Agabus alexandrae sp. n. from Morocco, with a molecular

phylogeny of the Western Mediterranean species of the A.

guttatus group (Coleoptera: Dytiscidae)

IGNACIO RIBERA, CARLES HERNANDO and PEDRO AGUILERA

A new species of the Agabus guttatus group, A. alexandrae sp. n., is described from the Moyen Atlas in Morocco. A phylogeny of most of the Western Mediterranean species of the group based on Cytochrome Oxydase I and 16S rRNA sequences (approximately 1,250 bp) shows that A. alexandrae sp. n. is sister to the remaining species of the group. The evolution- ary history of the species of the group is tentatively reconstructed, based on a Maximum Likelihood estimate of divergence time of approximately 2% per MY for the combined 16S and COI genes, similar to other estimates of evolutionary rates of the mitochondrial DNA of Coleoptera.

I. Ribera, Department of Entomology, The Natural History Museum, Cromwell Road, London SW7 5BD, U.K.

C. Hernando & P. Aguilera, Museu de Zoologia, P.O. Box 593, 08080 Barcelona, Spain.

Introduction

The genus Agabus, with ca. 180 species (Nilsson 2000), is one of the most diverse of the family Dytiscidae. It has a Holarctic distribution, with some isolated species in the Oriental and Ethiopian regions (Nilsson 1992; 2000). The phy- logenetic position of the genus within the tribe Agabini, and of the species relationships within the genus, are still largely unknown, partly due to the paucity of morphological characters suitable for a phylogenetic analysis. In its most recent redefinition, the genus Agabus itself could only be characterised by a single sinapomorphy, the con- tinuous bead on the front margin of the clypeus (Nilsson 2000).

Within the genus Agabus three subgenera are currently recognised, each of them with several species groups. The Agabus guttatus group as defined by Nilsson & Holmen (1995) is part of the subgenus Gaurodytes (species with parameters stylate, Nilsson 2000). It includes species with pronotum without anterior beading, most species with anterior row of punctures broadly interrupted

medially; posterolateral row of punctures on pronotum with a sublateral gap; metatibia with anteroventral row of punctures complete and almost continuous; and male mesotarsomere 3 without adhesive setae. Twenty-five species are currently recognised within this group, with a strict Palaearctic distribution centred in the Mediterranean (Nilsson 2000). Most species are rheofilous, living in mountain streams and springs, often in habitats with a reduced amount of water (Foster & Bilton 1997).

A couple of specimens of an undescribed species of the group were found in the lake Afenourir, in the Moroccan Moyen Atlas, in 1999. The place had been visited by the authors in 1997 and 1998, and was revisited again in spring 2000, but no further specimens of the species were found, despite intensive search (that resulted in a list of ca. 100 species of aquatic Coleoptera). In an attempt to establish the relationships of this new species a molecular phylogeny of most of the Western Mediterranean species of the group was constructed using the mitochondrial genes 16S

254

rRNA and Cytochrome Oxydase I (COI). The

evolutionary history of the lineage is also dis- cussed based on the present distribution of the

species and the estimated time of divergence among them.

Material and methods

Taxon sampling. - All the Western Mediterranean

species of the A. guttatus group were included in the phylogenetic analysis (Tab. 1), with the excep- tion of A. picotae Foster & Bilton 1997 (very closely related to A. heydeni), A. africanus Pederzani & Schizzerotto, 1998 (closely related to A. binotatus), A. cephalotes Reiche, 1861, and A. maderensis Wollaston, 1854 (Nilsson 2000). The

phylogenetic relationships among the species of the group are unknown, as well as the possible monophyly of the Western Mediterranean species.

For most of the species more than one specimen from different populations were included, in par- ticular for A. biguttatus (the most widespread species, and the most similar morphologically to the new species), for which ten specimens from south Spain, Morocco and the Canary Islands were included (Tab. 1). Outgroup sequences were obtained from Ribera et al. (2001) (with the

exception of Agabus aubei), and include several species of Agabini, of the genera Agabus, Ilybius, Ilybiosoma and Platambus (Tab. 1).

