Occurrence records of Holaptilon were harvested from previous studies (Abu-Dannoun and Katbeh-Bader 2007: one record; Kolnegari and Vafaei 2018: three records; Handal et al. 2018: one record; Yılmaz and Sevgili 2023: three records) and complemented with freshly collected samples (79 records) from Iran, old collected samples (four records) plus freshly collected samples (two records) from Israel, and records extracted from the iNaturalist portal (these records were checked carefully for correct identification and curated by the first author), i.e. five records with IDs: 179338809, 147924203, 147924202, 147924201, 147924199, 81544213). A total of 87 records was obtained and plotted on a distribution map using QGIS v. 3.22. The complete list of records is reported in Table S1.

Fresh specimens and oothecae were collected over a seven-year period (2015–2021) from 15 districts in six provinces of Iran (Mashhad, Khorasan Razavi Province; Arak, Markazi Province; Yasuj, Kohgiluye va Bouer Ahmad Province; Dehdez, Mal agha, Bagh Malek, and Gharibiha village, Khozestan Province; Dasht Arjan, Fasa, Tadovan, and Sahlak, Fars Province; Sooro, Kangan, Tombak, and Jam, Busheher Province). Sampling methods included net sweeping and hand-picking during daytime through careful observation under stones or observing mantids running on the surface of stones, and at night-time mostly on the ground or attracted to some source of light. Once collected, specimens were put in separate plastic containers and later placed in separate plastic jars (15 × 15 × 10 cm) after arriving at the laboratory. Some rocks and sticks were added to the container in order to help the mantids climb and hang, especially during moulting.

The specimens of the four species newly described in this study are deposited in the following institutions or private collections: ESPC Evgeny Shcherbakov, Private Collection, Ramenskoye, Russia; ZMCBSU The Zoology Museum of Shiraz University, Shiraz, Iran; ZMPC Zohreh Mirzaee, Private Collection, Müncheberg, Germany; SDEI Senckenberg German Entomological Institute, Müncheberg, Germany. Other examined material is deposited in the following institutions or private collections: HSC Hasan Sevgili collection (Department of Molecular Biology and Genetic, Ordu University, Turkey; SMNHTAU The Steinhardt Museum of Natural History, Tel Aviv, Israel.

Specimens in the lab were observed throughout their entire lifetime. At adult stage, males and females were placed together to observe their mating behaviour. Once laid, the oothecae were collected to record the following data: number of hatching nymphs, number of instars to reach adult stage, male and female rate per ootheca, male and female longevity, and spermatophore feeding. Information from the oothecae oviposited in the lab was compared with oothecae oviposited in their natural habitat. The individuals were kept at room temperature (25–27°C), with relative air humidity (RH) kept at 50–55% by misting the room on a regular basis. An HTC2 digital terrarium hygrometer (Dongguan, China) was used to monitor RH. According to some occasional observations of the specimens preying on small ants and small flies in nature, laboratory specimens were fed with one to two fruit flies (Drosophila melanogaster Meigen, 1830) or small ants (Trichomyrmex spec.) every third day. Later instars were fed with living mealworm larvae (Tenebrio molitor Linnaeus, 1758) twice a week, all after testing the appreciation and response of this mantid toward these unconventional preys.

After completing their life cycle, the individuals were preserved in 96% ethanol for morphological and molecular studies. Specimens were examined and measurements were taken under a Leica M205 C stereomicroscope with an ocular micrometer. The classification system used in this study follows Schwarz and Roy (2019). Descriptive terminology of external morphology of male, female and preparation of male genitalia followed Brannoch et al. (2017) The ultimate segments of male abdomina were dissected under a stereomicroscope, in order to separate the genitalia from the terminalia. Genitalia were macerated as a whole in 10% KOH solution. Subsequently, they were washed first with distillate water, then in ethanol (70%), and finally in glycerine for 24 hours in order to ensure the complete removal of ethanol. Finally, prepared genitalia were placed in a vial with glycerine drops for further study. The specimens were photographed with a system comprising a Stone Master Stack Unit, an Olympus OM-D E-M1 Mark II camera, and Zeiss Luminar lenses (16 mm, 25 mm, 40 mm, 63 mm). Olympus Capture and Stone Master v.3.8 software was used to acquire digital photos. Stacking was performed with Helicon Focus (v.7.6.1), and a scalebar was added with ImageJ (v.1.53t).

Most of the traditional taxonomy in Mantodea, especially at the genus/species level, is based on external morphological characters, and all current species delimitation in the genus Holaptilon is based on few qualitative characters like the shape of the pronotum and the tip of the supra-anal plate (= 10^th abdominal tergite), the pattern of the fore femora inner colour spot, and the number of posteroventral tibial spines (Yılmaz and Sevgili 2023). Considering the type specimens of all three already known Holaptilon species, the current taxonomy is based on just ten specimens, a clearly insufficient number to measure inter- and intraspecific variability within this genus. To evaluate this variability, all of these characters were considered and studied in a larger number of specimens. Moreover, all the further traditional measures and ratios used as discriminants in Mantodea, including shapes of male and female genitalia, were also included in our analyses.

To measure morphological differentiation among Holaptilon species, volumes and overlaps of morphological traits and their respective centroid distances were calculated using the R package Hypervolume (Blonder et al. 2014). Three sets of measurements, including spine counting on both left and right front legs (AFS, PFS, ATS, PTS, DS), ratios of different body parts in relation to each other, and measurements of different parts of the body (Tables S2–S4) were used. In total, 16 measurements (see below) for all Holaptilon specimens were considered in this study. All spines from both the right and left frontal legs (see below) of each specimen were counted and the distances between different body parts measured and then the ratios were calculated.

