In the late Pliocene, the closure of the Isthmus of Panama was a trigger to the diversification of many species in the Neotropics. The separation of previously continuous populations created two groups of extant species, which live now in the Atlantic and Pacific drainage systems. This vicariant event opened a unique opportunity for studies on evolution, divergence and speciation processes (Knowlton et al. 1993, Knowlton and Weigt 1998, Lessios 2008). The Central American land bridge is a well-dated biogeographic barrier and is a relatively recent event, about 3.1 million years ago (Keigwin 1978, Coates et al. 1992, Coates and Obando 1996, Anger 2013). Since then, the Atlantic and Pacific marine ecosystems became gradually separated, whereas the gene flow was blocked between organisms on either side.

In spite of the geographic separation, some species are difficult or impossible to distinguish using traditional morphological features, and are thus called “sibling species” (see Knowlton 1993 and references cited therein). These sibling species refer to pairs of species that are genetically closely related, but reproductively isolated (Mayr 1963, Steyskal 1972, Knowlton 1986). Others authors refer to “sibling” as “geminate species” (Jordan 1908, Stillman and Reeb 2001, Marko 2002), in which individuals were separated necessarily by a geographic barrier, and each member of the pair occurs along one coast of the Americas (Lessios 1998, Miura et al. 2010). Other non-morphological features have been used to distinguish these species such as “karyology, hybridization experiments to detect postzygotic incompatibility, distribution patterns, resource use, breeding season, life history and development, mating behavior (including visual, acoustical, and chemical signals), color pattern, and various biochemical characters” (Knowlton 1986). Consequently, a pair of species, reproductively isolated and very similar in morphology, is not necessarily considered as sibling species, and an interdisciplinary approach is necessary to evaluate this conclusion.

Molecular tools have been used to contribute with species delimitation in several cryptic decapods (Schubart et al. 2001a, b, Kitaura et al. 2002, Lai et al. 2010, Pileggi and Mantelatto 2010, Mantelatto et al. 2011, Negri et al. 2012, Torati and Mantelatto 2012). Phylogenies based on molecular data has evidenced probable cases of misidentification of sibling species based on morphology (Lessios 2008, Rossi and Mantelatto 2013). For some freshwater species, the isolation by the closure of the Isthmus of Panama might be questionable, since species of the genus Macrobrachium Spence Bate, 1868 can disperse over greater distances than the width of the Isthmus (Steeves et al. 2005, Bauer and Delahoussaye 2008, Bauer 2011) and may also use the Panama Canal as passageway for both sides (Hildebrand 1939, Abele and Kim 1989).

Most studies on decapods sister species focused only in marine species of the genus Alpheus Fabricius, 1798 (Knowlton et al. 1993, Knowlton and Weigt 1998, Wehrtmann and Albornoz 2002), while our knowledge of the impact of the Isthmus of Panama on freshwater-invading decapods is extremely limited (Anger 2013). Prawns of the genus Macrobrachium are widely distributed in rivers of tropical and subtropical regions with more than 240 recognized species worldwide (De Grave and Fransen 2011). Although its greatest diversity has been found in the Indo-Pacific region, in the Americas there are more than 55 valid species, representing an area of great importance concerning the diversity of the family Palaemonidae (Holthuis 1952, Pileggi and Mantelatto 2012).