Within the A. guttatus group, the species con- sidered to be 'A. nitidus' by Milldn et al. (1992) was included as a separate entity. In Foster & Bilton (1997) and Nilsson (2000) A. nitidus is considered to be a synonymy of A. biguttatus, based mostly in the examination of northern and central European material. At present it is unknown weather 'A. nitidus' from SE Spain cor-

responds to the true A. nitidus, but due to its well characterised morphology and genetic differentia- tion (even in coexisting populations, note that one of the specimens of A. biguttatus included in the

analysis was collected in the same locality and date as 'A. nitidus') it was included as a separate species.

DNA extraction, PCR and sequencing. - Spe- cimens were collected in absolute ethanol, and some muscular tissue was used for DNA isolation

using a Phenol-Chlorophorm extraction as described in Vogler et al. (1993). Sequences of 16S rRNA were amplified in a single fragment of

ca. 510 bp, using primers 16Sa (5' ATGTTTTTGTTAAACAGGCG) for the 5' end of the gene, and 16Sb (5' CCGGTCTGAACTCA- GATCATGT) for the 3' end. A single fragment of ca. 720 bp of COI (from the middle of the region E3 to the COOH end, Lunt et al. 1996) was ampli- fied using the primers 'Jerry' (5' CAACATT- TATTTTGATTTTTTGG) for the 5' end of the

gene, and 'Pat' (5' TCCAATGCACTAATCTGC- CATATTA) for the 3' end (Simon et al. 1994). All

sequences generated in this study were deposited in GenBank (Acc. Nos AY039257-AY039277, Tab. 1).

The following cycling conditions were used: 1 to 2 min at 95°C; 30 seconds at 94°C, 30 seconds at 47-50°C (depending on the melting tempera- tures of the primer pair used), and 1-2 min at 72°C (repeated for 35 to 40 cycles); 10 min at 72°C.

Amplification products were purified using a GeneClean II kit (Bio 101, Inc.) Automated DNA

sequencing reagents were supplied by Perkin Elmer Applied BioSystems Ltd. (ABI PRISM Big Dye Terminator Cycle Sequencing Ready Reaction Kit). Sequencing reactions were purified by ethanol precipitation and were electrophoresed on an ABI3700 sequencer. In all cases both strands of the gene were sequenced, and sequenc- ing errors/ambiguities were edited in the

Sequencher 3.0 software package (Gene Codes

Corporation).

Phylogenetic analysis. - Sequences were not

length variable for the ingroup, with the exception of a deletion of a single 'T' in the 16S rRNA of

Agabus dilatatus in position 325. 16S rRNA

sequences of the outgroup differed in length only minimally, with a maximum difference of only four base pairs (see Results). Alignment therefore was performed manually, by maximising sequence similarities. Phylogenetic analysis was

performed with PAUP4.0 (Swofford 1999), using parsimony procedures for tree reconstruction, and

using gaps as a missing character. Constraint trees for determining Bremer Support values (Bremer 1994) and partitioned Bremer Support values were generated with Treerot (Sorenson 1996). The

significance of the Incongruence Length Difference (ILD) (Farris et al. 1994) was assessed with the Partition Homogeneity Test as imple- mented in PAUP (using a heuristic search with 100 random-addition replicates).

To estimate divergence times branch lengths for

255

Table 1. Material included in the molecular phylogeny, with locality data, haplotype and GenBank accession num- bers.

the ingroup only were fitted using Maximum Likelihood (ML) as implemented in PAUP. Optimum ML models were selected using

Modeltest 3.04 (Posada & Crandall 1998). To estimate node ages we fitted ML branch lengths assuming a molecular clock and compared the

256

likelihood to that obtained assuming no clock (Felsenstein 1981). As the ML ratio was not sig- nificant (see Results), an ultrametric tree was esti- mated enforcing a molecular clock.

To calibrate the branch lengths two of the most extreme published estimations of the evolutionary rate of insect mitochondrial DNA were used. The faster rate estimation was the standard approxi- mately 2% divergence per million years (MY) for mitochondrial DNA (Brower 1994), correspon- ding to a base rate (per branch) of 0.01 substitu- tions/site/R4Y. The slower rate estimation was that of Gomez-Zurita et al. (2000) for the gene 16S rRNA, estimated for the split among northern Moroccan and south Iberian species of the genus Timarcha (Coleoptera, Chrysomelidae), with a rate of 0.0038 substitutions/site/MY (0.76% diver-

gence per MY).