Mesocoxal muscle tissue was excised from 34 Holaptilon specimens preserved in 96% ethanol. Total genomic DNA was extracted using the E.N.Z.A.^® Tissue DNA Kit protocol for animal tissue. Four loci were targeted for amplification and sequencing: 16S ribosomal DNA (16S, 508 bp), Isocitrate dehydroxynase (IDH, 718 bp), histone 3 protein-coding for the nucleosome (H3, 330 bp), and cytochrome c oxidase I (COI, 658 bp). Primer sequences and amplification protocols for these four loci are provided in Table S5. PCR products were visualised using gel electrophoresis to confirm the correct amplification or detect undesired contamination. Amplicons were purified with Thermo Scientific Exonuclease I and the FastAP Thermosensitive Alkaline Phosphatase Clean-up Kit. They were sequenced at Macrogen Europe with complements and sufficient overlap with adjacent regions to ensure accuracy of sequence data. Sequences were manually edited and aligned using Geneious R10 (https://www.geneious.com) for nucleotide editing and contig assembly. ClustalW (Thompson et al. 1994) was used to perform multiple sequence alignments of protein-coding genes on MEGA-X (Kumar et al. 2018), which was then converted to Fasta and Nexus formatted files for use in various analysis programs.

We analysed 34 Holaptilon individuals; GenBank accession numbers and sampling localities of specimens are listed in Table S1. In this study, three different datasets were used to analyse the phylogenetic relationship among Holaptilon groups. Dataset number one (1166 bp) contained the combined gene fragments of the COI barcode (658 bp) and 16S (508 bp) concatenated into two partitions. Dataset number two (1048 bp) contained the combined gene fragments of H3 and IDH concatenated into two partitions. Dataset number three (2214 bp) contained the combined gene fragments of all four mentioned genes concatenated into four partitions. A selection of specimens was sequenced for 18S (955 bp) and 28S (490 bp), but because these genes turned out to be (almost) invariable in Holaptilon, these sequences were not used for further analyses. All novel DNA sequences, i.e., 33 COI, 31 H3, 33 16S, 32 IDH, 5 18S, and 7 28S sequences, were deposited in GenBank (accession numbers OR536777–OR536809, OR541990–OR542033, OR545419–OR545480). Gonypetyllis semuncialis Wood-­Mason, 1891 was chosen as outgroup for the phylo­genetic analyses because this genus is morphologically similar to Holaptilon and it belongs to the same family (Gonypetidae) and subtribe (Gonypetyllina).

The best-fit model of nucleotide substitution for all sequences partitioned by gene, GTR+I+G for mitochondrial genes, and HKY for nuclear genes, were selected under the corrected Akaike Information Criterion (AICc; Hurvich and Tsai 1989) using jModelTest ver. 2.1.10 (Darriba et al. 2012). IQ-TREE 1.6 was used to perform a maximum likelihood (ML) phylogenetic tree estimation (Nguyen et al. 2015). In ten independent runs, 1000 ultrafast bootstrap replicates (UFBoot) (Hoang et al. 2018) and 1000 Shimodaira-Hasegawa approximate likelihood ratio test replicates (SH-aLRT) (Anisimova et al. 2011) were calculated. Bayesian analyses were conducted utilizing MrBayes v3.2.7 (Huelsenbeck and Ronquist 2001; Ronquist and Huelsenbeck 2003), employing the most suitable DNA substitution model under the Akaike Information Criterion (AIC). The analysis involved four chains, consisting of two hot and two cold chains, executed in three independent runs for 200,000,000 generations, sampling every 2000th generation. The first 20% of generations were discarded as burn-in. The majority-rule consensus tree was created using the “sumt” command. TRACER (Rambaut et al. 2018) was used to check that analysis has reached an effective sample size (ESS) over 200 in order to ensure correct chain convergence. Posterior probabilities (pp) were obtained for each clade, where pp ≥ 0.95 indicated significant support for clades. The run with the best log-likelihood score was selected. Consensus trees generated were visualised in FigTree 1.4.2 (http://tree.bio.ed.ac.uk/software/figtree), and then edited using Inkscape vector graphics editors (ver. 1.2). Three different protocols, Assemble Species by Automatic Partitioning (ASAP) (Puillandre et al. 2021), Bayesian implementation of the Poisson Tree Processes model (bPTP) for species delimitation (Zhang et al. 2013), and the multi-rate PTP (mPTP) (Kapli et al. 2017) were used with pairwise genetic distances for species delimitation analyses. For ASAP, FASTA-configured files of all, nuDNA (H3, IDH), and mtDNA (COI, 16S) datasets were used for analyses on the ASAP website (https://bioinfo.mnhn.fr/abi/public/asap) with two replacement samples used to estimate distances, namely the simple p-distance model and the K2P model. bPTP was run on the bPTP online server (https://species.h-its.org) with default values (100,000 Markov chain Monte Carlo [MCMC] generations, thinning = 100, burn-in = 0.1, and seed = 123), using the NEXUS formatted input files of all, nuDNA (H3, IDH), and mtDNA (COI, 16S) datasets. The multi-rate PTP (mPTP) method was also used according to the study by Kapli et al. (2017) which incorporates the potential divergence in intraspecific diversity to the PTP and implements a fast method to compute the maximum likelihood delimitation from an inferred phylogenetic tree of the samples. The specific settings used in the mPTP analysis included a total of 100,000 MCMC generations, 10,000 generations as burn-in, and thinning every 100 generations. An uninformative prior was applied to the number of species, and the genetic model used was GTR+G. Species were delimited based on a threshold of 0.95 for posterior probabilities (https://mptp.h-its.org).

