Barriers and retreats

The geographic distribution of the obtained genetic clusters strongly supports significant impact of orographic structures as dispersal barriers. The extant distribution of Buthus scorpions mostly does not exceed elevations higher than 2500 m asl [23], and individuals show low dispersal ability. Thus, high mountain ranges likely acted (and act) as strong barriers to gene flow resulting in genetically distinct clusters found within the Atlas Mountains. This is reflected in our data by fine structured genetic lineages of the sampled Buthus individuals. A similar pattern was observed for the same taxonomic group using various molecular markers such as allozymes, nuclear and mitochondrial DNA all of which showed significant genetic differentiation patterns even among local populations [24], [25]. Similar results were found for other organisms with low to moderate mobility [7]. These extremely sedentary species apparently diverged into several narrowly distributed genetic clusters, restricted to the slopes of mountains and river valleys (e.g. the Ourika and Draa). These lineages are found at different altitudes and might be adapted to various ecological niches. We assume that all actual distribution ranges of Buthus lineages also include their last centre of differentiation. Thus, our study area may comprise 12 allopatric refugia during the last ice age revealing a much more fine grained biogeographical sub-structure than previously observed in the Maghreb e.g. [7]–[9]. Consequently, the biogeographical structuring of Buthus in Morocco is much more complex than the structures observed in other Mediterranean sub-centres [26], most probably as a consequence of strong gene flow barriers represented by high mountain ranges and a more constant climatic situation in North Africa compared to the northern part of the Mediterranean. The combination of our data with previously analysed Buthus individuals (Figure 1c) underlines this superposition of the Maghreb region as a speciation and differentiation centre: The European Buthus are polyphyletic, but restricted to one branch of the entire tree and make the North African Buthus paraphyletic. This may be indicative of multiple colonisation events from North Africa to Europe.

The trigger of genetic differentiation

This distribution of genetic lineages supports an allopatric differentiation scenario as opposed to sympatric speciation processes [27]: The lineages originating from the oldest vicariance events are most strongly differentiated from each other. Perhaps this long period of time elapsed since the onset of differentiation allowed for the evolution of characteristics enabling sympatric occurrence of representatives of different genetic groups. Furthermore, the close geographic proximity among the well differentiated lineages G, H and I suggests some areas of range overlap among them. If reproductive barriers are in effect, species might even occur in sympatry without niche differentiation when resources are abundant. Thus, the time elapsed since the onset of the evolution of these lineages must have been sufficient that neither the inter-lineage competition for space nor the strong stenotopy of Buthus scorpions has been able to prevent the geographic intermixing of these lineages. However, differentiation among them still might not be sufficient for syntopic occurrences so that these groups exclude each other at each single locality. Such a scenario of competitive exclusion would reinforce a strictly geography-dependent distribution pattern of lineages.

Discordance between morphospecies and COI phylogeny

In contrast to the stringent geographic pattern, our findings are not congruent with the current taxonomic classification. The different species identified based on morphological characters often cluster within different genetic lineages which, in turn, often comprise more than one species. In addition, many specimens could not be identified due to incomplete or doubtful identification keys, and a lack of diagnostic traits. Hence, the current taxonomy of North African Buthus species is in need of revision. Additional diagnostic characters must be identified and new species keys be developed. However, to actually define the status of the detected genetic lineages, further morphological, behavioural and ecological studies are necessary. Besides, it is well known that gene trees and species trees may differ, e. g. due to shared ancestral polymorphisms, incomplete phylogenetic lineage sorting and introgression [28]. Evidence of recombination of mtDNA has been found in Buthus, which could also influence our results [29]. Taking our results as an exclusive basis for taxonomical conclusions would therefore doubtlessly be premature. In future studies, the addition of nuclear markers is indispensable to solve the question how prominent hybridization and introgression are among these Buthus lineages and species.

Sampling and identification

Most specimens were collected at daylight under stones. Overnight collections were facilitated by black light as all scorpions fluoresce in UV light due to a specific protein in their exoskeleton [30]. Collected specimens were stored in absolute ethanol until DNA extraction. All sampling sites are shown in Figure 2 and listed in Table 1 which also gives details on the sampling date and location. Species were identified with keys and additional species descriptions [10], [12], [31]–[33] based on typical morphological characters used for scorpion identification [cf. 32]–[33]. All species identifications were performed by the same author (IS).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Geographical distribution of all sampling sites and the 12 genetic groups and three subgroups around the Atlas Mountains in Morocco.

