We collected 111 specimens from two sites in the Sétif region of northeastern Algeria. The first, El-Ourecia, is a wasteland (36° 14'43.5" N; 05° 24' 05.9" E; elevation 1040 m). Soil salinity is very low; the electrical conductivity (EC) level (1/5) is 0.182 deciSiemens per meter (dS/m). The second, Bazer Sakhra, is a sebkha (36° 04'18" N, 05° 39'56" E; elevation 915 m). Soil salinity is high (EC [1/15] = 18.43 dS/m). All specimens were frozen at -20 °C until their use in the analyses. Taxonomic identification was carried out using taxonomic keys that employ external morphology (Chopard (1943), Mestre (1988), Defaut (2006), Belmann & Luquet (2009), Fontana et al. (2019), and Louveaux et al. (2020)). Six species were identified: Oedipoda miniata mauritanica (Lucas 1849); Acrotylus patruelis (Herrich-Shaeffer 1838); Acrotylus insubricus insubricus (Scopoli 1786); Sphingonotus (Neosphingonotus) azurescens (Rambur 1838) (see specific discussion later); Sphingonotus (Neosphingonotus) finotianus (Saussure 1885); and Sphingoderus carinatus. (Saussure 1888). . Fifty-one individuals were used in the DNA barcoding analysis, where each species was represented by at least seven specimens. The only exception was A. patruelis, for which we only had a single specimen because it is rare in the study area (see Supp. Table I for details). All the grasshoppers (n = 111) were used in the chemotaxonomic analyses.

Genomic DNA was extracted using the Wizard® Genomic DNA Purification Kit (Promega, Madison, WI, USA) in accordance with the manufacturer's instructions. The antennae were used or, in their absence, the legs. Prior to the extraction, the samples were stored in 96% alcohol at 4 °C. The barcoding region of the COI gene (650 bp) was amplified using the modified primers described in Moulton et al. (2010): MDG-F (5'TYTCAACWAAYCAYAARGAYATYGG-3') and MDG-R (5'TADACTTCWGGRTGWCCRAARAATCA-3') for Oedipoda miniata mauritanica, Sphingonotus (Neosphingonotus) azurescens, Sphingonotus (Neosphingonotus) finotianus, and Sphingoderus carinatus. For Acrotylus patruelis and A. insubricus insubricus, COI-F and COI-R were used instead (Husmann et al., 2012). The PCR reaction mixture had a total volume of 40 μl and contained 2 ng of DNA template, 1 × Qiagen Multiplex PCR Master Mix, and 2 μM of each primer. PCR amplification was performed using an ESCO Swift Maxi® Thermocycler. The initial denaturation step at 95 °C (15 min) was followed by 35 iterations of the following cycle: 94 °C (30 s), 52 °C (1 min 30 sec), and 72 °C (2 min); there was a final extension step at 72 °C (10 min). The PCR products were sequenced by the GenoScreen Institute (http://www.genoscreen.fr) using BigDye 3.1 and a 96-Capillary 3730 xl DNA Analyzer (Applied Biosystems, Foster City, CA, USA). Sequences were inspected, corrected, and aligned using Geneious® Pro v. 5.6.6 (http://www.geneious.com/). They were then deposited in GenBank.

The 51 aligned COI sequences were analyzed using neighbor-joining (NJ), maximum-likelihood (ML), and Bayesian inference (BI) methods. The NJ method was applied using PHYLO_WIN (Galtier et al., 1996). The ML method was applied using PhyML (Guindon & Gascuel, 2003). MrAIC was used to find the appropriate sequence evolution model for the data (GTR + G) (Nylander, 2004). BI was carried out using MrBayes v. 3.1.2 (Huelsenbeck & Ronquist, 2001). The program was run for 15 million generations. No prior assumptions were made about the topology. All the sequences were analyzed using the COI reference sequences available for Oedipodinae in GenBank and BOLD (https://www.boldsystems.org/). These reference sequences are indicated in the phylogenetic tree (GenBank or BOLD accession code followed by species name). Two sequences were used for the outgroup: one from Chortophaga viridifasciata (JQ513034) and one from Lilaea fuliginosa (KM243529).