The high morphological similarity between some American species led Holthuis (1952) to designate Atlantic and Pacific Macrobrachium “sister species”. Until now, morphological similarities between the transisthmian “sibling species” have impeded the identification of the following pairs of species: Macrobrachium occidentale × Macrobrachium heterochirus, Macrobrachium americanum × Macrobrachium carcinus, Macrobrachium digueti × Macrobrachium olfersii, Macrobrachium hancocki × Macrobrachium crenulatum, Macrobrachium tenellum × Macrobrachium acanthurus and Macrobrachium panamense × Macrobrachium amazonicum. These species occur primarily in Central America, with the first species of each pair is found in the Pacific drainage and the second in the Atlantic side. Larvae of these species require saline water (i.e., 10–35 ppt) to complete their life cycle, and exhibit other adaptive features, such as extended larval development and amphidromous life histories (Hedgpeth 1949, Bauer and Delahoussaye 2008, Bauer 2011, 2013). Moreover, these prawns show great morphological modifications during ontogenesis, and as other congeneric species they present controversial systematic issues, with high interspecific conservatism and males with intraspecific variation, as found among distinct morphotypes (Holthuis 1952, Moraes-Riodades and Valenti 2004, Pileggi and Mantelatto 2010, Vergamini et al. 2011). Considering the doubt whether the previously indicated species of Central American Macrobrachium are sister taxa or not, our study aimed to evaluate in a molecular phylogenetic context the relationships among 12 transisthmian Macrobrachium “sibling species” from the Americas in order to assess the validity of their current species level.

Our phylogenetic analysis included 12 transisthmian American species of Macrobrachium, 7 from other American Macrobrachium species, and 2 from palaemonid-related groups. We generated 95 new sequences: 26 mitochondrial 16S sequences, 25 mitochondrial COI sequences, and 44 nuclear 18S sequences. The analysis of the 181 sequences from the three genes produced an alignment of 1.645 bp.

The optimal model for the concatenated data set was the TPM1uf model of sequence evolution (Kimura 1981) plus gamma distributed rate heterogeneity with a significant proportion of invariable sites (TPM1uf +I+G) with the following parameters: assumed nucleotide frequencies A = 0.3028, C = 0.2125, G = 0.1909, T = 0.2937; proportion of invariable sites I = 0.6020; the variable sites followed a gamma distribution, with shape parameter = 0.6700.

The topology obtained by maximum likelihood from concatenated genes (16S, 18S and COI) analyses confirmed that the transisthmian sibling species (Macrobrachium heterochirus × Macrobrachium occidentale – Sibling 1, Macrobrachium carcinus × Macrobrachium americanum – Sibling 2, Macrobrachium olfersii × Macrobrachium digueti – Sibling 3, Macrobrachium crenulatum × Macrobrachium hancocki – Sibling 4, and Macrobrachium acanthurus × Macrobrachium tenellum – Sibling 5) are closely related by well-supported clades (Fig. 1). Sibling 6 (Macrobrachium amazonicum × Macrobrachium panamense) did not form a separate sister clade despite being phylogenetically close. The position of Palaemonetes argentinus showed a stable condition in an external branch. However, the other outgroup (Cryphiops caementarius) was maintained within the Macrobrachium clade in the phylogeny (Fig. 1). The results did not reveal geographical separation among populations of the same species inside each group (Siblings 1–5). Macrobrachium michoacanus (see the arrow in the phylogeny) seems to be close related to Macrobrachium hancocki in Sibling 4 group.

The relation among the sibling groups is supported by morphological traits. The species included in Siblings 1 and 2 exhibit similar shapes of the rostrum with the upper margin somewhat arched over the eye and with the apex directed upward (Fig. 1C). Species of the Siblings 3 and 4 with Macrobrachium michoacanus and Macrobrachium surinamicum show similar rostrum, being almost straight and usually with more than 10 teeth in the upper margin (Fig. 1B). In the same way the species of Siblings 5 and 6 and Macrobrachium jelskii have a distinct rostrum, which is elongated, slender, with apex curved upward, with many teeth in the upper and lower margin (Fig. 1A).

In general, distance analyses revealed that the percentage of intraspecific variation was lower than interspecific variation (Table 2). Considering the relation between distinct sibling species, the genetic variability ranged from 4.4% (Sibling 3 × Sibling 4) to 16.9% (Sibling 4 × Sibling 6) for 16S, from 11.3% (Sibling 3 × Sibling 4) to 23.9% (Sibling 2 × Sibling 5) for COI, and from 1.1% (Sibling 5 × Sibling 6) to 11.3% (Sibling 2 × Sibling 5, 6) for 18S (Table 2). Inside each sibling group, the genetic variability varied between 1.5% (Sibling 3) and 8.7% (Sibling 6) for 16S, between 8.3% (Sibling 2) and 16.9% (Sibling 5) for COI, and between 0.0% (Sibling 2, 3) and 1.1% (Sibling 1) for 18S (Table 2).

