal article of tarsi, urotergite X and coxites IX with light pigmentation; styli IX pigmented in distal three quarters; other styli with less pigment. Ovipositor with yellowish hue. Terminal filaments with rings of pigment, with the pigment present in all annuli except the annuli bearing the large macrochaetae. Pigmentation is much reduced in juvenile specimens. Macrochaetae: Bifid apically, or simple, light to darker brown in colour. Often quite thick e.g., the stronger macrochaetae of the pronotal collar measured about 50 microns in diameter in the holotype (HW 1.30) compared to 38 microns in the similar-sized holotype of H. sclerophylla (HW 1.28) from Broulee (compare Figs 7,8). They also appear to be more densely packed.
Scales: Unevenly rounded or ovoid, with numerous parallel ribs that do not extend beyond the margin (Fig. 9); in alcohol dorsal scales with dark brown ribs; ventrally mostly hyaline. Lanceolate scales not observed. Scales absent from flagellum of antennae, mouthparts and terminal filaments. Head: Wider than long (Fig. 10) with marginal rows about three macrochaetae wide along the sides of the vertex decreasing to two wide in front of the antennae and six strong curved macrochaetae along the anterior margin, the lateral rows extend back above the eyes, as well as a small 1+1 peri-antennal groups not quite isolated from the marginal rows at the level of each antenna. Clypeus with numerous setae, some long and thin, others more robust arranged in 1+1 lines in the proximal lateral regions but not forming combs. Labrum with thin setae only. Eyes dark, composed of about 12 ommatidia.-Antennal scape with a subdistal rosette of setae, very conspicuous from above (Fig. 11), and numerous setae along the sides and over the ventral face; pedicel short, 0.51 times the length of the scape (range 0.47-0.55), with many setae mostly distally and on the ventral face; repeating intervals of distal end of antennae (Fig. 12) of eight annuli, the most apical annulus of each interval (T-annulus) with a trichobothrium and at least one small inconspicuous rod-like basiconic sensillum (type B of Adel, 1984), the second and forth annuli also each with a sausage-shaped type C sensillum (confirmed also to be present on same annuli in holotype of H. sclerophylla).-Mandibles typical for genus with welldeveloped molar and incisor areas; a group of about nine strong setae distally adjacent to the pectinate molar area and a bush of 50+ setae and macrochaetae externally.-Maxilla (Fig. 13) with three large macrochaetae externally proximal to the palp, the lacinia with three strong teeth, one shorter than the rest, seven lamellate processes and a row of eight simple setae, the galea longer than the lacinia with setulae on the outer face. Palp with rosettes of distinctly stronger setae (some "carrot-shaped") subapically on the three basal articles, all articles with numerous fine setae, apical article of maxillary palp 4.6 times longer than wide (range 4.1-4.9) and 1.23 times longer than penultimate article (range 1.16-1.27), the ultimate article in both sexes with three "branched" papillae, those in the female less robust than those in the male.-Labium (Fig. 14) short and broad with rows of strong setae on the prementum and submentum; glossae and paraglossae quite broad with short curved setulae; labial palp short, apical article eccentric suboval, 1.1 times as long as wide (range L/W 0.8-1.3) with 2+3 papillae of compact type in a "cluster formation" where the slightly larger distal papillae curve around the two smaller proximal papillae and a curved club-like thin-walled basiconic sensillum and at least one rod-like basiconic sensillum, which can also be confirmed as present on the holotype of H. sclerophylla.