CHCs were extracted from all the specimens using pentane (Blomquist et al., 1984); solvent quantity varied between 2 and 5 ml because it depended on specimen size. Immersion time was 10 min, and, during this period, the mixture was agitated several times. The resulting extract was evaporated under nitrogen flow. Between 40–80 μl of 10‑5 g/ml of n-eicosane (n-C20) was added as an internal standard to calibrate the gas chromatograph (GC) profiles; the exact quantity again depended on specimen size. Then, 2 μl of the extract was injected into a GC (Agilent Technologies 6850, Santa Clara, CA, USA) equipped with a flame ionization detector (FID) and an apolar HP-1 capillary column (Agilent, 100% methylsiloxane; 30 m x 320 μm x 0.25 μm). The flame was supplied with air (450 ml/min) and hydrogen (40 ml/min). The carrier gas was helium at a pressure of 1.2 ml/min. The temperature of the detector as well as that of the injector was 250 °C. The oven’s initial temperature was 70 °C. Temperature first climbed by 30 °C/min until it reached 150 °C; it then climbed by 5 °C/min to reach a final temperature of 320 °C. The injection occurred in splitless mode (30 sec), and the GC analysis cycle lasted 62 min. The CHCs were identified using a GC-mass spectrometer (MS) (Autosystem XL coupled with a TurboMass 5.4; Perkin Elmer, Waltham, MA, USA); an electron ionization setting of 70 eV was employed. The parameters of the GC cycle were the same as those described above. The mass spectra were identified using known mass spectra from the literature (Lockey 1988; review in Blomquist & Bagnères, 2010).

The relative percentages of the CHCs were calculated by performing GC analysis on the individual specimens; the cuticular mixtures were identified by performing GC-MS analysis on pools of specimens representing the same species. Principal component analysis (PCA) and hierarchical cluster analysis were carried out using the relative percentages of CHCs. An analysis of variance (ANOVA) was performed to test for intraspecific differences between the sexes (SPSS v. 18).

The three phylogenetic methods yielded trees with congruent topologies that contained the same three major clades. Consequently, we have chosen to present only the BI tree (Fig. 1). The first clade was formed by all the specimens of O. miniata mauritanica. The second clade contained two well-supported monophyletic groups that corresponded to the two Acrotylus species. The third clade comprised the vast majority of the specimens representing the two Sphingonotus species as well as the specimens for S. carinatus. In the first clade (i.e., corresponding to O. m. mauritanica), four haplotypes were present that corresponded to four variable nucleotide sites. In the second clade, there were two groups. The first group contained a single haplotype (AA10) that was associated with A. partuelis; the second group contained six haplotypes (corresponding to eight variable nucleotide sites) that were associated with A. insubricus insubricus. In the third clade, there were three well-supported monophyletic groups that corresponded to the three species. The first group was composed of eight specimens divided into two haplotypes (corresponding to a single variable nucleotide site); this group was associated with the BOLD reference sequence for Sphingonotus lucasii (OBAL053-18). The second group was composed of nine specimens displaying a single haplotype that were associated with S. (Neosphingonotus) finotianus. The third group consisted of eight specimens divided into two haplotypes (corresponding to a single variable nucleotide site); this group was associated with S. carinatus.

Fig. 1. Phylogenetic tree built using COI gene sequences and Bayesian inference. The branch support values are Bayesian posterior probabilities. The reference sequences are labeled (Genbank or BOLD accession code followed by species name). The brackets on the right side of the tree indicate the species identity of the specimens.

Using the GC and GC-MS analyses, CHC profiles were obtained for the six study species (Fig. 2); they contained four CHC classes: n-alkanes, monomethylalkanes, dimethylalkanes, and trimethylalkanes. No unsaturated CHCs were seen (Supp. Table II).

Fig. 2. Total ion chromatograms (GC-MS) for Oedipoda miniata mauritanica (A), Acrotylus insubricus insubricus (B), Acrotylus patruelis (C), Sphingonotus azurescens (D), Sphingonotus finotianus (E), and Sphingoderus carinatus (F).

Within species, individuals displayed certain similarities in compound classes, although relative compound quantities differed (Fig. 3).

