of southern Iberian lineages of Triops mauritanicus
Michael Korn & Andy J. Green & Margarida Machado & Juan García-de-Lomas & Margarida Cristo & Luís Cancela da Fonseca & Dagmar Frisch & José L. Pérez-Bote & Anna K. Hundsdoerfer
Abstract We investigated the phylogeography of the main lineages in the tadpole shrimp Triops mauritanicus Ghigi in the south-western Iberian Peninsula, using mito- chondrial 12S and 16S rDNA sequences. Our results indicate that a fourth, hitherto unknown main phylogenetic lineage occurs in Iberia, so that in total, the species is divided into six distinct clades, comprising T. m. mauritanicus, T. m. simplex Ghigi, and four as yet unnamed lineages that appear to be endemic to Iberia. Percentages of sequence
divergence among the main clades in T. mauritanicus reach the range reported for recognized species in other notostracan lineages. A thorough morphological investigation also revealed that the differentiation among these lineages is higher than previously thought, and that populations of three of the main clades within T. mauritanicus can be reliably separated from each other and from the remaining lineages based on the morphology of adult males. The remaining clades also show a significant level of morphological
Electronic supplementary material The online version of this article (doi:10.1007/s13127-010-0026-y) contains supplementary material, which is available to authorized users.
M. Korn (*)
Limnological Institute, University of Konstanz, Mainaustr. 252,
Laboratório Marítimo da Guia/Centro de Oceanografia (FCUL), Av. N. Sra. do Cabo, 939,
2750-374 Cascais, Portugal
J. L. Pérez-Bote
Grupo de Investigación en Ecosistemas Acuáticos Continentales, Área de Zoología, Facultad de Ciencias,
Universidad de Extremadura,
06071 Badajoz, Spain
differentiation, but include a certain proportion of populations for which the additional application of molecular methods is needed for a reliable determination. The geographic distribu- tions of 12S haplotypes are indicative of frequent dispersal events and gene flow among populations belonging to the same main lineage, but give no evidence of recent migration events among different main lineages, suggesting that there is no gene flow among the latter. Our data thus suggest that the six main lineages within T. mauritanicus represent distinct species. We therefore describe the Iberian lineages as T. baeticus Korn n. sp., T. emeritensis Korn & Pérez-Bote n. sp., T. gadensis Korn & García-de-Lomas n. sp., and T. vicentinus Korn, Machado, Cristo & Cancela da Fonseca n. sp., and reinstate T. simplex Ghigi to full species status. Our data confirm the general, previously recognized pattern of a lower dispersal probability in gonochoric Triops taxa. However, we found evidence that passive dispersal in Triops may be further complicated by a strong habitat dependence of dispersal probability, mediated by prevailing dispersal vectors.
Keywords Phylogeography .Gene diversity .Passive dispersal .Dispersal probability .Gene flow .Waterbird
Low levels of morphological differentiation in many passively dispersed aquatic invertebrates had led to the assumption that these taxa had a cosmopolitan distribution and that there was little restriction to frequent long-distance dispersal (Bohonak and Jenkins 2003). The application of molecular tools has greatly changed our knowledge on these taxa, and led to the general conclusion that many zooplankton species have small geographic distributions and often show a high level of genetic substructure (e.g. Colbourne et al. 2006; Ishida and Taylor 2007), indicating that high potential for dispersal does not necessarily translate to high effective dispersal rates or gene flow (Bohonak and Jenkins 2003). Similar cryptic diversification has also been reported in large branchiopods (e.g. Korn et al. 2006; Korn and Hundsdoerfer 2006; Sassaman et al.
1997). The factors that may interact to uncouple dispersal from gene flow are summarized in the Monopolization Hypothesis formulated by De Meester et al. (2002). This hypothesis suggests that for many freshwater organisms, the impact of new immigrants was reduced by a numerical effect (high population growth rates and large resting propagule bank) and a fitness effect (local adaptation of residents), leading to the monopolization of resources by first colonizers, enhancing priority effects and reducing gene flow, so that neighbouring populations may common- ly show pronounced genetic differentiation despite high dispersal capacity. Bohonak and Jenkins (2003) argue that
the Monopolization Hypothesis may not be generally applicable to the majority of freshwater invertebrates, and stress that generalizations about overland dispersal in freshwater taxa are not valid, and that specific information is needed for each taxon. The case of the Notostraca supports this argumentation, for example, because indirect evidence suggests that closely related species in this group show different dispersal probabilities (Korn et al. 2006; Sassaman et al. 1997: Fig. 3). These differences appear to be linked to reproductive modes, and gonochoric taxa (i.e. those that have an obligately outcrossing mode of repro- duction, with separate male and female individuals) have lower inferred dispersal probabilities.