Agabus alexandrae sp. n.

(Fig. 2, 4)

Description. - Length 9.5 mm (male) 9.7 mm (female), maximum width 5.3 mm (male) 5.4 mm (female). Body oval, vey convex, black, head with two rufous spots. Antenna brown, last segments darkened at apex. Palpi black, apex of last seg- ment brown. Legs black. Elytra with weak medio- lateral and subapical yellow spots.

Body surface very shiny; surface of head and

pronotum finely rugose; elytra with polygonal, weakly impressed irregular meshes, more

impressed apically, most of meshes with interior

punctures. Elytral apex with sparse, irregular punctures. Three well impressed, irregular series of setiferous punctures in each elytron. Anterior and posterior submarginal transverse rows of

punctures of pronotum widely interrupted in the middle, irregular. Metacoxal plates coarsely retic- ulated, with irregular polygonal meshes. Abdominal stemites with progressively transverse striae towards apex; stemites 2 to 4 with a medial row of setiferous punctures.

Clypeus finely but continuously bordered. Lateral beads of pronotum of uniform width; hind

angle slightly obtuse. Elytral margins finely bor- dered. Prostemal apophysis lanceolate, pointed, with continuous lateral beads. Metastemal wings broad. Apical metatibial spur shorter than first metatarsomere.

First two tarsomeres of anterior and middle legs

of male dilated, with adhesive setae. Anterior claws elongated, falcifom; interior claw slightly longer, with a medial tooth. Last abdominal ster- nite with strongly impressed longitudinal striae. Penis regularly curved in lateral view, asymmetri- cal in dorsal view (Fig. 2, 4).

Etymology. - Named after Alexandra 'Ali' Cieslak, for her enthusiasm to collect Agabus and Hydroporus.

Type locality. - Lac Afenourir, Moyen Atlas, Morocco, 33°17'12.4"N 5°15'09.8"W, 1,800 m a.s.l.

Type material. - Holotype a (Natural History Museum, London): '8 MOROCCO 9.4.1999 / Azrou: Lac Afenourir / I. Ribera leg.'. Paratype (IR Coll.), 19, same locality and data as holotype. Genomic DNA extraction of both specimens stored in the collection of the Molecular Systematic Laboratory, Department of Entomology, Natural History Museum, under reference numbers IR212 (holotype 0") and IR213 (paratype 9).

Ecology. - The two known specimens were col- lected in a temporary, well vegetated shallow

pond close to the main lake Afenourir. This sug- gests the possibility that the ancestral habitat of the A. guttatus group could be stagnant rather than

running water, with an early transition to running waters (see the phylogenetic analysis below). However, these two specimens could also have been vagrants from some of the nearby streams - a more typical habitat of the species of the group. More captures will be necessary to establish with

certainty the habitat of A. alexandrae sp. n.

Remarks. - Among the Western Mediterranean

species, the external morphology of A. alexandrae

sp. n. seems most similar to that of A. bigutattus, from which can be differentiated by its more reg- ular oval shape, with the maximum width in a more rear position. The edeagus is however clear-

ly different, more regularly curved in lateral view, and less asymmetrical in dorsal view (Fig. 1-4).

The shape of the aedeagus, as well as the gen- eral body shape and colour, allows a clear differ- entiation from the examined Central and Eastern Mediterranean species of the group (A. balcanicu.s

Hlisnikovsky, 1955; A. brandti Harold, 1880; A.

freudi Gueorguiev, 1975; A. glazunovi (Zaitzev, 1953); A. lobonyx Guignot, 1952; A. ommani Zaitzev, 1908 and A. svenhedini (Falkenström, 1932) could not be examined, although because of the characters given in the original descriptions and the distribution of the species - most of them from the Eastern Palaearctic - any confusion is most unlikely).

257

Figures 1-4. Aedeagus of Agabus biguttatus (1, 3) and A. alexandrae sp. n. (2, 4), lateral and dorsal views, traced from photographs. Scale bar 0.5 mm.