Fig. 3. Hierarchical cluster analysis of the CHC profiles of Oedipoda miniata mauritanica (A), Acrotylus insubricus insubricus (B), Sphingonotus azurescens (C), Sphingonotus finotianus (D), and Sphingoderus carinatus (E).

In Oedipoda miniata mauritanica and S. azurescens, n-alkanes were more common in females; in S. finotianus, S. carinatus, and Acrotylus insubricus insubricus, they were more common in males. Monomethylalkanes were also generally more common in females. However, in S. azurescens and S. carinatus, they were more common in males. There were only minor differences in dimethylalkanes between the sexes. That said, dimethylalkanes were far more prominent in females in A. i. insubricus and S. carinatus. Across all six species, trimethylalkanes were generally more common in males. The only exceptions were seen in S. carinatus (more common in females) and in A. i. insubricus (no sex-specific difference) (Fig. 4).

Fig. 4. Relative percentages of the different CHC classes observed in Oedipoda miniata mauritanica (Omin), Acrotylus insubricus insubricus (Ains), Acrotylus patruelis (Apat), Sphingonotus azurescens (Sazu), Sphingoderus carinatus (Scar), and Sphingonotus finotianus (Sfin).

In general, in Oedipoda miniata mauritanica, Acrotylus patruelis and A. insubricusinsubricus, the most prominent CHC classes were the monomethylalkanes and n-alkanes. The latter were most prominent in S. azurescens, S. finotianus, and S. carinatus.

Comparing relative CHC levels among species, n-alkanes were more common in S. azurescens; monomethylalkanes were more common in Acrotylus patruelis; and dimethylalkanes were more common in S. carinatus; and trimethylalkanes were more common in S. finotianus.

A PCA was carried out using 64 variables, which corresponded to the CHC profiles of all the specimens (i.e., the GC-MS date (Table I), and the identities of the six study species. Taken together, axis 1 and axis 2 explained 53.18% of the total variance (33.04% and 20.14%, respectively; Fig. 5). Three clusters formed largely along taxonomic lines: one with individuals from Oedipoda miniata mauritanica; another with individuals from both Acrotylus insubricus insubricus and A. patruelis; and a third with individuals from S. azurescens, S. finotianus, and S. carinatus. Furthermore, O. m. mauritanica had a very distinct profile, characterized by monomethylalkanes, dimethylalkanes, and trimethylalkanes with much shorter carbon chains. The clustering of S. azurescens, S. finotianus, and S. carinatus can be explained by the similarities in their CHCs classes, namely the mono- and dimethylalkanes. (Table I).

Fig. 5.Principal component analysis of the CHC profiles of Oedipoda miniata mauritanica (Omin), Acrotylus insubricus insubricus (Ains), Acrotylus patruelis (Apat), Sphingonotus azurescens (Sazu), Sphingoderus carinatus (Scar), and Sphingonotus finotianus (Sfin). The results for males (1) and females (2) are shown.

Table I. Carbon chain length range for the cuticular hydrocarbon classes seen in Oedipoda miniata mauritanica (Omin), Acrotylus insubricus insubricus (Ains), Acrotylus patruelis (Apat), Sphingonotus azurescens (Sazu), S. finotianus (Sfin), and Sphingoderus carinatus (Scar).

Sex-specific differences in CHCs profiles were seen in Oedipoda mauritanica and S. carinatus (ANOVA: p < 0.05) but not in S. azurescens, S. finotianus, Acrotylus insubricus insubricus, or A. patruelis.

The COI-based phylogenetic analysis generally confirmed our morphology-based classifications. Indeed, the sole member of the second clade in the phylogenetic tree was identified as Acrotylus patruelis. Thus, the difference between A. patruelis and A. insubricus was clear from the barcoding data.