The Notostraca comprise two genera with worldwide distributions, Triops Schrank, 1803, and Lepidurus Leach,
1819. Both occur almost exclusively in temporary bodies of water and can even inhabit ponds that remain dry for several years, as they are capable of enduring prolonged dry phases via resting eggs (e.g. Fryer 1988; Longhurst 1955). Among the European regions inhabited by Triops, the Iberian Peninsula is of special interest to studies on dispersal abilities and phylogeography, since several highly divergent phylogenetic lineages of possibly subspecific status have been recorded in a rather small geographic region in south-west Iberia (Korn et al. 2006). The predominant species of Triops in the Iberian Peninsula is T. mauritanicus. Originally established as a species by Ghigi (1921) it was later treated as a subspecies of T. cancriformis (Longhurst 1955), but has been returned to full species status (Korn et al. 2006). The northern African populations of the former T. c. simplex (originally described as a separate species; Ghigi 1921) are presently recognized as a subspecies of T. mauritanicus (Korn et al. 2006).
In this study, we use 12S and 16S rDNA sequences to investigate the phylogeography of the main lineages within Triops mauritanicus in the south-western Iberian Peninsula, and to infer dispersal abilities in these gonochoric taxa. In addition, we conduct a thorough morphological investiga- tion. The different datasets on genetic divergence, phylo- geography, inferred patterns of gene flow and morphological diversification are used to re-evaluate the taxonomy of the group.
Material and methods
We attempted to acquire as many different samples of Triops mauritanicus from southern Iberia as possible (for locality data, see Appendix section, Table A1). We used both wild-caught specimens and specimens raised in the laboratory from eggs from sediments. Samples were
preserved in 70–99.8% ethanol until extraction. Tissue vouchers have been deposited in the tissue collection of the Museum of Zoology (Museum für Tierkunde), Senckenberg Dresden, Germany, under the MTD-TW numbers listed in Table A1. Voucher specimens from the morphological analyses have been deposited in the invertebrate collection of the same museum, under the numbers MTD Crus 3046–
2765–2770, 2775–2778, 2781–2787 and 2792–2801 were included in morphological analyses (these are samples from Korn et al. 2006; MTD numbers originally referred to populations but were later partially redistributed in order to store individual specimens separately)]. Sequences are available from GenBank under accession numbers FN691389–FN691428 (12S) and FN689861–FN689867 (16S). Earlier sequences of T. mauritanicus and T. cancriformis were retrieved from GenBank and also included in the phylogenetic analyses (AM183829– AM183832, AM183836 – AM183840, AM183842, AM183854–AM183861, AM183867; samples listed in Tables 1, A1).
Table 1 Overview of sequences retrieved from GenBank, including short names used in present study for haplotypes of Triops c. cancriformis
Taxon Accession no. Gene Haplotype
T. c. cancriformis AB084514 12S T.c.c. 4
T. c. cancriformis AY159564 12S T.c.c. 3
T. c. cancriformis DQ369308 12S T.c.c. 5
T. c. cancriformis AY159575 16S T.c.c. 4a T. c. cancriformis AY159577 16S T.c.c. 6a T. c. cancriformis AB084514 16S T.c.c. 8
T. longicaudatus AY639934 12S T. granarius AY115602 12S L. a. apus AF494483 12S L. a. lubbocki AY159567 12S L. arcticus AY159569 12S L. lemmoni AY115604 12S T. longicaudatus AY639934 16S T. granarius AY115612 16S L. a. apus AY159584 16S L. a. lubbocki AY159583 16S L. arcticus AY159585 16S L. lemmoni AY115614 16S
aOur 12S haplotype T.c.c. 4 corresponds to H1 in Mantovani et al. (2004), so that short names in Fig. 2b–c for the dataset with 12S and 16S sequences combined are T.c.c. 4 4 and T.c.c. 4 6 for Mantovani et al.’s (2004) specimens Tcsi2, AY159575 and Tcsa2, AY159577, respectively
Determination of specimens
The characters given by Korn et al. (2006) were used to assign samples of Triops to species. All specimens from southern Iberia had large furcal spines, and thus could be determined unambiguously as T. mauritanicus. The present study’s ‘S. Iberian’ lineage within T. mauritanicus corresponds to the ‘S. Spanish’ haplotype group in Korn et al. (2006). Eurasian and North African populations of T. cancriformis are referred to as T. c. cancriformis here, to avoid confusion with certain populations in southern Africa (Hamer and Rayner 1995) whose actual status (subspecies of T. cancriformis?) and species affiliation remain to be investigated.
DNA extraction, PCR amplification and sequencing
DNA methods followed Korn et al. (2006), with the exception that PCR products were sequenced on an ABI
3130 sequencer (Applied Biosystems) at the Museum of
Zoology, Senckenberg Dresden.