Over 150 sequences from three different gene regions were used in the present study in order to estimate phylogenetic relationships among freshwater prawns of the genus Macrobrachium, which previously were assumed to be transisthmian sibling species. The results revealed that all geminate species studied herein were valid taxonomic entities. Likewise they confirmed the role of the Isthmus of Panama as an effective barrier contributing in the separation of sibling species by the mechanism of allopatric speciation. However, in other cases the separation happened before the closure of the Isthmus probably by the mechanism of sympatric speciation. Our multigenic phylogeny produced consistent groups in most of the pairs of geminate species i.e., sister taxa geographically separated: Macrobrachium heterochirus × Macrobrachium occidentale, Macrobrachium carcinus × Macrobrachium americanum, Macrobrachium olfersii × Macrobrachium digueti, Macrobrachium crenulatum × Macrobrachium hancocki and Macrobrachium acanthurus × Macrobrachium tenellum. The constitution of these clades corroborates the morphological proximity of each pair of species as mentioned by Holthuis (1952).

The genetic divergence analyses showed the separation of each sibling group from others, which suggest a consistent relation in comparison with other congeners species (Table 2). Considering that for 16S the divergence in decapods is presumed to be around between 0.6 to 0.9% per Myr (Schubart et al. 2000), we can estimate the divergence time of the sibling species according to the closure of the Isthmus. For Siblings 1 and 5 the time of divergence between the species was approximately from 5.1 to 7.8 and 7.11 to 11.5 Mya, respectively. These estimates predate the closure of the Isthmus, which suggest that the speciation process separated these two species already before the closure. Considering that these amphidromous species are dependent of estuarine water for successful larval development, a sympatric speciation hypothesis seems to be unlikely. However, in these cases the possibility of occurrence of this event is plausible, probably due to environmental changes (Knowlton et al. 1993, Knowlton and Weigt 1998, Morrison et al. 2004). A genetic differentiation could have arisen from a mutational step and the two subpopulations, whose geographic distribution ranges overlap completely, became isolated because both occupy completely different ecological niches (Bush 1994). Analogous events were reported for other crustacean (Malay and Paulay 2009, Jarman et al. 2011). In addition, the estuary can contribute to restriction of the gene flow between the species by distinct selective regimes or habitat fidelity of the species, generating potential speciation in complete or partial isolation (Stanhope et al. 1992, Bilton et al. 2002). Therefore, the sympatric speciation may have occurred in these sibling species by the mechanism of microallopatry (Fitzpatrick et al. 2008). The difficulty in separating Macrobrachium heterochirus from Macrobrachium occidentale, and Macrobrachium acanthurus from Macrobrachium tenellum using morphological, ecological and genetic characters (Tables 2, 3, 5), allied with the consistent position in the phylogeny (Fig. 1) provide convincing arguments to consider them as sibling species. The phylogenetic position of Siblings 1 with 2 and Siblings 5 with “6” (here marked between quotes due its artificial position, not characterized as sibling) followed the morphological grouping based on the shape of the rostrum (Fig. 1A, C) indicating that this character is determinant for taxonomic studies.