Thorax: Pronotum (Fig. 15) with strong setal collar of short, apically bifurcated setae and cilia with a row of longer macrochaetae spaced along the back of the collar including 1+1 long thin simple setae about one third in from each side, the density and length of the smaller more marginal macrochaetae of the collar decreases only slightly towards the middle but not the size and density of the larger spaced macrochaetae in the posterior row; lateral margins also with numerous shorter but quite robust, apically bifurcate setae as well as several larger more erect submarginal macrochaetae; trichobothrial areas open and in contact with the lateral margins, the anterior one (Fig. 16) located just anterior to the mid-point along the margin, with or without a large submarginal macrochaeta, when present laterad to the trichobothrium (this macrochaeta missing on the left side of the holotype) and with a cilium and a few setulae; posterior trichobothrial area (Fig. 17) near posterior lateral corner with two submarginal macrochaetae between the trichobothrium and the margin as well as a few setae and setulae; posterior margin slightly concave with 1+1 combs (Fig. 18) each of two macrochaetae, the more postero-mediad lying flatter than the other, each comb associated with a setula and a few cilia.-Mesonotum with lateral chaetotaxy similar to pronotum but less dense (Fig. 19), except the submarginal macrochaetae anterior to the trichobothrial areas are grouped into combs of two, each associated with a cilium and a few setulae, the more posterior of these on the left side of the holotype of only one macrochaeta, both trichobothrial areas (Fig. 20) of similar configuration to those of pronotum except anterior area has a macrochaeta mediad of the trichobothrium. The posterior combs of the mesonotum are unusual in that they both consist of only a single macrochaeta in the illustrated holotype (Fig. 21) whereas in the paratype K.261204 one consists of two macrochaetae and the other of only one and in the other paratypes (K.261213 and K.261210) two macrochaetae are present on both sides.-Metanotum (Figs 22, 23) similar to mesonotum; both posterior combs consist of two equal sized insertion points. The trichobothrium appears to be absent from the right posterior area of the holotype.
Presternum narrow, with transverse row of strong setae and numerous cilia and setulae (Fig. 24).-All thoracic sterna and coxae with hyaline scales. Prothoracic sternum pointed cordiform, almost as long as wide at its base (range L/W 0.91-1.00) and reaching to about two thirds the length of the coxa, rounded apically and with a medial furrow (Fig.  24), most of lateral margins with numerous small marginal setae and cilia, with 6-7 larger submarginal macrochaetae forming weak combs parallel to the edges in the distal third.-Mesosternum (Fig. 25) 1.08 times longer than broad (range 1.01-1.12) with an acutely rounded apex, with setae and cilia along the distal quarter of the margins, with 1+1 distal combs of four apically bifurcate macrochaetae and 1+1 subposterior more pointed macrochaetae.-Metasternum (Fig. 26) 0.79 times longer than wide (range 0.77-0.80) with less pointed, even slightly concave, apex, each comb of four macrochaetae.
Legs fairly long (Figs 24, 26), tibia L/W ratio of legs PI 3.0 (range 2.9-3.2), PII 3.4 (range 2.8-4.0), PIII 3.9 (range 3.5-4.2); tarsi L/W ratio PI 7.0 (range 6.3-8.0), PII 7.0 (range 6.3-7.7), PIII 8.5 (range 7.4-9.7). Legs increasingly longer from front to back, mean ratio PI/PIII (tibia 0.62, tarsus 0.72). PI with transverse comb of about six macrochaetae laterally on the precoxa. Coxa of all legs covered with hyaline scales and with strong macrochaetae and numerous cilia and setulae in a row about two macrochaetae wide along the external margin, a strong seta on the inner margin subapically and group of about six curved setae at the apex over the articulation. Trochanter lacking scales, with fine and one stronger seta over the surface. Femur with numerous setae over most of the surface and along the margin; anterior distal end with two strong, quite deeply bifurcate stout  I  1-3  2  --II  3  2-3  2  1  III  2-4  3 More posterior seta insertion smaller macrochaetae and another simple stout macrochaeta more proximal; posterior margin with several strong macrochaetae as illustrated. Tibia with numerous long setae over the ventral surface, with three stout macrochaetae on or near the anterior margin and six or seven stout macrochaetae along the posterior margin; apical spur with several setae. Tibia of PIII with a long thin, laterally projecting trichobothria-like seta inserted dorsal to the proximal stout macrochaeta on the anterior margin, which is about two times as long as the tibia is wide. Tarsus with four articles, all with numerous setae, some on the ventral surface quite long and strong. Pretarsus with long curved lateral claws and a strong curved shorter medial claw.