Phylogeny of the Western Mediterranean

species of the A. guttatus group

16S rRNA. - All the ingroup sequences of the 16S rRNA fragment had the same length, 512 bp. Among the outgroups, the longest were I. albar- racinensis and L chalconatus (513 bp), and the shortest L lugens (508).

All specimens of A. biguttatus had the same

haplotype, with the exception of the ones from the

Canary Islands (Gomera and Tenerife, Tab. 1). The two specimens of A. binotatus, A. heydeni and A. alexandrae had the same haplotype respective- ly (Tab. 1).

The aligned sequences had 515 characters, of which 418 were constant and 63 parsimony informative. Minimum distance within the

ingroup corresponded to 'A. nitidus' - A. bigutta- tus 1 (see Tab. 1 for the haplotype codes) (uncor-

rected 'p' distance 0.008, corresponding to 4 bp). The maximum ingroup distance was that of A.

heydeni - A. biguttatus 2 (p = 0.025, correspon- ding to 13 bp). The maximum overall distance was between P. maculatus and L angustior (p = 0.11, corrresponding to 54 bp).

A Tree-Bisection-Reconnection (TBR) search of 1,000 replicas produced 32 trees of a consisten-

cy index (CI) of 0.62 and a length of 181 steps. The topology of the strict consensus tree was com-

patible with that of the combined analysis (see below), with the only difference of the sister rela-

tionship between A. heydeni and A. binotatus. The

position of A. aubei, A. didymus and A. bipustula- tus among the outgroups, and A. guttatus, A. dilatatus and A. alexandrae within the ingroup, was unresolved (i.e. forming a polytomy at the base of the genus Agabus and the A. guttatus group respectively).

258

Figure 5. Phylogram of one of the two most parsimonious trees of the combined (16S rRNA + COI) analysis (in the strict consensus tree node 15 is collapsed). In square brackets, node numbers. Bootstrap support values (> 50%) on top left of the nodes (see Tab. 2 for Bremer support values).

259

Cytochrome Oxydase L - All the studied se- quences had 723 base pairs, with no length varia- tion. Of them, 494 were constant and 171 parsi- mony informative.

Both specimens of A. alexandrae and A. bino- tatus had the same haplotype, and there were four different haplotypes within A. biguttatus (Tab. l ).

The minimum distance within the ingroup cor- responded to A. biguttatus 1 and 2 (p = 0.0014, corresponding to 1 bp). The maximum distance within the ingroup corresponded to A. dilatatus - A. biguttatus 2 (p = 0.07, 54 bp). The maximum distance overall corresponded to P. maculatus - L

angustior (p = 0.15, 112 bp). A TBR search of 1,000 replicas resulted in a

single tree of 628 steps with a CI = 0.51. The

topology of the ingroup was the same as that of the combined analysis (i.e. with A. guttatus sister to A. binotatus), but the relationships among the outgroups were slightly different (mainly the posi- tion of the other species of Agabus).

Combined analysis. - A TBR search with 1,000 replicas on the combined data set (all characters equally weighted) resulted in two trees of 815 steps and a CI = 0.53 (Fig. 5). The Partition

Table 2. Bremer Partitioned Support values for the nodes of one of the two combined ( 16S rRNA + COI) trees. See Fig. 5 for the number of the nodes. Values refer to the relative Bremer Support Value (i.e. the dif- ference between the value obtained for the constrained node and the value obtained for the unconstrained tree, for each gene separately and for the combined analysis). Unresolved nodes in the strict consensus tree have a Bremer Support Value of 0 (node 15).

Homogeneity Test was not significant (p = 0.72, 100 random-addition replicates), indicating a high congruence between both genes.

In the combined tree all included genera are

monophyletic, with the possible exception of Agabus (the position of A. didymus is unresolved). The species formerly included in the Agabus chal- conatus and A. erichsoni groups (L subtilis, L albarracinensis and L chalconatus), and trans- ferred to flybius by Nilsson (2000), were placed as sister to L angustior.