Sequence alignment, nucleotide composition, and substitution patterns
Computerized alignments (obtained with Clustal W in the program BioEdit; Hall 1999) were modified by hand using BioEdit (alignments available from http://purl.org/phylo/ treebase/phylows/study/TB2:S10349). In both, the 12S and the 16S datasets, four Lepidurus and two Triops sequences (T. longicaudatus and T. granarius) were included as outgroups (Table 1). Nucleotide composition, substitution frequencies, pairwise transition/transversion frequencies, and pairwise distances (uncorrected p-distances) were calculated with PAUP* 4.0b10 (Swofford 2003). To enable an assessment of the overall range of sequence divergence found among the sublineages of the ingroup (T. mauritanicus and T. c. cancriformis), we compared the mean genetic distances between all T. mauritanicus lineages and between these lineages and T. c. cancriformis (calculated with MS-Excel). MEGA version 3.1 (Kumar et al. 2001) was used to illustrate parsimony-informative characters and singletons. The pro- gram ForCon 1.0 (Raes and Van de Peer 1998) was used to convert input files between formats. To assess saturation effects in this dataset, pairwise comparisons of transitional and transversional changes were plotted against pairwise distances in DAMBE version 4.2.13 (Xia and Xie 2001; for all distance correction methods implemented in DAMBE, we consistently found that the data were not saturated).
The 428 12S ingroup sequences obtained were collapsed to
53 haplotypes (Tables 1, A1), 48 of which were detected
within Triops mauritanicus. A second dataset comprised 26 ingroup 16S haplotypes (18 within T. mauritanicus; collapsed from 119 ingroup sequences). A third dataset used to investigate relationships among the ingroup lineages consisted of combined 12S and 16S sequences of those samples for which both gene fragments were available. This combined 12S and 16S dataset comprised
38 mitotypes (29 within T. mauritanicus). Data analysis for all three datasets was performed using maximum parsimony (MP; settings gapmode=new; add=cl) and maximum likeli- hood (ML) as implemented in PAUP* 4.0b10 (Swofford
2003). Additional ML searches performed with the program RAxML (Stamatakis et al. 2008) consistently resulted in tree topologies identical (with respect to the relationships among the main ingroup lineages) to those of ML phylograms obtained with PAUP*. The best evolutionary models for the data were selected using the program Modeltest (Posada and Crandall 1998; best-fit models were: TVM+G for 12S, HKY+I+G for 16S, and GTR+G for the combined dataset, selected by AIC; parameter values can be obtained from the first author upon request). As a measure of branch support, bootstrap values were calculated with MP in PAUP* (settings nreps = 1,000, maxtree = 10,000), and with ML in GARLI (version 0.95; Zwickl 2006; setting bootstrapreps =
100). Bayesian analysis was additionally applied to the combined 12S and 16S mitotype dataset using MRBAYES
3.1.2 (Huelsenbeck and Ronquist 2001); the settings were four runs with six chains of 5,000,000 generations, sampling every 500, and a burn-in of 1,000. The analysis was partitioned by gene (evolutionary models as specified above, but parameter values were estimated: no priors).
Geographic distribution of Iberian lineages
The geographic distribution of the Iberian lineages within T. mauritanicus was derived from mitochondrial sequence data obtained from 422 specimens originating from a total of 103 populations. Different samples obtained from the marisma habitat (natural temporary marshes; see Serrano et al. 2006) of Doñana National Park were considered as belonging to different populations (Table A1).
Genetic diversity of populations and lineages
Analysis of molecular variance (AMOVA) was calculated in Arlequin 2.000 (Schneider et al. 2000) to investigate the level of differentiation in the main phylogenetic lineages of Triops mauritanicus. We included only populations in the analysis for which sequence data from at least six individuals were available (populations 001–039, 082–087 and 094–097; Table A1). The ‘S.Iberian’ lineage was broken down to subgroups referring to four main habitat types: (1) marismas of Doñana National Park (populations
001–009); (2) ponds surrounded by forest or shrubland
(populations 010–020); (3) ponds in open habitat within
25 km distance to the marismas (populations 021–030; these ponds were situated within 2 km distance from the marismas before vast areas of marshland were transformed to farmland during the 20th century—the sampling sites are now mainly used as pastures or fields); (4) ponds in open habitat more than 75 km from the marismas of Doñana National Park. In the last category, we included only Portuguese samples (populations 033–039) in order to compare data obtained from geographical areas of similar size (i.e., populations 031 and 032 from eastern Sevilla province were excluded from the AMOVA, but were included in this habitat category to investigate gene diversity, see below). A pond was considered to be situated in open habitat when wide areas of meadows or fields were flanking the pond at least to one side. Significance tests were based on 1,023 permutations.