Individuals that were morphologically identified as belonging to S. azurescens, S. finotianus, and S. carinatus occurred in three well-supported monophyletic groups in the third clade. Consequently, questions are raised about the S. azurescens (JQ513072) and S. lucasii (OBAL053-18) reference sequences—are those species assignments actually correct? Our specimens that grouped genetically with the S. lucasii reference sequence were morphologically classified as S. azurescens. The latter species is identified based on the following traits. It has a pronotum that is nearly smooth, weakly rough, and finely punctuated. The pronotum’s median carina is prominent in the prozone but reduced to a single line in the metazone. The species’ posterior femurs have a dark inner side with two clear bands. The upper carina of the posterior femur descends in its apical half. The posterior tibiae are yellowish white with a bluish tinge. The tegmina have a broad brown spot in their basal third and a narrow brown stripe in their middle. The wings have a broad brown fascia that varies in length; it starts at the middle of the anterior wing edge and closely follows the edge. Given that our specimens clearly displayed these traits, we assumed that they were indeed S. azurescens.

According to Chopard (1943), S. azurescens has been confused with a neighboring species, notably with S. lucasii.Louveaux et al. (2020) reiterated that confusion between these species may occur and that their distribution ranges need to be confirmed. The species appear to occur in different regions of North Africa (e.g., northern and southern Algeria). They are geographically close.

Moussi et al. (2018) stated that there was a match between the molecular and morphological classifications of many Oedipodinae species; however, Sphingonotus remained a problematic taxon in their phylogenetic analysis. These authors interpreted this result as showing that COI-based species delimitation remains difficult, at least for most Sphingonotini genera, perhaps because of nuclear copies of mitochondrial sequences (pseudogenes), even though DNA barcoding is a valuable tool for identifying species in these genera. They suggested additional genetic data may be required to achieve better levels of resolution. Here we assume our eight specimens as S. azurescens rather than as S. lucasii.

The length of the CHC carbon chains observed in this study are consistent with those seen in previous research on insects. According to Nelson & Blomquist (1995), this length ranges approximately from 21 to 50 carbons, and the shorter chain compounds are volatile. In general, the number of methylated compounds was high, and there was a remarkable absence of unsaturated products (alkenes). These results agree with those obtained for Locusta migratoria cinerascens (Genin et al., 1986).

The cluster analysis revealed species specific variation in CHC profiles. Some past studies (Nielsen et al., 1999; Buczkowski et al., 2005; Parkarsh et al., 2008; Bontonou et al., 2013) have found that variation in CHCs can be linked to factors such as age, diet, habitat, and diverse environmental characteristics, while other studies have seen no such correlation (Toolson, 1982; Vander Meer et al., 1989; Dahbi et al., 1996). Consequently, CHCs could contribute to creating a species-specific chemical signature.

The PCA results highlighted the profile similarities among S. azurescens, S. finotianus, and S. carinatus. The three species formed a single cluster that was distinct from the cluster formed by Oedipoda miniata mauritanica and from the cluster formed by Acrotylus insubricus insubricus and A. patruelis. These results show that, although there were similarities, the CHC profiles were nonetheless relatively species specific, notably in their number of compounds. The clusters arising from the chemical data generally corresponded to the clusters arising from the genetic data.

There was a significant sex-specific difference in relative CHC percentages in S. carinatus and Oedipoda miniata mauritanica but not in S. azurescens, S. finotianus,Acrotylus insubricus insubricus, or A. patruelis. However, it could be that more volatile compounds not analyzed here are also at play. Sexual selection could be involved (Thomas & Simmons, 2008), given that sexual dimorphism in CHC profiles was seen in only some species and that profile variation was seen within genera. An explanation could be that hydrocarbons are involved in reproduction, acting as sex pheromones as they do in other insect taxa, such as mosquitoes (Amira et al., 2013), Drosophila (Bontonou et al, 2013), Blattella germanica (Morakchi et al., 2010), and cricket Teleogryllus oceanicus (Thomas & Simmons, 2008).

In this study, we successfully explored intraspecific diversity within Acrotylus insubricus and used our combined taxonomic approach to distinguish between the two Acrotylus species. For Sphingonotus finotianus and S. carinatus, we increased the taxonomic specificity and showed that the species display clear genetic and chemical differences. We also helped flesh out the genetic and chemical taxonomy of the two other study species.

The genetic and chemical data we obtained have helped clarify differences within and between genera. They were also able to clearly distinguish between the six study species. Our findings indicate that such multifaceted analyses could make a significant contribution by complementing current morphotaxonomic techniques and thus help clarify species systematics.