Abdomen: Urotergite I usually with 2+2 combs each of one to three macrochaetae (usually two) located quite close together, urotergites II-VII with 3+3 combs of macrochaetae as in Table 7 (Figs 9, 27-29) noting that the more posteromedial insertion point of each submedial comb was occasionally, quite a bit smaller than the other insertion point, but mostly of about the same size; each comb also associated with up to five marginal setae, five setulae and four cilia. Urotergite VIII with 2+2 combs, lacking the sublateral comb; urotergite IX ( Fig. 30) with two infralateral setae on each side as well as a setula and a few cilia. Urotergite X short, parabolic in both sexes (Fig. 31), L/W at base about 0.52 (range 0.51-0.55) with many strong setae along entire margin and obscure 1+1 submarginal macrochaetae in the posterolateral corners. Urosternite I glabrous, urosternites II-VIII (Fig. 32) with 1+1 single macrochaetae (Fig. 33), each associated with 0-2 small marginal setae as well as a few cilia and/or setulae. Coxites of segment VIII in ♀ (Fig. 34) with one or two small macrochaetae, one or two small marginal setae and a cilium mediad of the stylus base and some small setae, setulae and a cilium laterad of the stylus base. Styli in two pairs in the ♀ (VIII-IX); all styli with several noticeably longer and stronger setae apically (Fig. 34) as well as stronger setae along the middle of the ventral face. Styli IX 2.4 times as long as styli VIII (range 2.2-2.8).
Coxite IX of ♀ (Fig. 34), the internal process acute apically, about three times longer than the external process (range 2.9-3.2) and 1.5 times as long as broad at its base (range 1.5-1.6), not reaching to half the length of the stylus; external and internal margins of internal process and external margin and apex of outer process with many moderately strong setae directed both up and down.-Ovipositor (Fig.  34), very long and thin (up to 2.34 HW), surpassing the apex of stylus IX by more than the length of the stylus (excluding terminal macrochaetae), composed of 34-39 divisions. Distal divisions of gonapophyses VIII and IX with only short fine setae and setulae.
Cerci not well preserved in slide material, with basal divisions shorter than long, gradually becoming longer distally, equally wide as long by about the eighth division after which they become even longer with more annuli each with a rosette of setae and some with trichobothria with the large macrochaetae restricted to the most distal annulus of each division; the most distal surviving divisions with up to four annuli, this annulus without pigment.-Medial filament of similar arrangement.
Male: As for female except only one pair of styli (segment IX). Coxites IX (Fig. 35) with acute inner process about 1.3 times longer than wide at its base and about three times longer than the external process, reaching to just under half the length of the stylus. Both process also with several strong setae mostly apically emerging from both the dorsal and ventral surfaces of the processes close to or on the margin. Parameres a little longer than wide, with about thirty fine setae (Fig. 36). Penis typical for genus with numerous glandular setae apically, each set on a protuberance.
Subadult stages: the single juvenile specimen available, K.377726 (HW 0.63) had a single pair of styli (IX) and no indication of urosternite VIII dividing into separate coxites nor any nascent genitalia (ovipositor or parameres). The thoracic sternites conformed to those of the adults and the long thin setae on the tibia of PIII was present and very long (about three times the width of the tibia), tergite X was round but shorter than in the adults and the feathered papilla of the maxillary palp could not be seen.
Habitat. Heterolepisma coorongooba was collected from leaf litter protected from rain under a fallen but still elevated, log.
Etymology. The species name is derived from the proper noun Coorongooba referring the creek that flows through the valley from where it was collected.