The species of the A. guttatus group were mono- phyletic and closest to A. bipustulatus. Within the group, A. alexandrae was basal, and sister to two main clades: that formed by the species with a denticle in the apex of the aedeagus (A. guttatus complex: A. dilatatus, A. binotatus, A. heydeni and A. guttatus, see Foster & Bilton 1997), and that formed by the A. biguttatus complex (A. bigutta- tus sensu lato and A. `nitidus').

When gaps were considered as a 5th character, the search resulted in 4 trees of 830 steps and a CI of 0.53. The topology was identical, with the ex- ception of the collapse of nodes 8 and 9 (Fig. 5) (i.e., the position of A. dilatatus was unresolved, resulting in a polytomy at the base of the A. gutta- tus gr.).

Bootstrap values were very high for most of the nodes (Fig. 5). Bremer Support values were equal- ly high for most of the nodes (Tab. 2). Partitioned Bremer Support values were negative (i.e. indicat- ing that the gene in question did not supported the final topology) for the sister relationship between A. guttatus and A. binotatus (node 5); the position of A. dilatatus (node 9) and for the position of A. didymus (node 15). The highest support values (other that for haplotypes of the same species) were those of the genera Ilybiosoma (node 13) and Ilybius (node 12), and the A. guttatus gr. (node 18). The genus Agabus (node 15) had no support, due to the inconsistency in the position of A. didymus.

Rate of evolution. - For the combined data set

(ingroup only), the optimal ML model as selected by Modeltest used the observed base frequencies, an estimated substitution model and equal rates of variation for all sites. Estimations for both genes independently gave similar models, with the dif- ference of unequal rates for COI. The maximum likelihood of the single combined tree under the

assumption of a molecular clock was not signifi- cantly different to that of the same tree and model

260

Figure 6. Ultrametric tree obtained with a Maximum Likelihood model on the combined data set. Numbers above branches refer to branch lengths. The scale shows an estimated rate of evolution for the combined data set of approx- imately 0.01 substitutions/site/MY (see text).

without enforcing a molecular clock (-logL (no clock) = 2597.6; -LogL (clock) = 2603.9; 2 x dif- ference in LogL has a chi-square distribution with

9 degrees of freedom, p = 0.18). Both when the model estimated for the combined data was

applied to the genes separately, and when the

261

models estimated for each gene separately were

applied to the combined data set, there were no

significant differences between the likelihood

enforcing and not enforcing a molecular clock. A

single ultrametric tree was thus produced using the ML estimated model for the combined data set

(Fig. 6). The use of the standard 2% divergence per mil-

lion years (MY) for insect mitochondrial DNA (Brower 1994) gives a minimum estimated age of the A. guttatus group of 3.6 MY (Fig. 6). The ori-

gin of the species (with the exception of A. alexan- drae) would be much younger, 2.3 MY for A. nitidus, and ca. 1.7 MY for A. biguttatus, A. bino- tatus, A. guttatus and A. heydeni. The use of the much slower estimate of Gomez-Zurita et al. (2000) for the genus Timarcha gives an age of the

divergence of A. alexandrae of 9.4 MY, and an approximate age of the most recent species of ca. 4.2 MY. However, this latter estimate was based on the gen 16S rRNA, which has a much lower evolutionary rate than Cytochrome Oxydase I. If

only the gene 16S is taken into account, the esti- mated branch length of A. alexandrae is 0.013, which would correspond to 3.4 MY - much closer to the 3.6 MY obtained with the standard rate.

Discussion

Phylogenetic analysis. - The phylogenetic rela-

tionships of some of the Western Mediterranean

species not included in the analysis can be deduced from their morphology. Agabus cephalotes has an apical tooth in the aedeagus (Foster & Bilton 1997), and it is likely to be included in the A. guttatus complex. Similarly, A. africanus has been hypothesised to be sister to A. binotatus (Pederzani & Schizzeroto 1998). The

position of A. maderensis is less clear, although it seems likely to be a member of the A. biguttatus complex, probably close to A. biguttatus.

The south Portugese A. picotae is undoubtedly closely related to A. heydeni (Foster & Bilton 1997), a species with a scattered distribution in Iberia and Morocco, with main populations in central Spain, north and central Portugal and Sierra Nevada (Rico et al. 1990). Until more molecular data is available it is not possible to ascertain their taxonomic status.