The time of divergence between both species of the Sibling 3 was approximately from 1.66 to 3.16 Mya for 16S gene, which supports the efficiency of the barrier in the separation of sibling species by mechanism of allopatric speciation. The morphologically close relation of the “olfersii complex” (see Villalobos 1969 for revision) was corroborated in the phylogeny, where Siblings 3 and 4 form sister groups with the same shape of the rostrum (Fig. 1B), as evidenced in previous molecular results (Rossi and Mantelatto 2013). The entity of the results obtained together with morphological and ecological similarities of Macrobrachium olfersii and Macrobrachium digueti suggest that both are sibling species, but the inclusion of other species from the Macrobrachium olfersii complex in the analysis is necessary to confirm this proposition. Among the sibling species proposed by Holthuis (1952), only one pair (Macrobrachium surinamicum × Macrobrachium transandicum) was not analyzed in our study due the impossibility to obtain specimens of Macrobrachium transandicum. In our phylogeny, Macrobrachium surinamicum was included inside the clade of Macrobrachium olfersii complex (Villalobos 1969, Rossi and Mantelatto 2013) corresponding to a species with the rostrum almost straight, usually with more than 10 teeth in the upper margin (Fig. 1B).

For Siblings 2 and 4 the time of divergence between the species varied from 2.11 to 4.66 and 1.88 to 3.5 Mya for 16S gene, respectively. These data place them exactly in the range of the closure of the Isthmus, precluding the definition that the separation of the species may have been caused by this vicariant process. Pileggi and Mantelatto (2010) mentioned that Macrobrachium americanum could be a synonymous of Macrobrachium carcinus based on a single molecular 16S phylogeny. However, and as suggested by the authors, a more extensive sampling of Macrobrachium americanum will be necessary to verify this proposition. Our results that include five specimens of Macrobrachium carcinus and seven of Macrobrachium americanum from distinct localities revealed that both species are sibling species.

In the same way, our data as well as the morphological and ecological similarities evidenced the close relationship between Macrobrachium crenulatum and Macrobrachium hancocki; however, the addition of data from more specimens and other species from the Macrobrachium olfersii complex is necessary to confirm them as sibling species, i.e., sister taxa geographically separated (Rossi and Mantelatto, unpubl. data). Another unpredictable result was the close relation of Macrobrachium michoacanus with Macrobrachium hancocki (Fig. 1, Sibling 4). With both occurring on the Pacific side, this result may be interpreted as an indication that the relation of phylogenetically closely related congeners living on either side of the Isthmus must be older than the biogeographic barrier separating them (Anger 2013). New diversifications succeeding the closure of the Isthmus occurred at the same side, which can be demonstrated by the higher proximity between these sympatric species than Macrobrachium hancocki and Macrobrachium crenulatum, the hypothetical Sibling 4. However, analysis of additional material is necessary to verify the phylogenetic position of Macrobrachium michoacanus. Following the other sibling species, relationship of the systematic position with the shape of the rostrum was maintained (Fig. 1B–C) supporting the high reliability of this morphological character.

Our results regarding Macrobrachium amazonicum × Macrobrachium panamense did not confirm a separate sibling group despite the close phylogenetic relation among these species. Our multigenic phylogenetic hypothesis (Fig. 1) indicates Macrobrachium panamense as a sister group of the Sibling 5, and Macrobrachium amazonicum as a sister species of this group (Macrobrachium panamense + Sibling 5). Genetic divergence analyses of the “Sibling 6” pair (8.33 to 14.5 Mya for 16S genes) suggest that the time of their divergence predates the closure of the Isthmus, indicating that both did not share the same ancestral. In addition, the wide geographic distribution of Macrobrachium amazonicum in the large South American river basins must be related to geological events driven by the rising Andes along the western portion of these basins (supposedly its native area of occurrence) (see Magalhães et al. 2005 for revision). Macrobrachium jelskii as an external clade of Sibling 5 and “Sibling 6” is in agreement with morphological similarities among these species, mainly of the shape of the rostrum (Fig. 1A), despite Macrobrachium jelskii being the unique species of the group to present abbreviated larval development. The position of a more external group (Macrobrachium potiuna, Macrobrachium brasiliense, Macrobrachium borellii) with abbreviated larval development in the phylogeny indicates that the ancestral species of this entire group possibly had a life cycle independent of salt water as suggested in previously studies (Murphy and Austin 2005, Pileggi and Mantelatto 2010). Macrobrachium amazonicum plays a key role in this puzzle since it presents inland and coastal populations (Vergamini et al. 2011, Meireles et al. 2013), suggesting that the species originated in freshwater environments and entered subsequently in estuarine habitats (Pereira and García 1995, Pileggi and Mantelatto 2010).