Comments. The morphology of this species is very close to that of H. sclerophylla, differing from it only in the absence of the most anterior pair of styli in both sexes. It differs from the Broulee lineage in the length of the ovipositor (2.17-2.34 times HW versus 1.49-1.95), but not so much in the number of divisions (34-39 versus 32-37) and the more robust macrochaetae. The Megalong, North Nowra and Glenbrook lineages also have more robust macrochaetae and a longer ovipositor (but more divisions) leaving the fewer styli as the only unambiguous defining character. previous; 1♂ (HW 0.95) (gbs004902 in 100% ethanol) same data as previous; 1♂ (HW 0.88) (gbs004903 in 100% ethanol) same data as previous; 1 juvenile ♀ (HW 0.83) (gbs004904 in 100% ethanol) same data as previous; 1 juvenile ♂ (HW 0.75) (gbs004905 in 100% ethanol) same data as previous; 1 juvenile ♀ (HW 0.75) (gbs004906 in 100% ethanol) same data as previous; 1 juvenile ♂ (HW 0.70) (gbs004907 in 100% ethanol) same data as previous; 1 juvenile ♂ (HW 0.70) (gbs004908 in 100% ethanol) same data as previous; 1 juvenile ♀ (HW 0.83) (gbs004909 in 100% ethanol) same data as previous; 1 juvenile ♂ (HW 0.68) (gbs004910 in 100% ethanol) same data as previous; 1 juvenile ♂ (HW 0.75) (gbs004911 in 100% ethanol) same data as previous; 1 juvenile ♀ (HW 0.68) (gbs004912 in 100% ethanol) same data as previous; 1 juvenile ♂ Diagnosis. This species differs from other described species of Heterolepisma that also have 2+2 combs on urotergite I and three pairs of styli in the female and only two in the male, by the presence of lanceolate scales on the femora, tibiae and clypeus, the straight anterior margin of the head devoid of macrochaetae, the single macrochaeta mediad of the anterior trichobothrium on the pronotum, the mesosternum is also slightly different with a wider glabrous apex and the combs more compact, the posterior margin of the metasternum is rounded with a comparatively wide glabrous gap with small 1+1 combs of two to three macrochaetae. The parabolic shape of urotergite X is narrower and the terminal filaments evenly pigmented.

Description
Appearance: Medium to large silverfish, scale covering in life uniform or slightly mottled grey with brown antennae, terminal filaments slightly darker than antennae with only a small portion of each annulus bearing the larger macrochaetae lighter in colour (Fig. 37).
Body length: H+B up to 9.9 mm (♀) 8.25 mm (♂); maximum HW 1.20 mm; thorax: length up to 2.85 mm (or 0.26-0.32 H+B); width up to 1.93 mm, usually slightly widest at the mesonotum; antennae damaged in all specimens, maximum preserved length of antenna 5.6 mm (or 0.68 H+B); terminal filaments damaged in all specimens, maximum preserved length of cercus 3.6 mm (or 0.48 H+B); maximum preserved length of median dorsal appendage 5.00 mm (or 0.60 H+B). Body neither elongate nor broad (Fig. 38) with thorax slightly wider than abdominal segment I, the following abdominal segments about the same width until the fourth or fifth after which it tapers posteriorly.
Pigmentation: Pigment light chestnut-brown in alcohol preserved specimens, stronger around peri-antennal and supra-ocular lines of macrochaetae and along the band of setae on the clypeus (especially laterally); pedicel and scape very lightly pigmented distally, rest of flagellum uniformly lightly pigmented becoming somewhat darker distally; all articles of maxillary palp with pigment except the most distal article, densest on article three especially distally; labium with lines of pigment around macrochaetae across the mentum, present on distal three articles of labial palp being stronger on the edges, pigment of ultimate article mostly in basal half but with a noticeable line above the more distal row of papillae. Nota with some pigment anteriorly and along margins. Legs with pigmentation along outer edge of precoxa of PI, along the length of the outer margin among the macrochaetae and only very faintly along the inner margin, distally; trochanter with light patch on margin distally; femur pigmented, darkest distally along dorsal margin and around bulge on ventral margin; tibia pigmented along edges being a little darker distally; first tarsal article pigmented distally. Urotergite X with very faint pigment along anterior lateral margins. Styli IX slightly pigmented in distal three quarters; other styli with very little or no pigment. Ovipositor with yellowish hue. Terminal filaments lightly pigmented basally becoming a lot darker distally. Some individuals show greater or lesser levels of pigmentation, with less pigmentation in juvenile specimens.
Scales: Unevenly rounded or ovoid, with numerous parallel dark brown ribs, that do not extend beyond the margin (Fig.  39); in alcohol dorsal scales and the more lateral scales of the urosternites with dark brown ribs; ventrally mostly hyaline but those towards the lateral margins with light brown ribs. Lanceolate scales present on clypeus, femora and tibia. Scales absent from flagellum of antennae, mouthparts and terminal filaments.