All 34 samples used for our molecular analyses were tested for an infection of the Wolbachia surface protein-coding gene (WSP, 549 bp) using the primer pair wsp81F (5′-TGG TCC AAT AAG TGA TGAAGA AAC-3′) and wsp 691R (5′-AAA AAT TAA ACG CTA CTC CA-3′) with the following PCR protocol: 95°C for 5 min, followed by 38 cycles at 95°C for 30 s, 55°C for 90 s, 72°C for 2 min, and terminated with a final extension step at 68°C for 30 min. The PCR products were loaded onto a 1.4% agarose gel and stained with GelRed (Biotium, Fremton, USA) to check for positive or negative infection. We included tests on Wolbachia to ensure that our molecular markers were not impacted by this entero-parasitic bacterium. Wolbachia can influence mitochondrial DNA (mtDNA) patterns, resulting in misleading phylogenetic signals among closely related species and specimens of the same species (Ritter et al. 2013). It can either obscure species diversity due to mtDNA introgression among species (Narita et al. 2006; Whitworth et al. 2007), or conversely, it can induce extensive mtDNA divergence due to prolonged reproductive isolation among lineages infected and not infected with Wolbachia, which may even lead to the formation of new species (Bordenstein et al. 2001).

We estimated the divergence time using BEAST 2 version 2.7.5 (Bouckaert et al. 2019). Two datasets were utilised: one from mitochondrial genes (COI, 16SrDNA) and another from nuclear genes (H3, IDH), which were concatenated to prevent coalescence and zero-length branches. Tree models were set as linked, the site model and clock model were set unlinked. The subsets and substitution models were determined using jModelTest version 2.1.10. For the mitochondrial dataset, the GTR model with estimated base frequencies and a gamma distribution (with 4 categories) was chosen. The nuclear dataset employed the HKY model with estimated base frequencies and the same gamma distribution setup. As fossils for Holaptilon or closely related genera are unavailable, the tree was calibrated using standard gene substitution rates as done in previous studies (Papadopoulou et al. 2010; Wendt et al. 2022). Consequently, a clock rate of 0.0177 was applied to the mitochondrial dataset, and 0.00177 to the nuclear dataset, using an optimised relaxed clock (Papadopoulou et al. 2010). To assess the potential effects of different models, two separate analyses were conducted using Yule and Birth-Death tree priors. Each analysis comprised four independent Markov Chain Monte Carlo (MCMC) runs, each running for 50 million generations and sampling trees every 5,000 generations. After discarding the initial 10% of trees as burn-in, convergence was verified using Tracer version 1.7.1 (Rambaut et al. 2018). The final trees were combined using TreeAnnotator v.1.10.4 and edited in FigTree v.1.4.4 (http://tree.bio.ed.ac.uk).

To study the historical shifts in the geographical ranges of Holaptilon, two models were used for biogeographical range expansion, the Dispersal-Extinction-Cladogenesis (S-DEC) model and the Dispersal-Vicariance (S-DIVA) model on RASP 4.3 (Yu et al. 2020). The input for this analysis was an ultrametric tree created using BEAST v2.7.5. To refine the analysis, the outgroup (Gonypetyllis semuncialis) was removed from the tree using the tool provided by the RASP software. Five geographical regions were here defined based on our understanding of the species current distribution: (A) south-western part of Iran, (B) north-western part of Iran, (C) Turkey, (D) north-eastern part of Iran, and (E) Israel/Jordan.

To address uncertainties arising from the tree’s structures, all trees sampled from BEAST analysis, except for the initial 500 trees, were incorporated. For S-DIVA analysis, the “Allow Reconstruction” feature was selected, which allowed a maximum of 100 reconstructions utilising three random steps. This was followed by conducting up to 1000 reconstructions for the final tree. Each node in the analysis was granted the potential for up to four distinct areas. The outcomes of the most appropriate S-DIVA reconstructions were then summarised by employing the pruned maximum-clade-credibility tree derived from this Bayesian phylogenetic study. For S-DEC analysis, the probability of dispersal between areas was considered as equal, and all values in the dispersal constraint matrix were set to 1 with four as the maximum number of areas.

A brief and preliminary evaluation of the conservation status of each species was done referring to the IUCN Red List Categories and Criteria and Guidelines for Using the IUCN Red List Categories and Criteria (version 15.1, 2022). The aim of this analysis is to give a preliminary assessment that can be used as a base for a future standard IUCN Red List evaluation. The Extent of Occurrence (EOO) is defined as minimum convex in which no internal angle exceeds 180 degrees and which contains all sites of occurrence, the number of known presence locations, the presence of human activities or habitat degradation, and the fragmentation of populations, were considered to evaluate a Threat Level according with IUCN standards.

The genitalia of 25 Holaptilon female and 30 male specimens were examined in this study. Some female genital characters (i.e., the shape of the gonapophyses, gonoplacs, and gonocoxae) were proven to be useful in delimiting genera and species-level boundaries in some praying mantids (Brannoch and Svenson 2016). Our results showed that the female genitalia of the 25 analysed female specimens exhibited strong variability and weak discriminant differences in shape (Fig. 1).