Phylogenetic analyses were based on two mitochondrial and one nuclear genes in order to provide a broad spectrum of inference and insights into the evolutionary history of Macrobrachium in the Americas. Although the mitochondrial markers may offer strong evidence for genus and species-level relationships, they have high mutation rates, which can cause increasing saturation when older splits are analyzed (Simon et al. 1994, Avise 2004). Therefore, analyses were carried out with sequences from conserved and variable genes to access phylogenetic information across a range of evolutionary time (Crandall et al. 2000). The genes concatenated analysis improve the diversity of evolutionary time, consequently revealed a more consistent phylogeny compared to previous morphological and molecular phylogenetics studies (Pereira 1997, Murphy and Austin 2005, Pileggi and Mantelatto 2010). The inclusion of the member of genus Cryphiops within Macrobrachium species was maintained in the phylogeny, and raises the question whether Macrobrachium is a monophyletic group (Pereira 1997, Murphy and Austin 2005, Pileggi and Mantelatto 2010, Carvalho et al. 2013, Rossi and Mantelatto 2013).

The results of our multidisciplinary approach suggest that species pairs 1-5 refer to siblings, in which each pair of species is difficult to distinguish using traditional morphological characters, although they are genetically distinct, closely related, and reproductively isolated (Mayr 1963, Steyskal 1972, Knowlton 1986). In contrast, our data did not validate “Sibling 6” by molecular analysis, although morphology, ecology, and geographic distribution patterns seem to suggest that they are sibling species (Holthuis 1952). Moreover, the speciation processes of the species of the pairs 2, 3, and 4 seem to have occurred after the rise of the isthmus barrier, probably in Pliocene and Pleistocene by the mechanism of allopatric speciation. However, the isolation of pairs 1 and 5 may have happened before the rise of the isthmus barrier, probably in Miocene by the mechanism of sympatric speciation.

An intriguing case refers to the occurrence of two species (Macrobrachium hobbsi and Macrobrachium olfersii) on both sides of the Central American land bridge (Anger 2013). The identification of these species may be incorrect or is related to the possibility of a passageway that has started with the opening of the Panama Canal, a scenario that has been already reported (Hildebrand 1939, Abele and Kim 1989). The possible expansion of the distribution range of Macrobrachium through the Panama Canal may occur, especially considering the dispersal potential of these amphidromous species (Bauer and Delahoussaye 2008, Bauer 2011, 2013). However the findings of the previous genetic study with Macrobrachium olfersii revealed the absence of gene flow between Pacific and Atlantic populations. Moreover, Macrobrachium digueti and specimens from Pacific considered as Macrobrachium olfersii did not show divergence enough to split them, and the differences were within the range of interspecific values. Therefore, on the Pacific coast only Macrobrachium digueti occurs naturally, which is considered, like Macrobrachium olfersii, a sibling and cryptic species (Rossi and Mantelatto 2013).

This is the first phylogenetic study using molecular methods devoted entirely to the American transisthmian Macrobrachium sister species. Molecular markers confirmed that the Isthmus of Panama is an effective barrier contributing to the separation of sibling freshwater prawns species by the mechanism of allopatric speciation. However, some species seemed to have evolved before the closure of the Isthmus by the mechanism of sympatric speciation. Our phylogenetic analysis revealed consistent groups in most of the studied pairs endorsing the supposed sibling species. In contrast, the position of one pair (Macrobrachium amazonicum × Macrobrachium panamense) seems to be artificial once they did not share a recent common ancestor. The results presented here contribute to resolution of some doubts about the relationships of geminate American species. Our results support the conclusion that these sibling species are valid taxonomic entities, but not all transisthmian species are the closest living relatives with each other.