Head: Wider than long ( Fig. 40) with marginal rows about two macrochaetae wide along the sides of the vertex, but without macrochaetae along the anterior margin, the lateral rows extending back along the margin to the eyes and extending as a single short row above the eyes, as well as a small peri-antennal group isolated from the marginal. Clypeus with numerous setae, some long and thin, others more robust but not forming combs; with a few lanceolate scales scattered among the setae. Labrum with thin setae only. Scales on top of head, those along the anterior margin overhanging the margin. Eyes dark, composed of 12 ommatidia.-Antennae long, about ⅔ H+B, scape with a subdistal rosette of setae, very conspicuous from above (Fig. 41), and numerous setae along the sides and over the ventral face (Fig. 42); pedicel short, 0.42 times the length of the scape (range 0.33-0.53), with many setae mostly distally and on the ventral face; the most apical annulus of each interval in the distal end of the flagellum with a few small inconspicuous rod-like basiconic sensilla (type B of Adel, 1984) (Fig. 43).-Mandibles (Figs 44,45) typical for genus with well-developed molar and incisor areas; a group of about nine strong setae distally adjacent to the pectinate molar area and a bush of 50+ setae and macrochaetae externally.-Maxilla (Figs 46-48) with three large macrochaetae externally proximal to the palp, the lacinia with three strong teeth, one shorter than the rest, seven lamellate processes and a row of eight simple setae, the galea slightly longer than the lacinia with setulae on the outer face. Palp with rosettes of somewhat stronger setae subapically on the two basal articles, all articles with numerous fine setae, apical article of maxillary palp short, being only 0.22 times HW (range 0.20-0.25) and 4.5 times longer than wide (range 3.6-6.0) and 1.3 times longer than penultimate article (range 1.2-1.6), the ultimate article in both sexes with three "branched" papillae, those in the female much less robust and with fewer "arms" than those in the male.-Labium (Fig. 49) short and broad with rows of strong setae on the prementum and submentum, glossae and paraglossae quite broad with short curved setulae; labial palp short, apical article eccentric suboval, 1.05 times as long as wide (range L/W 0.9-1.2) with 2+3 papillae of compact type (Fig. 50) in a "cluster formation" where the slightly larger distal papillae curve around the two smaller proximal papillae and at least one curved club-like thin-walled basiconic sensillum (N.B. one palp of holotype does not show usual shape, possibly as a result of damage in the previous instar).
Thorax: Pronotum (Fig. 51) with narrow setal collar of short, apically bifurcated setae and cilia, quite weak in the medial part of the margin; lateral margins ( Fig. 52) also with numerous small to medium sized, apically bifurcate setae as well as several larger more erect submarginal macrochaetae; trichobothrial areas open and in contact with the lateral margins, the anterior one (Fig. 53) located near the mid-point along the margin, almost always with one large macrochaeta located mediad of the trichobothrium (this macrochaeta missing on the left side of the holotype) and with a few cilia and setulae; posterior trichobothrial area (Fig. 54) near posterior lateral corner with a submarginal macrochaeta between the trichobothrium and the margin as well as a few setulae and cilia; posterior margin slightly concave with 1+1 combs each of one macrochaeta with a smaller seta mediad and posterior to it, the insertion of this setae smaller than that of the macrochaeta on most specimens (the smaller insertion missing from the left side of the holotype), the size of the smaller insertion becomes increasingly smaller on posterior segments such that it only appears as a very small submarginal seta on urosternite VIII; these posterior notal combs are associated with two cilia and some setulae.-Mesonotum with lateral chaetotaxy similar to pronotum (Fig. 55), the posterior trichobothrial area (Fig. 56) in the posterolateral corners with a large macrochaeta between the trichobothrium and the margins as well as some marginal and submarginal setae, cilia and setulae; the anterior trichobothrial area (Fig. 57) about ¾ the distance posteriorly along the margin, with a submarginal macrochaeta (m -1 ) between it and the margin, also with a few setulae and cilia; anterior to this trichobothrial area are two combs (m -2 , m -3 ) each of two macrochaetae, a further two or three submarginal macrochaetae more anterior along the margin; posterior combs as for pronotum (Fig. 58).-Metanotum (Fig. 59) similar to mesonotum (the comb at position m -2 on the left side of the holotype composed of only one macrochaeta, suggesting that a degree of variation exists within the species).