A similar situation was obtained for male genitalia (25 males from Iran, including one paratype of H. brevipugilis, four males of H. pusillulum from Israel, and photos of a single male’s genitalia of H. yagmur; Yılmaz and Sevgili 2023). All specimens were grouped into seven morphotypes (Ha–Hg) based on the length and shape of afa, sclerotisation of pba anteriad of afa, and the shape of the posterior edge of vla (Table 1). The Ha morphotype (11 specimens) is characterised by having long, saber-like afa, which has constant width and a gently curved or straight apex, pba anteriad of afa is not sclerotised; posterior edge of vla is gently angular (Figs 2a and 2b). The Hb morphotype (5 specimens) is characterised by short, finger-like afa, which markedly narrows towards the apex and often is curved near it; pba anteriad of afa is markedly sclerotised; posterior edge of vla is markedly truncated (Figs 2c and 2d). The Hc morphotype (2 specimens) is characterised by short, finger-like afa, with a curved apex, pba anteriad of afa wholly sclerotised; posterior edge of vla is oblique (Figs 2e and 2f). The Hd morphotype (2 specimens) is characterised by short, finger-like afa with a gently curved apex, pba anteriad of afa less sclerotised; posterior edge of vla is truncated, but the truncation is not as clear-cut as in morphotype Hb (Figs 2g and 2h). The He morphotype (6 specimens) is characterised by very short, finger-like afa, which is curved at the apex, pba anteriad of afa that is not sclerotised, with the posterior edge of vla oblique (Figs 2i and 2j). The Hf morphotype (3 specimens) is characterised by long, hook-like afa which is strongly sclerotised (Figs 2k and 2l). The Hg morphotype (1 specimen) is characterised by short, straight afa, pba anteriad of afa is sclerotised (Yılmaz and Sevgili 2023). Note that all these differences may show intraspecific variability.

Taken together, the results point out to six groups of specimens that we treat as species: the Ha morphotype (= H. abdullahiisp. nov.), the Hb morphotype (= H. iranicumsp. nov.), the Hc morphotype (= H. tadovaniensissp. nov.), the Hd morphotype (= H. brevipugilis), the He morphotype (= H. khozestanisp. nov.), the Hf morphotype (= H. pusillulum), the Hg morphotype (= H. yagmur), which will be properly described or redescribed in the section “Taxonomy”.

Characteristics regarding the head, thorax, forelegs, and supra-anal plate of 51 adult specimens were examined and compared. Holaptilon species exhibited a generally homogenous morphology and the traditionally used morphological characters mentioned above all showed comparatively small differences. The only apparently constant differentiating character were the five posteroventral fore-femoral spines in H. abdullahiisp. nov., whereas all other species of Holaptilon seem to possess only four. The number of other spines in the forelegs of Holaptilon species is variable. Variability in the number of spines was also recorded in the left and right forelegs of the same individual (Table S2). As a result, it appears that the external morphology of the species in this genus is highly variable and requires a multidisciplinary approach to understand its species boundaries. Therefore, the only morphological characters that are reliable as diagnostic characters for species are the number of posteroventral spines of the fore-femora and the characteristics of male genitalia.

The morphometric ratios PL/HH and secondarily HH/HW were the most discriminative in both males and females (Fig. S1). These ratios, when plotted together, demonstrated at least a clear separation between the taxon groups iranicum + tadovaniensis and abdullahii + brevipugilis + pusillulum, while the placement of khozestani within the latter group seemed to be more defined in males (Fig. S2). This indicates that the male individuals within the “khozestani” group tend to have distinct or well-defined values for morphometric ratios compared to males from other taxon groups. On the other hand, the females within the “khozestani” group might have more overlapping values with females from other groups, making it somewhat more difficult to clearly distinguish them based on these specific morphometric ratios. The dendrogram, at least for females, was very similar to the molecular phylogeny (Fig. 3), supporting, despite the evident variability, the existence of separate lineages inside the genus Holaptilon. While H. iranicum seemed quantitatively well separated from the type species H. pusillulum, the boundaries among the other species were less well defined using only traditional morphology, but the integration of the dendrogram using morphometric ratios (Fig. 3) with the results of the analysis of molecular data (Fig. 4) in this study advanced our understanding of the phylogenetic relationships among the investigated taxon groups. By employing discriminative morphometric ratios (PL/PW and HH/HW), clear distinctions were observed between the groups “iranicum + tadovaniensis” and “abdullahii + brevipugilis + pusillulum” with potential sexual dimorphism in the “khozestani” group. The integrative analysis demonstrated a concordance between morphometry-based and genetic-based groupings, providing compelling evidence for the validity of taxonomic distinctions and emphasising the importance of considering both data types for accurate phylogenetic inferences. For all these considerations, a traditional dichotomous key for species identification, based on morphology alone may be misleading for this genus. A key to the species is however here proposed to give a preliminary help for faunistic studies and general field work.

