Material and methods
Sample collectionnext section
We used specimens from most of the coastal and inland regions of Brazil where this species has been reported to date. Thirteen populations of M. amazonicum were analyzed, from throughout the country (Fig. 1). The populations were classified according to the Brazilian National Hydrographic Division (Brasil, 2003) and the influence of brackish water (coastal: those restricted to river systems close to the seacoast, with brackish water influence; and inland: those found in inland river systems with no connection to the coast).
Coastal populations covered the following Hydrographic Regions (HR): Amazonian, Tocantins-Araguaia, and Eastern/Northeast Atlantic (Fig. 1). The inland populations were divided in two groups: Amazonian HR, and Paraguay and Paraná HR (Fig. 1). Inland populations sampled in the state of São Paulo and along the northeastern Brazilian coast were classified as introduced, because of their unnatural distributions (Magalhães et al., 2005).
Some specimens were obtained from field collections, carried out in compliance with current applicable state and federal laws of Brazil (DIFAP/IBAMA, 126/2005; permanent license for collection of Zoological Material No. 11777-1 MMA/IBAMA/SISBIO). These specimens were incorporated into the Crustacean Collection of the Biology Department (CCDB) of the Faculty of Philosophy, Sciences and Letters of Ribeirão Preto (FFCLRP) and the University of São Paulo (USP) (Appendix). Complementary specimens were acquired by donation or loan from crustacean collections, or were collected and sent to us by collaborating researchers from several institutions in Brazil (Appendix). Donated material was preserved directly in 80% ethanol and deposited in the CCDB. The identifications were based on the diagnostic morphological traits of M. amazonicum (Heller, 1862; Holthuis, 1952; Gomes-Corrêa, 1977; Melo, 2003).
Based on the proposed phylogeny for Macrobrachium by Pileggi and Mantelatto (2010), we identified the species that are more closely related and reliably distant from M. amazonicum, to compose the outgroup in our analyses (Appendix).
DNA extraction, amplification, and sequencing
All sequences used in this study were generated from our own extractions for this project. When possible, the analyses used three to ten specimens from each collection site, in order to limit the chance of misidentifications and variability. Genetic vouchers, from which tissue samples were obtained, were deposited in appropriate collections (Appendix). All procedures followed Mantelatto et al. (2007, 2009a) and Pileggi and Mantelatto (2010), with appropriate modifications. Total genomic DNA was extracted from the abdomen or from the pereiopod muscle tissue.
A polymerase chain reaction (PCR) was conducted in a Thermo® PxE 0.2 Thermal Cycler, using the universal primers for invertebrates: 16Sar (5′-CGCCTGTTTATCAAAAACAT-3′) and 16Sbr (5′-CCGGTCTGAACTCAGATCACGT-3’) (Palumbi et al., 1991) for the 16S rRNA (the large subunit of the ribosomal rRNA), and COI-a (5’-AGTATAAGCGTCTGGGTAG TC-3’) and COI-f (5’-CCTGCAGGAGGA GGAGACCC-3’) (Palumbi and Benzie, 1991) for the COI gene. PCR products were purified using Microcon 100® filters and a SureClean Plus kit, and were sequenced with the ABI Big Dye® Terminator Mix in an ABI Prism 3100 Genetic Analyzer® following Applied Biosystems protocols. All sequences were confirmed by sequencing both strands. The consensus sequence for the two strands was obtained using BioEdit Version 18.104.22.168 (Hall, 1999). Sequences were edited using BioEdit and aligned in Clustal W (Thompson et al., 1994) with interface in BioEdit, with default parameters. All sequences were submitted to GenBank (Appendix).
It is recommended that, at least preliminarily, the phylogenetic relationships that delimit a monophyletic group be resolved, so that an analysis can be undertaken with only one segment of this group (Amorim, 2002). Considering Macrobrachium as a natural group (Murphy and Austin, 2005; Lui et al., 2007; Pileggi and Mantelatto, 2010), our phylogenetic analysis focusing on M. amazonicum populations can be considered relevant and justified.
The gaps from the 16S mtDNA sequences, which are due to real gaps in the alignment, were removed in order to obtain non-aligned sequences. No gaps were found in the alignment of COI sequences. These sequences were analyzed in POY Version 4.0 (Varón et al., 2007) using the direct optimization method, with parsimony as the optimality criterion (Wheeler, 1996). This methodology has given consistent results in recent molecular phylogenies of crustaceans (Mantelatto et al., 2009b; Pileggi and Mantelatto, 2010). Topologies were constructed through random addition sequence, followed by a combination of refinement parameters. Sensitivity analysis was carried out using different cost matrices, as suggested by Wheeler (1995). All data sets for the parsimony analysis were analyzed under 10 parameter sets for a range of indels, transition, and transversion ratios. The matrix digits (111, 112, 113, 211, 212, 221, 411, 412, 812, and 821) correspond to the ratio of indel/transversion/transition values, respectively.
Distance analyses were carried out by the static alignment procedure for both gene sequences. Ambiguous regions of the sequences were removed. Substitution models used in distance matrix calculations were previously selected under the Akaike Information Criterion (AIC) (Posada and Buckley, 2004) among 56 available alternatives of the program ModelTest Version 3.7 (Posada and Crandall, 1998). Matrix data were grouped by Neighbor Joining (NJ) (Saitou and Nei, 1987) in PAUP Version 4.0 beta10 (Swofford, 2003) using the maximum-likelihood distance correction set. The consistency of topologies was measured by the bootstrap method (Felsenstein, 1985) with 1000 replicates; only confidence values > 50% were reported. In order to estimate intra- and interspecific divergence rates, genetic distances were also calculated in PAUP using the p distance. All positions were compared directly for each pair of sequences, one at a time.
In this analysis, we considered COI sequences from coastal and inland populations of M. amazonicum. The haplotype number was calculated in DnaSP Version 4.10.9 (Rozas and Rozas, 1999). The haplotype and nucleotide diversities were calculated for each population using Arlequin Version 3.1 (Excoffier et al., 2005).
Haplotype networks were constructed by the statistical parsimony method in TCS (Version 1.21) (Clement et al., 2000) and by the Median-Joining method in Network (Bandelt et al., 1999), with data preparation in DnaSP. Networks were constructed in two phases. First, introduced populations with an unnatural distribution (Appendix, Fig. 1) were not included because their origins are unknown, and the results could be skewed or masked by their presence in the analysis. In a second phase, when the genetic variability among natural populations had been estimated, an analysis with all populations was carried out so that the probable origin of the introduced populations could be inferred.
Series of analyses of molecular variance (AMOVA) (Excoffier et al., 1992) were computed in Arlequin to examine the distribution of genetic variation. Analyses were run based on haplotype frequencies with no hierarchical structure (all populations in a single group) and with regional subdivisions defined according to the results of the haplotype networks. The significance was tested using a nonparametric permutation procedure (Excoffier et al., 1992), incorporating 10,000 permutations.