Presternum narrow, with transverse row of strong setae and numerous cilia.-All thoracic sterna with hyaline scales. Prothoracic sternum pointed cordiform, only slightly longer than wide at its base (L/W 1.08 range 1.4-1.15) and reaching almost to the end of the coxa, rounded apically and with a medial furrow (Fig. 60), most of lateral margins with numerous small marginal setae and cilia, with 4-6 larger submarginal macrochaetae forming weak combs parallel to the edges in the distal quarter.-Mesosternum (Fig. 61) slightly longer than broad (1.09 range 0.98-1.18) with a truncate or evenly slightly concave posterior margin, with 1+1 distal combs of two to three submarginal macrochaeta associated with some marginal setae and cilia (Fig. 62).-Metasternum (Figs 63, 64) wider than long (L/W 0.75 range 0.69-0.84) but otherwise similar to the mesosternum; the gap between the combs 7.3 times the average width of each comb (range 5.4-11.1).
Legs fairly long (Figs 60, 61, 63), tibia L/W ratio of legs PI 2.6 (range 2.5-3.0), PII 3.0 (range 2.4-3.5, PIII 3.6 (range 3.1-4.1); tarsi L/W ratio PI 6.7 (range 5.4-7.6), PII 6.7 (range 6.0-8.0), PIII 8.8 (range 7.4-10.0). Legs increasingly longer from front to back, mean ratio PI/PIII (tibia 0.58, tarsus 0.68). PI with transverse comb of about six macrochaetae laterally on the precoxa. Coxa of all legs covered with hyaline scales and with strong macrochaetae and numerous cilia in a row about two macrochaetae wide along the external margin, a stout macrochaeta and some long fine setae on the inner margin subapically and group of about four to six stout curved macrochaetae at the apex over the articulation. Trochanter lacking scales, with fine and one stronger seta over the surface. Femur with numerous lanceolate scales on the anterior half and along the margin, rest of surface with fine setae; anterior distal end with three to six strong, quite deeply bifurcate stout macrochaetae as well as some strong setae; posterior margin with several strong macrochaetae as illustrated. Tibia with numerous setae and lanceolate scales over the ventral surface, with two stout macrochaetae on or near the anterior margin and three or four stout macrochaetae along the posterior margin (some paired with thinner macrochaetae on the dorsal side of the margin; apical spur with several setae. Tibia of PIII with a long thin, laterally projecting trichobothria-like seta inserted dorsal to the proximal stout macrochaeta on the anterior margin, which is more than twice as long as the tibia is wide. Tarsus with four articles, all with numerous setae (without lanceolate scales). Pretarsus with long curved lateral claws and a strong curved shorter medial claw (Fig. 65).
Urosternite I glabrous, urosternites II-VIII ( Fig. 74) with 1+1 single macrochaetae (Fig. 75) (although missing from left side of urosternite II on holotype), each associated with 0-1 marginal seta as well as a few cilia and/or setulae. Coxites of segment VII, VIII and IX in ♀ (Fig. 76) with group of several fine setae on the rounded corners on each side of the stylus insertion (Fig. 77). Styli in three pairs in the ♀ (VII-IX); all styli with several noticeably longer and stronger setae apically. Styli IX three times as long as styli VII (range 2.3-3.7) and about two and a half times as long as stylus VIII (range 2.0-2.9) and much more robust (Fig. 78).
Coxite IX of ♀ (Fig. 76), the internal process acute apically, about 4.2 times longer than the external process (range 3.3-6.0) and 1.8 times as long as broad at its base (range 1.6-2.1), reaching almost to half the length of the stylus; external and internal margins of internal process and external margin and apex of outer process with many moderately strong setae directed both up and down.-Ovipositor (Fig. 76)  Cerci (Figs 81,82) with basal divisions shorter than long, gradually becoming longer distally, equally wide as long by about the sixth division after which they become even longer with more annuli each with a rosette of setae and some with trichobothria with the large macrochaetae restricted to the most distal annulus of each division; the most distal surviving divisions with up to eight annuli.-Medial filament of similar arrangement (Figs 81, 83).