In total, five species were included in the analysis: H. abdullahiisp. nov. (N = 20 specimens), H. khozestanisp. nov. (N = 13), H. brevipugilis (N = 4), H. iranicumsp. nov. (N = 11), H. tadovaniensissp. nov. (N = 3). The morphological hypervolumes showed low levels of overlap among species (Figs S3, S4; Tables S6, S7), with the lowest overlap between H. tadovaniensissp. nov. and H. iranicumsp. nov. and the highest overlap between H. khozestanisp. nov. and H. abdullahiisp. nov. followed by H. brevipugilis and H. khozestanisp. nov. Species had the highest overlap with set 1 estimated with three principal component axes, and the lowest with set 3 estimated with four principal component axes. In our study, some species were represented by numerous individuals while others only by few individuals, due to the difficulty finding the specimens in their natural habitat. The number of specimens in each population or sample is a critical factor in analysing morphological variation and interpreting results, especially in analyses of morphological hypervolumes. Sample size can influence the robustness of conclusions and the amount of variation observed. The differences in sample sizes therefore have to be always considered in the interpretation of our results.

Our test for Wolbachia infection proved negative for all our 34 Holaptilon samples used in the molecular study, so we did not find evidence for an effect of Wolbachia infection on the phylogeny of Holaptilon.

All applied phylogenetic methods (Bayesian, Maximum likelihood) resulted in trees that did not differ much in their topology (Fig. 4), i.e., all groups of specimens were mostly assigned to the same main clades, and the relationships between these clades were mostly identical, except for the BI tree obtained using mitochondrial genes (COI, 16S), which suggests that H. brevipugilis might not form a monophyletic group. It is important to note that this latter relationship lacks robust support based on Bayesian analysis (Fig. 4b).

The H. iranicum clade in all three trees represented the first split and was sister to the clade H. tadovaniensis + all other Holaptilon taxa. The H. pusillulum clade remained the sister clade to the H. brevipugilis clade in the trees based on mitochondrial genes (COI, 16S) and the combination of all four genes (COI, 16S, H3, IDH), but was sister to the clade H. abdullahii and H. brevipugilis + H. khozestani in the tree based on nuclear genes (H3 and IDH). One combined tree (from ML tree and BI tree, which were identical in topology but with different support values) with well-supported nodes is given in Figure 4. The new species were resolved as monophyletic. Most of the nodes were well supported (0.89/90‒1/100), and all taxa were resolved as independent lineages with long branches and they clearly belong to different clades than the type species H. pusillulum.

The number of species defined by the species delimitation software for mtDNA and all markers combined were even higher than our assumptions above, i.e., ASAP multilocus (Kimura): ten species; bPTP multilocus: 14 species; mPTP multilocus: ten species. However, restricting the data to the nuDNA markers (H3, IDH) resulted in almost the same number of species as in our assumptions given above, ASAP: 6 species, bPTP and mPTP: 7 species (Fig. 4).

Our molecular dating based on concatenated mtDNA (COI, 16S) and nuDNA (H3, IDH) datasets suggested that the most recent common ancestor of Holaptilon likely lived in the late Miocene, 8.4‒6.1 Mya (Fig. 5b). The first divergence event in Holaptilon occurred approximately 6.06 Myr ago when the clade containing H. iranicum branched off from the rest of the Holaptilon group. Around 5.5 Mya, the clade containing H. tadovaniensis split off from all remaining Holaptilon taxa. Roughly 3.7 Mya, the split between the two sister clades H. abdullahii + H. khozestani and H. brevipugilis + H. pusillulum occurred.

Since the biogeographic analyses results were consistent for both the S-DIVA and S-DEC methods, one of them was utilised (Fig. 5a). The model assumed an origin in south-western Iran (A) with range expansion to the north-western part of Iran (B) and Israel/Jordan (E) starting 3.7 Mya. They separated from A to BE (dispersion) between 3.7 and 3.4 Mya, and then separated into B and E (vicariance), starting about 3.4 Mya. E remains stable with no changes until today. From B, the taxon expanded its range moving to BD and BC (dispersion), starting about ~2.8 Mya to ~1.0 Mya. In the last ~1My, two distinct vicariant events occurred, which separated B, C, and D into distinct areas (Fig. 5a).

Today, several species (i.e., H. iranicum, H. tadovaniensis, H. khozestani, H. brevipugilis) occur together in the Zagros Mountains forest steppe ecoregion (Dinerstein et al. 2017). Among these species, H. brevipugilis exhibits the widest distribution, spanning three ecozones, namely the Eastern Mediterranean conifer-broadleaf forests, the Zagros Mountains forest steppe, and the Central Persian desert basins. Conversely, H. abdullahii and H. pusillulum are confined to the South Iran Nubo-Sindian desert and semi-desert, and the eastern Mediterranean conifer-broadleaf forests, respectively. Although Holaption mantids have a vast geographical distribution (extent of occurrence EOO: approximately 1.5 Mio km²), encompassing four different ecoregions, they are often encountered in fragmented and scarce populations. Their places of occurrence frequently deviate from the typical vegetational pattern of the overall ecoregion but exhibit certain consistent microecological features, such as rocky grounds sparsely covered by vegetation, sometimes near water streams. Notably, these habitats are often found within non-natural or modified environments, such as cultivated fields or urban areas (Figs 6–8).

While five of the six species (i.e., H. abdullahii, H. iranicum, H. khozestani, H. pusillulum, H. tadovaniensis) as far as known today have well-defined and restricted distributions, only H. brevipugilis has a vast but possibly fragmented distribution spanning 20 degrees of longitude (39‒59°E) across the Middle East, while most likely maintaining a relatively narrow latitudinal range (34‒37°N). This large distribution of H. brevipugilis may encompass regional subspecies.