Numbers of sampling sites coincide with numbers given in Table 1. Letters of the groups correspond with Figure 1. The geographic coordinates of sampling sites are given in Table 1. Abbreviations: SV: Souss valley, DV: Draa valley, DaV: Dades valley, ZV: Ziz valley. Country border between Morocco and Algeria is indicated by a dotted line.

https://doi.org/10.1371/journal.pone.0029403.g002

PCR and sequencing

For all individuals, DNA was isolated from leg, caudal segment or telson muscle tissue using the Qiagen DNeasy kit with standard protocols for tissue samples. A 452 bp fragment of the mitochondrial Cytochrome Oxidase I (COI) gene was amplified using standard PCR procedures with the following primers: forward 5′ GGT CAA CAA ATC ATA AAG ATA TTG G 3′, reverse 5′ TAA ACT TCA GGG TGA CCA AAA AAT CA 3′ [34]. PCRs were performed in 20 µl volumes: 10 µl Mastermix (Thermozyme), 0.2 µl of each Primer (1 µM), 4.6 µl PCR grade water and 5 µl DNA template. The cycle programme comprised an activation step at 94°C for 4 min, followed by 40 cycles of 30 sec denaturation at 94°C, 30 sec annealing at 45°C and 1 min elongation at 72°C. Cycling was terminated by a final extension step at 72°C for 10 min. Amplicons were subsequently purified and sequenced in both directions on an automated sequencer (3730xl DNA Analyzer; Applied Biosystems, Carlsbad, CA, USA) at the University of Kiel. All sequences are deposited at the NCBI genbank (appendix SI).

Data analysis

We included 91 sequences of Buthus, each 452 bp in length, in our analyses. Four sequences of the Buthid genus Androctonus served as outgroup (one sequence from A. mauretanicus obtained from genbank: accession JF820097 and three sequences from Androctonus spec. (09-88-1; 09-89-1; 09-89-2)). Sequences were inspected and aligned using Geneious 5.0.3 [35].

General alignment statistics (diversity indices, etc.) were calculated using DNAsp v. 5.10 [36]. Estimates of population divergence were calculated as pairwise Fst under a Kimura2P substitution model in Arlequin v.3.5.1.2. [37]. Populations were defined a posteriori according to our phylogenetic results. Significance of Fst values was tested using 1000 permutations and a significance level of 0.05.

The best fitting model of sequence evolution for phylogenetic tree reconstruction was chosen using MrModeltest 2.3 [38] in PAUP 4.0 b10 [39] and was determined to be the GTR+I+G. Maximum Likelihood (ML) trees were constructed using MEGA version 5.01 [40]; 1000 bootstrap replicates were performed to obtain a statistical measure for branch support (Figure 1a). Additionally, we used the Bayesian algorithm implemented in MrBayes 3.1.2 [41] to obtain a second phylogenetic inference and evaluate the stability of our tree reconstruction (Figure 1b). Using the obtained substitution model we ran the Markov chain Monte Carlo algorithm with four chains and two independent runs for 20 million generations. Trees were sampled every 2,000 generations; the first 10% of generated trees were discarded as burn-in as recommended by Tracer v1.5 [42]. The remaining trees were used to generate a consensus tree.

In a second analysis, we included 84 sequences from Buthus scorpions obtained from genbank (see appendix SI) resulting in a total of 175 ingroup sequences. However, due to only partial overlap the alignment was reduced to 383 bp in length. Bayesian analyses of this dataset were performed as described above. While these additional analyses provide further insights into geographic structuring of genetic divergence, they do not help in evaluating species identity and monophyly (Figure 1c). When these sequences were submitted to Genbank, the vast majority of Moroccan Buthus species known today had not yet been described and most Moroccan Buthus were characterized as subspecies of B. occitanus (B. o. occitanus – Europe, B. o. mardochei – Morocco, B. o. tunetanus – Tunisia) [31].