Male: As for female except only two pair of styli (segments VII and IX). Coxites IX (Fig. 84) with acute inner process about 1.8 times longer than wide at its base (range 1.70-1.93) and about four times longer than the external process which has a small preapical constriction, reaching to about half the length of the stylus. Both processes with several strong setae mostly apically emerging from both the dorsal and ventral surfaces of the processes close to or on the margin. Parameres small, slightly longer than wide, with about eight fine setae (Fig. 85). Penis typical for genus with numerous glandular setae apically, each set on a protuberance (Fig. 84).
Subadult stages: Very small specimens (HW 0.68) only have styli on coxites IX; the coxites of segment VIII are already clearly divided in a juvenile ♀ of HW 0.53. The ovipositor is just beginning to appear in a ♀ of HW 0.83 and styli VIII are present but not styli VII; a ♀ with HW of 0.88 had an ovipositor that just attained the end of styli IX but still lacked styli VII; by HW 0.93 all styli were present and the ovipositor was much longer than the end of stylus IX. In a ♂ with HW 0.96 one stylus VIII was developed but the other was only represented by a small triangular appendage. Presumably specimens of both sexes could be considered as sexually mature once HW is greater than 0.93-0.96 mm or once all styli are clearly developed.
Habitat. Heterolepisma cooloola was fairly common in the Rainbow Beach area with specimens collected in dry leaf litter accumulating in places largely protected from rainfall such as within burned out ground level tree hollows. It was also taken from the bark of ti-trees, Casuarina and Eucalyptus in heathland and on the edge of rainforest using pyrethrum sprays.
Etymology. The species name is derived from the proper noun Cooloola referring the locality in which it was collected. Comments. Heterolepisma cooloola is in many ways similar to H. sclerophylla (small triangular prothoracic sternum, glabrous urosternite I, urosternites II-VIII with 1+1 macrochaetae, three pairs of styli in the female and two in the male) but in other aspects it shares characters with H. parva Smith from Barrow Island (notably the glabrous anterior margin to the frons, a macrochaeta mediad of the anterior trichobothrium on the pronotum and the presence of lanceolate scales). Heterolepisma parva is however, one of the species with a medial comb on urosternite I and 1+1 combs of several macrochaetae on urosternites II-VII (VIII), a group that Mendes (pers. comm.) has suggested may be a separate group within Heterolepisma. The combined molecular and morphological data support the view that the presence of lanceolate scales and the absence of macrochaetae from the anterior margin of the frons are more significant to phylogeny than the arrangement of styli and the shape of the thoracic sternites in Heterolepisma.

Molecular data
While previous studies have presented COI sequence data for Zygentoma, they have focussed on higher-level systematics. The present study is the first to use DNA barcodes for species delimitation in Zygentoma, and the first to produce barcode-standard compliant data (Hanner, 2009). Espinasa et al. (2007) reported 16 COI sequences from 14 species, all of Nicoletiidae except for two lepismatid specimens (Thermobia domestica Packard) and one tricholepidiid (Tricholepidion gertschi Wygodzinsky) but since then these authors have utilized 16S data for species level systematics, due to its ease of PCR amplification. Other reported Australian studies have used 12S (e.g., Smith et al., 2012). Phylogenetic analyses of DNA barcode data suggest that H. sclerophylla, as currently defined, is an assemblage of several cryptic species, as there are six well-defined barcode clades (lineages), each with >4% divergence in COI sequences and each geographically restricted. Intra-clade divergences are also large, and despite the well-supported phylogeny there is no clear "barcode gap" (distinction between intra-clade and inter-clade distances) for three of the six NSW populations. This is in contrast to the 28S data, which distinguishes only four lineages from NSW, with essentially no variation within each lineage. Despite this, the 28S data generally aligns well with morphological evidence, clearly identifying H. cooloola as a distinct species, supporting also the description of H. coorongooba even though it only appears to differ from H. sclerophylla in the number of styli. Similar genetic distances are observed in 28S data among H. sclerophylla populations from North Nowra, Glenbrook/Burralow/ Nattai and Megalong, however the Broulee and Wellington populations have identical 28S sequences. The low levels of variation in 28S sequences among NSW populations accord with the lack of unambiguous morphological differences among these COI lineages.
Unsurprisingly, given that they are both mitochondrial genes and therefore linked markers, the limited 16S data shows a similar pattern to COI with large distances among Megalong and North Nowra lineages in NSW, a large distance between H. cooloola and remaining samples, and between H. cooloola and the single other Queensland sample.