Holaptilon individuals were observed in 16 districts of five provinces of Iran during different years by the first author (2015‒2022). The oothecae were found deposited under stones of different sizes. Oothecae were small, yellowish, with a spongy texture, delicate, and without apex (Fig. 13). Size range was 5–8 mm; larger oothecae contained eight eggs, the smaller ones five to six. Females tended to lay their oothecae in small, already existing holes in stones.

Nymph emergence observations of H. abdullahiisp. nov. were made from mid-March onwards, adults were observed from the end of May to the end of July (males) and the end of August (females). Nymph emergence observations of H. iranicumsp. nov. were made from mid-May, adults were observed from the first week of August to the end of September (males) and the end of October (females). Nymph emergence observations for H. khozestanisp. nov. were made from the first week of July, adults were observed from the end of August to the end of October (males) and the end of November (females). Nymph emergence observations for H. tadovaniensissp. nov. were made from mid-February onwards, adults were observed from the end of July to the end of August (males) and mid-September (females).

Adult males (N ≥ 15) were running between or on rocks during the hottest time of the day in August. They were observed actively running between stones during the day and on the ground at night. Seven of the 15 observed males were approaching females from the backside and suddenly jumped on their backs. Females (N ≥ 21) were mostly found under stones. All known habitats had natural water sources like permanent rivers or springs.

H. abdullahiisp. nov. females were exclusively observed beneath stones, while the behaviour of males differed. Some males (N = 6) were discovered while overturning stones, others were observed running on the ground towards light sources at night (N = 1). Additionally, some males were seen running between or on stones during midday in Tombak (N = 2), Kangan (N = 2), and Soroo (N = 1). The elevational range of the locations where specimens of this species were found varied from 10 to 637 m above sea level (asl).

One female of H. iranicumsp. nov. from Yasuj was found under Astragalus sp., whereas both females and males from Sahlak were observed among grasses. In Arjan and Fasa, males and females of this species were discovered under stones. Elevation ranged from 1000 to 1800 m asl.

H. khozestanisp. nov., females and males were primarily observed under stones (N = 8), but three males were found running on the ground. The elevation ranged from 800 to 2000 m asl.

Regarding H. tadovaniensissp. nov., one female and two males of this species were discovered under stones (N = 3). The elevation was approximately 1000 m asl.

The oothecae of different species were collected from their natural habitats. They hatched eight to nine weeks after being collected. Oothecae sizes, incubation duration, number of eggs per ootheca, and number of hatched nymphs are given in Table 2. Two oothecae of H. khozestani were parasitised by Palachia pulchra Bouček, 1969 (Fig. 13) (Mirzaee et al. 2021). No mantid nymph emerged from these two oothecae; 12 and 9 parasitoids emerged instead, respectively. Each male individual moulted five times and each female six times to reach adult stage. The observed male courtship display was the same as observed by Kolnegari (2020) for H. brevipugilis, and it consisted of bending the abdomen up and downward, moving the antennae fast, sometimes with trembling boxing forelegs alternately, and jumping on the back of the female. The trembling boxing forelegs were mostly observed when males were facing females (90% of males). Most of the males tried to approach females from the back (11 males). Six males also ran towards females and jumped on them from the front (Fig. 14). Seven males approached females laterally. All 24 males were successful in the mating process. Mating duration was four to seven hours. No cannibalism was observed. Spermatophore feeding behaviours were observed for all 31 females (Fig. 14d).

Three males of H. iranicum from Arjan district were introduced to three females of H. khozestani. Two of these males tried to escape from the females, and they avoided facing them. One male approached the respective female and jumped on its back but was unable to successfully mate (Fig. 14b). After three hours attempting, the male’s reproductive organ still failed to connect with the female ’s reproductive organ, resulting in an unsuccessful attempt. As a result, the male was removed from the female. Two males of H. abdullahii were introduced to two females of H. iranicum, both of which attempted to escape from the females.

Beier (1964) described the type species H. pusillulum and mentioned that the femora of this genus have four very short external spines. However, our study revealed that H. abdullahiisp. nov. possesses five short external spines on the femora. Additionally, Beier (1964) noted that the anterior tibiae of this genus have 13 short external spines, while our research indicated that the number of spines on the anterior tibiae is variable, ranging from 9 to 13. Furthermore, Beier (1964) mentioned that this genus belongs to the Gonypetini tribe, which is primarily distributed in the Oriental region. However, our study demonstrated that not all genera within this tribe, such as Holaptilon, are restricted to the Oriental region, but are also distributed in the Palearctic region. In addition, the genus Elaea Stål, 1877, also belonging to Gonypetini, is distributed from Africa to SW Asia.

In previous studies on Holaptilon, a rather limited number of 2–5 specimens were utilised to characterise new species in this genus. Our study represents the first comprehensive revision of the genus, counting on the examination of 87 specimens. Based on the examination of this material, we show that all the characters previously used as diagnostic characters at the species level were actually within the range of intraspecific variation, not allowing species delimitation.

In fact, a notable amount of variability was observed across different traits. Historically, male genitalia attributes served as traditional tools for describing and classifying mantid species. However, given the pronounced variability observed, this approach has not been effectively employed for certain species within the genus Holaptilon, such as H. pusillulum, H. brevipugilis, and H. yagmur.