The specific identity of A. 'nitidus' is clear, both for the age of the separation with A. biguttatus (similar or older to that of other well defined

species, Fig. 6) and for the fact that both coexist

(although it would be necessary to study more

specimens to be certain). Similarly, differences between the populations of A. biguttatus from Gomera and Tenerife and that of Morocco and the Iberian Peninsula seem to be deep enough to war- rant a specific status, to which the name A. con-

sanguineus Woollaston, 1864 could be applied. According to Machado (1987) two forms can be found in the Canary Islands, A. nitidus and A.

biguttatus. More data is necessary to assess the

possible differences between the two, and if any, to which would correspond A. consanguineus.

Evolutionary history. -According to the estimated rates, the split between A. alexandrae and the

remaing species of the group would correspond to ca. 3.5 MY. This is younger that the age of the end of Messinian crisis, with the separation of the Iberian Peninsula and north Morocco after the re-

opening of the Gibraltar straits (5.33 MY, Krijgsman et al. 1999) (which was used by Gomez-Zurita et al. 2000 as a calibration point for their estimation). If we accept this estimation, all

speciation events across the Mediterranean (and in islands) would be the result of colonisation, either from Europe or North Africa (depending on the

geographical origin of the group). Similarly, the separation of the populations of A. heydeni from Morocco and Central Spain would have taken

place approximately 100,000 years BP, even if the

species can be considered a mountain specialist and has so far never been recorded to fly.

Similarly, our data suggest that the large range of A. biguttatus (the most extense of the group) is the result of a recent expansion. Differences

among COI haplotypes were small (with the

exception of the populations in the Canary Islands, with an estimated age of isolation of no less than 1.6 MY), but had some geographical structure: all Iberian Specimens had the same hap- lotype, and all Moroccan specimens another one with the exception of A. biguttatu.s 8, with a unique haplotype (Tab. 1). The topology of the tree suggests the paraphyly of the Moroccan pop- ulations, although the small amount of variation (between 1 and 3 base pairs) does not allow any strong conclusion.

According to our evolutionary hypotheses, A. binotatus would have a continental origin, with a

posterior colonisation of the Mediterranean islands (the present distribution includes Corsica,

262

Sardinia, Sicily and mainland Italy, Franciscolo 1979). The species is too young to have been orig- inated as a Corsican or Sardinian endemic by a vicariant split. The last connection of Corsica with mainland was at the end of the Messinian crisis 5.33 MY ago, as seen above, and the geological formation of the islands dates approximately 12- 15 MY (see e.g. Steinfartz et al. 2000 and refer- ences therein).

Acknowledgements We thank Hans Fery (Berlin), S. Ericsson and Anders Nilsson (Umea) and Keith Miller (Cyprus) for providing some specimens, and Alfried Vogler, Txus G6mez- Zurita and Andr6s Millan for comments and sugges- tions. Grant support was through NERC GR9/4735 to Alfried Vogler (NHM). IR is a Leverhulme Special Research Fellow.

Bremer, K. (1994) Branch support and tree stability. Cladistics 10: 295-304.

Brower, A. V. Z. (1994) Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochon- drial DNA evolution. Proceedings of the National Academy of Sciences, USA 91: 6491-6495.

Farris, J. S., Källersjö, M., Kluge, A. G. & Bult, C. (1994) Testing significance of incongruence. Cladistics 10: 315-320.

Felsenstein, J. (1981) Evolutionary trees from DNA se- quences : a maximum likelihood approach. Journal of Molecular Evolution 17: 368-376.

Foster, G. N. & Bilton, D. T. (1997) A new species of Agabus from south-west Portugal (Coleoptera: Dytiscidae). Koleopterologische Rundschau 67: 113- 118.

Franciscolo, M. E. (1979) Coleoptera Haliplidae, Hygrobiidae, Gyrinidae, Dytiscidae. Fauna d'Italia 14: 804 pp. Edizioni Calderini, Bologna.

