Material and Methods
Between 2000 and 2004, 71 specimens of freshwater crabs (Brachyura: Gecarcinucidae, see Klaus et al., 2006) were collected from 24 sampling points (Fig. 1) from the Indonesian island of Sulawesi. Most of the specimens were collected from the Malili lakes (Table 1). The specimens were preserved and stored in ethanol (75-99%). In addition, tissue was obtained from museum collections of the Raffles Museum Zoological Reference Collection Singapore (ZRC) and the Museum für Naturkunde Berlin (ZMB).
Fig. 1. Maps with the location of the freshwater-crab sampling points: A) The island of Sulawesi, B) the Malili region (modified from von Rintelen and Glaubrecht, 2005).
We amplified and sequenced 658bp of the cytochrome oxidase I (Cox1) gene from 30 specimens of Nautilothelphusa zimmeri (10 from Lake Matano, 8 from Lake Mahalona and 12 from Lake Towuti), 24 specimens of Parathelphusa ferruginea (6 from Lake Mahalona, 16 from Lake Towuti and 2 from tributaries to Lake Towuti), six of Parathelphusa pantherina (Schenkel, 1902) from Lake Matano, two of Syntripsa matannensis (Schenkel, 1902) from Lake Matano, and three of Syntripsa flavichela Chia and Ng 2006 from Lake Mahalona and Lake Towuti. As river species, we included two specimens of Parathelphusa pallida (Schenkel, 1902) and one specimen each of Parathelphusa possoensis (Roux, 1904), Parathelphusa sarasinorum (Schenkel, 1902), Parathelphusa celebensis (de Man, 1892) and Parathelphusa lokaensis (de Man, 1892). Among those, P. lokaensis is the only Sulawesi freshwater crab species not occurring within the Malili region and was used as an outgroup, because it may share a relatively recent common ancestor with the fauna from and around the lakes.
DNA was extracted from leg muscle tissue with the Puregene method (Gentra Systems) and used for polymerase chain reactions (PCR) with the primers LCO 1490 (GGT CAA CAA ATC ATA AGA TAT TGG) (Folmer et al., 1994), COL6b (ACA AAT CAT AAA GAT ATY GG) and COH6 (TAD ACT TCD GGR TGD CCA AAR AAY CA) (Schubart and Huber, 2006). PCRs were carried out with the following profile: 40 cycles with 45 sec 94°C, 1 min 48-50°C and 1 min 72°C for denaturing, annealing and extension respectively. PCR products were purified with the Quick-Clean kit (Bioline) or Microcon filters (Millipore), cycle sequenced with the Big Dye Terminator v1.1 (AB Applied Biosystems) and analysed with the automated capillary DNA sequencer ABI PRISM® 310 Genetic Analyzer (Applied Biosystems). Sequences of all haplotypes have been submitted to the European Molecular Biology Laboratories (EMBL) and are accessible under FM 177599 - FM 177643.
Alignment of the sequences was carried out with the program BioEdit (Hall, 1999). Absence of insertions/deletions and stop codons suggested lack of nuclear copies and resulted in one obvious alignment. The aligned sequences were analyzed for the best fitting model of DNA evolution with the software Modeltest 3.6 (Posada and Crandall, 1998). The resulting model of evolution GTR+I+G with the corresponding parameters (gamma = 0.871; I = 0.523) was used for Minimum Evolution (ME) and Bayesian Inference (BI) analyses. We used three different methods of phylogenetic inference for our dataset, Maximum Pasimony (MP) and Minimum Evolution (ME) using the software package PAUP* 4.0 (Swofford, 2001) and the Bayesian analysis (BI) using MrBayes v.3.0b4 (Huelsenbeck and Ronquist, 2001). MP trees were calculated with a heuristic search. Confidence values for the different groups within the trees were calculated with the bootstrap method (2,000 pseudoreplicates). Starting trees were obtained by stepwise random sequence addition with ten repetitions and five trees held per step. Branch swapping was conducted with the tree-bisection-reconnection (TBR) algorithm, holding multiple trees (MulTrees). Maximum number of trees was set to 5,000 (max trees). Otherwise, the default options of PAUP* 4.0 were used. Only minimal trees were retained and zero length branches were collapsed. The Bayesian analysis was run with four MCMC chains for 2,000,000 generations, saving a tree every 500 generations. After excluding a burn-in phase of 20,000 generations with possible random and suboptimal trees, the posterior probabilities of the phylogeny were determined for the remaining trees. Consensus trees were constructed using the ‘sumt’ option of MrBayes. As a measure of genetic differentiation between taxa, the FST value was calculated with an AMOVA (Excoffier et al. 1992) using the software Arlequin ver. 3.0 (Excoffier et al., 2005).