Although some aspects of male genital structures, specifically the curvature or sclerotisation of phallomere processes like afa, pba, or vla, might at times offer a secondary qualitative differentiation, the prevailing variability inherent in both male and female genital structures renders this trait inconclusive for species characterisation. The variation in male genitalia within the genus Holaptilon could be attributed to a combination of factors, including limited ranges without overlap and hence missing sexual selection among species; in turn, high species range overlap of closely related species should be associated with remarkable sexual selection. Investigating the interplay between these factors and their effects on male genitalia diversity would provide valuable insights into the complex mechanisms driving the observed variation in this character complex. With regard to colouration, our study revealed that the spots and colouration of the inner view of the fore-femora of Holaptilon species varies among specimens of the same species and is thus not suitable for distinguishing species (Fig. S5). Variability in the foreleg colouration and spot patterns was also observed in other Mantodea, such as Anasigerpes Giglio-Tos, 1915 (Roy, 1965). A comparable level of intraspecific variability is evident in other characters, such as the degree of concavity of the posterior margin of the meso- and metanotum or the number of posteroventral spines of the foretibiae (Fig. 12; Suppl. Material File 2: Table S2). Similarly, the shape of the thorax, supra-anal plate, and the numbers of posteroventral spines of the foretibiae, which previously were used to distinguish the three so far known Holaptilon species (Yılmaz and Sevgili 2023; Figs 11, 12, 15; Table S2) are also variable. Such a level of variation could be related to the relatively young clade status of the group. The limited time for divergence, high genetic diversity, incomplete reproductive isolation, adaptive potential, and the possibility of rapid speciation, all may contribute to the wide range of phenotypic traits and characteristics observed among the species in this genus.

Consequently, describing Holaptilon species based on a limited number of specimens and relying solely on external morphology to define and distinguish the species, as hitherto conducted, is unreliable and does not deliver adequate characters for a morphometric separation of species in this genus. In order to surmount these taxonomic challenges, we adopted alternative approaches, aiming to identify distinct morphological characters and understand morphological variability which can potentially offer valuable insights into the evolutionary narrative of Holaptilon species.

For this purpose, hypervolume analyses were employed for the evaluation of the morphological data regarding measurements of different body parts, ratios of the body parts with respect to each other, and also counting different spine types on the raptorial legs. The outcomes of our ecological, phylogenetic, and morphological analyses revealed low overlap (i.e., distinctiveness) between H. tadovaniensissp. nov. and H. iranicumsp. nov., and high overlap between H. khozestanisp. nov. and H. abdullahiisp. nov. followed by H. brevipugilis and H. khozestanisp. nov. These results could be explained by their ecological similarity, while the groups recovered in fact inhabit similar habitats—specifically their tendency to inhabit the spaces beneath stones in mountainous regions with a permanent water source (Fig. 7). This, in combination with the variation in external morphology among individual and their closely intertwined phylogenetic relationships, potentially contribute to the patterns recovered (Tables S6, S7, Figs S3, S4).

Most of the nodes of the phylogenetic trees had high support (1/100) (Fig. 4). The ASAP algorithm of nuDNA supported our hypothesis on the existence of 6 species, whereas the bPTP and the mPTP algorithm suggested an additional species split within H. khozestanisp. nov. However, the ASAP, bPTP, and mPTP algorithms of combined genes and mtDNA genes alone far overestimated the number of species; the detected species numbers were considerably higher for all combined genes and mtDNA, than the morphologically distinguishable ones and the ones distinguished by nuDNA. These high species number detections could also be an artifact due to the small sample size available from many populations: Most of the clades are represented by few individuals (several times only one), making a concise delimitation of species by an algorithm questionable. Adding more individuals might solve this problem, a difficult task in such a rare and difficult-to-spot genus as Holaptilon. Nevertheless, a general pattern exists among all outputs: All algorithms suggest that H. iranicumsp. nov., H. khozestanisp. nov., and H. brevipugilis are possibly split further into cryptic species, which form distinct clades within the tree separated by long branches due to high genetic distances. Therefore, their existence or at least separate evolutionary units within these three species (and thus a wider hidden variability and biodiversity in this mantid genus) cannot be excluded. This hypothesis needs to be tested more thoroughly using a larger number of samples.

The biogeographic analysis provided intriguing insights into the evolution and historical range dynamics of the genus Holaptilon, shedding light on its potential origin and the evolutionary pathways of its species. Unfortunately, it was not feasible to incorporate genera closely related to Holaptilon as outgroups in the RASP analysis due to a lack of available information and challenges associated with obtaining specimens for conducting the relevant analyses. Nonetheless, the results suggest a compelling narrative of range shifts and vicariant events that have shaped the distribution of Holaptilon over millions of years.

The biogeographical reconstruction proposed that the genus Holaptilon likely originated in the southern parts of Zagros mountains (south-western Iran, area A) dating back at least 6 million years (Mya). This is also supported by the fact that four of the six known species occur in the Zagros Mountains forest steppe ecoregion (Dinerstein et al. 2017); in addition, H. abdullahii is restricted to the desertic coastal region south of these mountains, i.e. the western edge of the South Iran Nubo-Sindian desert and semi-desert. The most recent common ancestor of all extant Holaptilon species should have reached the southern Zagros region immigrating from the east (where most extant relatives of the genus are found) in the late Miocene, dating back as far as 8.5 Myr, as indicated by the 95% highest posterior density (HPD) interval (Fig. 16a). This ancient origin is underlining the deep evolutionary history of the genus. An important turning point in the history of Holaptilon apparently coincided with the Messinian Salinity Crisis (5.96‒5.33 Mya) (Krijgsman et al. 1999). This period of environmental upheaval and increased aridity in the greater Mediterranean region (Hsü et al. 1977) likely triggered the fragmentation and isolation of populations within southern Iran, resulting in the basal split into three lineages of which two evolved into the species H. iranicum and H. tadovaniensis, the third into the other four species (Fig. 16b).