To further investigate genetic diversity in regard to ecological factors, the index of gene diversity H (Nei 1987) as implemented in Arlequin 2.000 (Schneider et al. 2000) was calculated for each of the 39 populations obtained from the ‘S.Iberian’ lineage for which we investigated a minimum of 6 individuals each (populations 001–039). A single-factor analysis of variance (ANOVA) was used to compare levels of gene diversity among the habitat categories as defined above. It is common knowledge that ponds situated in open habitat are more attractive to waterbirds (a major dispersal vector for branchiopod crustaceans, see e.g. Sánchez et al. 2007) than ponds surrounded by forest or shrubs (for a case study demon- strating a clear avoidance of habitats flanked by hedges, see Tourenq et al. 2001). As a result, probabilities of dispersal by waterbirds should be higher in the absence of wooded margins, and we predicted higher gene diversities in Triops populations situated in open habitat, including the marismas. Simple regression analysis was used to further investigate a possible correlation of gene diversity in Triops to waterbird abundances. As no exact data on waterbird abundances in our sampling sites were available, we estimated relative abundances of waterbirds for each of the sampling sites within Sevilla and Huelva provinces (populations 001–
032), at a scale of 5 abundance levels (from 1=low to
5=high). Estimates were based on regular observations during recent years (performed by A.J.G. while blind to the genetic data; see also Rendon et al. 2008).
In addition, we tested for a possible effect of land use for pastures on the gene diversity in Triops, as cattle could represent an important vector for dispersal on a local scale. We used an ANOVA with two levels of land use: no pasture use vs. pasture use (populations 001–032 and 035–039; no data available for 033 and 034). Finally, a possible correlation of gene diversity in Triops to abiotic factors
was investigated with a multiple regression analysis. Our data on abiotic factors for marisma sites were incomplete, as the central regions within the extensive marismas of Doñana could not be accessed during the flooding phase, so that Triops samples from these sites were raised from sediment samples collected during the dry season. Thus, abiotic factors were only investigated for non-marisma habitats. We included pH, conductivity, surface area of the ponds, and distance to marismas as dependent variables in the multiple-regression model. Surface area was measured on satellite photographs obtained from Google Earth version 4.0.2091 beta (Google, Inc.) using PixeLINK Capture SE (Version 3.1. obtained from www.pixelink. com); distances were measured directly in Google Earth (alternatively, for Portuguese ponds, surface area was derived from GPS signals or tracks taken in the field, using UTHSCSA Image Tool software, available at http://ddsdx. uthscsa.edu/dig/itdesc.html).
As part of the samples were obtained from sediments, we used three populations of T. mauritanicus to test if diversity measurements obtained from field-collected specimens differed from those obtained from lab-raised samples. For each of these populations, we obtained 12S sequences from six field-collected specimens and from an additional six specimens that were raised from eggs from sediment samples. Haplotype diversity was identical for two of the populations and differed by only one (four vs. five haplotypes) in the third sample, indicating that results are comparable and that it is a valid procedure to merge both types of data into a single data file for statistical analysis of diversity measurements (generalization to other studies may not be valid unless the sediment sampling procedure equals that used for the present study, which implies collecting numerous subsamples from different parts of a pond; subsamples can be pooled).
Evidence for recent passive dispersal and gene flow
To test if the present geographic segregation of the main phylogenetic lineages within south-western Iberia might be the result of a general dispersal limitation in gonochoric Notostraca (Korn et al. 2006), we inferred the order of magnitude of dispersal abilities from geographic haplotype distributions. We calculated the additive minimum geo- graphic distance between all occurrence sites of shared haplotypes. We call this the ‘accumulative minimum dispersal distance’ (AMDD) of the haplotype. For the geographic distance measurements performed in the present study, we made the simplifying assumption that shared haplotypes were always the result of dispersal (an indepen- dent evolution of identical 12S haplotypes at different sites is assumed to be a rare event, since evolutionary rates in the
12S gene are low enough to be applied to phylogenetic
studies at higher taxonomic levels, see, e.g. Ballard et al.
As the whole area covered by the marismas of Doñana is interconnected at certain flood events (Serrano et al. 2006), active dispersal may occur in addition to strictly passive dispersal among populations situated within the marismas. Consequently, the marismas were treated as a single site for our passive-dispersal distance measurements, referring to the original range and borderlines of the marismas around the year 1900 (Montes et al. 1998: Fig. 6.4), before large areas of natural marshes were transformed into farmland. In addition, we measured total dispersal distances (including obligately passive overland dispersal among strictly sepa- rated ponds as well as potentially active dispersal among marisma sites during high flood events), i.e. those obtained if all sampling localities that were originally situated (or are still situated) within the marismas were treated as separate sites.
For comparison to the results obtained for south-western Iberian lineages of Triops mauritanicus, we measured AMDDs for T. c. cancriformis. As no exact coordinates were available for some of the sites, AMDDs for this species were rounded off to the nearest multiple of 100 km.