Gómez-Zurita, J., Juan, C. & Petitpierre, E. (2000) The evolutionary history of the genus Timarcha (Coleop- tera, Chrysomelidae) inferred from Mitochondrial COII gene and partial 16S rDNA sequences Molec- ular Phylogenetics and Evolution 4: 304-317.

Krijgsman, W., Hilgen, F. J., Raffi, I., Sierro, F. J. & Wilson, D. S. (1999) Chronology, causes and progres- sion of the Messinian salinity crisis. Nature 400: 652- 655.

Lund, D. H., Zhang, D. X., Szymura, J. M. & Hewitt, G. M. (1996) The insect cytochrome oxidase I gene: evolutionary patterns and conserved primers for phy- logenetic studies.InsectMolecularBiology 5 : 153-165.

Machado, A. (1987) Los ditiscidos de las Islas Canarias (Coleoptera, Dytiscidae). Instituto de Estudios Canarios, La Laguna.

Millán, A., Velasco, J. & Soler, A. G. (1992) Los coleópteros Hydradephaga de la cuenca del rio Segura (SE de la Peninsula Ibérica). Aspectos faunís- ticos más relevantes (Coleoptera). Anales de Bio- logía, Universidad de Murcia 18: 39-45.

Nilsson, A. N. (1992) A revision of Afrotropical Agabus Leach (Coleoptera, Dytiscidae), and the evolution of tropicoalpine super specialists. Systematic Entomol- ogy 17: 155-179.

Nilsson, A. N. (2000) A new view on the generic classi- fication of the Agabus-group of genera of the Agabini, aimed at solving the problem with a para- phyletic Agabus (Coleoptera: Dytiscidae). Koleopte- rologische Rundschau 70: 17-36.

Nilsson, A. N. & Holmen, M. (1995) The aquatic Adephaga (Coleoptera) of Fennoscandia and Den- mark. II. Dytiscidae. Fauna entomologica scandina- vica 32: 188 pp. E. J. Brill, Leiden.

Pederzani, F. & Schizzeroto, A. (1998) Description of Agabus (Dichonectes) africanus n. sp. from north- west Tunisia and notes on the cohabiting species of Hydradephaga (Coleoptera Haliplidae, Gyrinidae & Dytiscidae). Atti dell'Accademia Roveretana degli Agiati (7) 8B: 87-95.

Posada, D. & Crandall, K. A. (1998) Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817-818.

Ribera, I., Barraclough, T. G. & Vogler, A. P. (2001) The effect of habitat type on speciation rates and range movements in aquatic beetles: inferences from spe- cies-level phylogenies. Molecular Ecology 10: 721- 735.

Rico, E., Pérez, L. C. & Montes, C. (1990) Lista faunís- tica y bibliográfica de los Hydradephaga (Coleop- tera : Haliplidae, Hygrobiidae, Gyrinidae, Noteridae, Dytiscidae) de la Peninsula Ibérica e islas Baleares. Asociación Espanola de Limnologia, Listas de la Flora y Fauna de las Aguas Continentales de la Pen- insula Ibérica n° 7, Madrid.

Simon, C., Frati, F., Beckenbach, A. T., Crespi, B., Liu, H. & Flook, P. (1994) Evolution, weighting, and phy- logenetic utility of mitochondrial gene sequences and a compilation of conserved polymerase chain reaction primers. Annals of the Entomological Society of Amer- ica 87: 651-701.

Sorenson, M. D. (1996) TreeRot. University of Michi- gan, Ann Arbor.

Steinfartz, S., Veith, M. & Tautz, D. (2000) Mitochondrial sequence analysis of Salamandra taxa suggests old splits of major lineages and postglacial recolonizations of Central Europe from distinct source populations of Salamandra salamandra. Molecular Ecology 9: 397-410.

Swofford, D. L. (1999) PAUP*: Phylogenetic Analysis using Parsimony. Version 4.0b2a. Sinauer Associates, Sunderland, Massachusetts.

Vogler, A. P., DeSalle, R., Assmann, T., Knisley, C. B. & Schultz, T. D. (1993) Molecular population genetics of the endangered tiger beetle, Cicindela dorsalis (Coleoptera: Cicindelidae). Annals of the Entomo- logical Society of America 86: 142-152.