While H. iranicum and H. tadovaniensis most likely remained close to their centre of origin (i.e., area A) throughout time without major range dynamics, it must have been the third basal Holaptilon lineage that expanded prior to 3.7 Mya out of the southern foothills of Zagros to north-western Iran (area B) and further westwards to Palestine (area E) (Fig. 16c). This large range must have been successively fragmented a little later into four parts, in each of which one Holaptilon species evolved (Fig. 16d). This range fragmentation most likely took place in a relatively short time window (i.e., at 3.7‒3.4 Mya), maybe triggered by the climate cooling known in the late Pliocene, although that seems to have happened somewhat later (i.e., 3.2 Mya) (Thunell 1979; Bertoldi et al. 1989). This period of isolation likely played a pivotal role in enhancing genetic divergence and promoting speciation, thereby reinforcing the genetic distinctiveness of populations occupying discrete regions.

At latest 1.0 Myr ago, another important range expansion has to be postulated for the by then already existing species H. brevipugilis from north-western Iran (area B), in western direction to south-eastern Turkey (area C) and in eastern direction to north-eastern Iran (area D), followed by (at least temporal) range fragmentation and vicariance about 1.0 Myr ago (Fig. 16e). These range dynamics might have been triggered by the cyclic fluctuations of the Pleistocene, possibly in particular by the Mid-Pleistocene Transition (1.2‒0.8 Mya) (Berends et al. 2021). However, the time elapsed since this initial vicariance event was apparently not sufficient for further speciation. However, due to the difficulties in spotting these mantids in the field, it is unknown whether the extant distribution of H. brevipugilis is mostly continuous from south-eastern Turkey to north-eastern Iran or highly fragmented, maybe in just three core areas (Fig. 16f).

The species belonging to this genus are restricted to geographically small areas, making them particularly susceptible to the risk of extinction due to the combined effects of climate change and human activities. However, they also exhibit some traits that make them excellent bioindicators, similar to other praying mantids. This means that changes in their population and behaviour can provide valuable insights into the overall health and ecological balance of their habitats (Battiston et al. 2020).

According to this research, all members of this genus appear to be univoltine, with only one generation per year. This is almost the same for most temperate mantid species (Hogue and Powell 1980).

The size, shape, and colour of mantid oothecae can be influenced by different biotic and abiotic factors, such as temperature, food availability, humidity, genetics, the presence of males (Robert 1937; Breland and Dobson 1947; Hurd et al. 1995), natural enemies that parasitise or prey on their eggs such as chalcidoids of the families Eupelmidae, and Torymidae (Mirzaee et al. 2021a), clerid beetles such as Trichodes (Coleoptera: Cleridae), dermestid beetles such as Attagenus, Dermestes, Orphinus, and Thaumaglossa (Kershaw 1910; Hawkeswood 2003; Mirzaee et al. 2022a), as well as other arthropods (Mirzaee et al. 2021b). These factors not only have the potential to influence the structure of oothecae, but also the population dynamics of mantids in the wild. The oothecae of Holaptilon species in this study were also affected by these factors, such as parasitisation by the parasitoids Palachia pulchra and deformation, for which we assumed some of the factors mentioned above (Table 2, Fig. 13).

Males in a variety of insects transfer sperm to females via an externally attached spermatophore, which the females then remove and consume. Males in most Mantodea genera transfer sperm to females via internally placed spermatophores. In Holaptilon species, the male inserts the spermatophore with its genitalia into the female’s genital chamber, from which it is expelled (while still attached) and consumed by the female. (Fig. 14d). This behaviour might be due to an adaptation of Holaptilon species to the rather harsh environmental conditions they live in, where every source of nutrition is extremely valuable. It also might be the reason why cannibalism is not observed in Holaptilon, as females receive some sort of “gift” when mating with males. Holwell (2007) documented this behaviour for the first time in four species of Ciulfina Giglio-Tos, 1915. This behaviour was documented among other females of Mantodea species, e.g., Hierodula tenuidentata (Mirzaee et al. 2022b), and seems to be a well-known phenomenon in Mantodea (Bischoff et al. 2001). Two species in this genus, H. iranicum and H. tadovaniensis, coexist in the same habitats, sharing remarkably similar morphology, including reproductive structures, but exhibit distinct genetic patterns (Fig. 4). The evolutionary divergence of these cryptic species requires in-depth exploration of their behavioural patterns. Notably, sexual selection in cannibalistic mantids can be a potent evolutionary driver, and the presence of courtship behaviours and the occurrence of unsuccessful mating attempts in laboratory conditions suggest that further research is needed to investigate this perspective.

The conservation status of the different species of Holaptilon is variable and a full knowledge on the real distribution, presence localities, population trends and threats for each single species is still lacking and further researches in this way are encouraged. However, in this preliminary evaluation, four species were assessed as Endangered (abdullahii, iranicum, khozestani, pusillulum), one as Data Deficient (tadovaniensis) and one as Least Concern (brevipugilis). On a general and preliminary level, these species seem to need urgent conservation efforts to improve their status.