An analogous situation was reported for the North American nicoletiid silverfish Texoreddellia texensis (Ulrich, 1902) which was initially considered to be a widespread and somewhat variable cave-dwelling silverfish. DNA sequences of the 16S gene (Espinasa & Giribet, 2009;Espinasa et al, 2016) have shown that distinct species exist and that, in most cases, subtle morphological differences could be found. Six species of Texoreddellia are now described, often with overlapping ranges. It would appear that Heterolepisma sclerophylla forms a similar example with several genetically distinct populations, however reliable, non-over-lapping, morphological differences between these populations have mostly not yet been identified. The large variability between individuals within any population, complicated by the continuous moulting, makes quantification of morphological differences difficult and it may be that each separate clade of COI sequences represents a distinct species but we are approaching the practical limits of morphology to separate lepismatid species. Differences in base pairs of 0.9-1.8% in the case of 28S and 7.2% for COI appear to be the lowest levels associated with morphological differences considered indicative of species.
On the other hand, Porco et al. (2012) found large intraspecific differences in COI (11.3-21.5%), corroborated by smaller differences in 28S (0.6-9.5 %), within several widespread species of Collembola. Vink and Brown (2014) found up to 8.3% divergence among COI sequences of Sminthurus viridis (Collembola) from New Zealand but no variation in 28S D2-region sequences from a small sample of the same specimens and concluded that only one species was present. Resch et al. (2014) similarly sequenced both COI and 28S rDNA from Protura, and found that whenever multiple geographic populations were sampled, large intraspecific divergences in COI were observed (up to 21.3%) despite the lack of corroborating morphological characters. Wide variability in COI may be a feature of ancient, low mobility, soil-dwelling taxa.
Unpublished preliminary molecular data suggest it is quite likely that other widespread Heterolepisma species, including H. buntonorum and an undescribed species related to H. highlandi Smith, may also have several genetically distinct lineages. This latter "species" occurs over much the same range as H. sclerophylla but is generally only taken on bark using pyrethrum sprays rather than from leaf litter. Wygodzinsky (1961) considered it very probable that H. howensis Womersley would prove to be a variety of H. zealandica noting that specimens of H. zealandica from the type locality possess four pairs of styli while those from the South Island possess only two pairs in both sexes. Tillyard's original illustration (1924) suggests only three pair of styli although he reports them to be present on segments I-IX. Smith (2014) also mentioned the possibility that H. sclerophylla might prove to be conspecific with H. howensis given the wide variety in morphology observed in the Australian mainland specimens. Although molecular data is currently lacking for the Lord Howe Island and New Zealand specimens, the results of this study suggest that the morphological differences reported are very likely to be indicative of distinct species. This brings into question the previous tendency to accept small morphological variations within species as a typical feature of the Zygentoma. We believe now that H. howensis, with its strongly truncate urotergite X, broader thoracic sternites, single submedial dorsal macrochaetae and a very short ovipositor (reaching only to half the length of stylus IX) is highly likely to be a valid and distinct species. Furthermore, it is quite likely that there is more than a single species of Heterolepisma in New Zealand.

Morphology
The genus Heterolepisma, as currently defined, now contains 26 described species with a wide, largely Gondwanan, distribution. The Australian material available display a number of morphological characters that allow them to be easily sorted into species groups but at this stage no clear picture is emerging as to the phylogeny of the genus.
It will be necessary to redescribe many of earliest described species of Heterolepisma to include characters that are now considered as important but which were overlooked in the early descriptions, notably the chaetotaxy of the frons, the presence of lanceolate scales, details of the trichobothrial areas, the number of combs on urotergite I and urosternite I. Ideally sequence data should be obtained from types or topotypic examples, however the habitat at the type locality is very likely to have changed greatly over the intervening century. Unless it becomes practical to obtain DNA from specimens stored for more than a century in 70-80% ethanol it may never be possible to obtain sequence data that could be linked unequivocally to the type specimens. Recent advances in "next generation" sequencing, making it possible to "re-assemble" the sequence from the remaining fragments (e.g., Ruane and Austin, 2017), offers some hope that this will not be the case.