Phylogeny and biogeography of cichlid fishes (Teleostei: Perciformes: Cichlidae)

Family level molecular phylogenetic analyses of cichlid fishes have generally suffered from a limited number of characters and ⁄or poor taxonomic sampling across one or more major geographic assemblage, and therefore have not provided a robust test of early intrafamilial diversification. Herein we use both nuclear and mitochondrial nucleotide characters and direct optimization to reconstruct a phylogeny for cichlid fishes. Representatives of major cichlid lineages across all geographic assemblages are included, as well as nearly twice the number of characters as any prior family-level study. In a strict consensus of 81 equally mostparsimonious hypotheses, based on the simultaneous analysis of 2222 aligned nucleotide characters from two mitochondrial and two nuclear genes, four major subfamilial lineages are recovered with strong support. Etroplinae, endemic to Madagascar (Paretroplus) and southern Asia (Etroplus), is recovered as the sister taxon to the remainder of Cichlidae. Although the South Asian cichlids are monophyletic, the Malagasy plus South Asian lineages are not. The remaining Malagasy lineage, Ptychochrominae, is monophyletic and is recovered as the sister group to a clade comprising the African and Neotropical cichlids. The African (Pseudocrenilabrinae) and Neotropical (Cichlinae) lineages are each monophyletic in this reconstruction. The use of multiple molecular markers, from both mitochondrial and nuclear genes, results in a phylogeny that in general exhibits strong support, notably for early diversification events within Cichlidae. Results further indicate that Labroidei is not monophyletic, and that the sister group to Cichlidae may comprise a large and diverse assemblage of percomorph lineages. This hypothesis may at least partly explain why morphological studies that have attempted to place Cichlidae within Percomorpha, or that have tested cichlid monophyly using only ‘‘labroid’’ lineages, have met with only limited success. The Willi Hennig Society 2004. Cichlidae is a species-rich clade of perciform fishes that has attracted much attention from systematists, particularly the ‘‘species flocks’’ of the East African lakes (Kornfield and Smith, 2000), which have been the focus of numerous microand macroevolutionary studies (e.g., Seehausen et al., 2003; Verheyen et al., 2003; are but two recent studies investigating the origin of the Lake Victoria cichlids). The current distribution of cichlids is essentially Gondwanan and they have a fossil record extending to the Eocene of Africa ( 46 Ma; Murray, 2000). Interestingly, when compared with basal African lineages such as Heterochromis and Tylochromis, these Eocene fossils are found to share derived features with the remaining African lineages; they appear to be nested well within the African clade (Murray, 2000, 2001), suggesting a significantly older origin for the family. Despite the great deal of attention that has been focused on this group, we still know little about either the diversification of the family or its placement within Percomorpha. There have been several attempts to reconstruct the familial-level relationships of cichlid fishes using nucleotide characters (see Sparks, 2004a, for a partial list of such studies), but only a single study has *Corresponding author: John S. Sparks, Department of Ichthyology, Division of Vertebrate Zoology, American Museum of Natural History, Central Park West at 79th Street, New York, NY 10024, USA. Tel.: +1 212 313 7791; Fax: +1 212 769 5642. E-mail address: [email protected] The Willi Hennig Society 2004 Cladistics www.blackwell-synergy.com Cladistics 20 (2004) 501–517 combined characters from both mitochondrial and nuclear genes (Farias et al., 2000). Although taxonomic sampling was extensive within the Neotropical assemblage, the study of Farias et al. (2000) exhibited limited sampling across all other geographic assemblages [e.g., no Malagasy cichlids were included in their analysis combining morphological and molecular evidence (Farias et al., 2000, fig. 4)]. Taxonomic sampling was limited, partly due to an absence of combinable data sets; many included species were sequenced for either mitochondrial or nuclear genes, not both, with the result that the data sets were considered not wholly combinable by the authors. Farias et al. (2000, fig. 4) contend that their results support a ‘‘robust phylogenetic hypothesis for the family’’, however, outside of the Neotropical assemblage, taxonomic sampling is extremely restricted and many recovered clades are not strongly supported. Sparks (2004a) only utilized nucleotide characters from mitochondrial genes in a study that focused on recovering relationships among the Malagasy and South Asian cichlid lineages. In that study, a comprehensive familial-level data set could only be assembled using the 16S fragment, which proved insufficient for robustly recovering early intrafamilial divergences; relationships among the major geographic lineages of cichlid fishes were only weakly supported. Sparks (2004a) recovered a monophyletic Malagasy-South Asian assemblage (with only weak support), and the South Asian cichlids were not monophyletic, based solely on the analysis of 16S nucleotide characters. The African and Neotropical lineages were each monophyletic, although the African clade received only weak support (Sparks, 2004a, figs 2,3). Molecular (and morphological) studies with a focus on recovering early divergences within Cichlidae—those between major Gondwanan assemblages—have met with limited success. All but one study (Farias et al., 2000) have relied on one or two molecular markers, which has proven to be insufficient for robustly recovering higherlevel intrafamilial relationships (e.g., Zardoya et al., 1996; Streelman et al., 1998; Farias et al., 1999, 2001; Sparks, 2004a). The combined molecular and total evidence phylogenies of Farias et al. (2000, figs 3, 4) do not include a single ptychochromine cichlid (Oxylapia, Ptychochromis, and Ptychochromoides), and include only two and one etropline (Paretroplus and Etroplus) species, respectively. No family level hypothesis of cichlid intrarelationships to date, based on equitable family wide taxonomic sampling, has shown strong support for a transoceanic sister-group relationship between major geographic assemblages of cichlid fishes (e.g., AfricanNeotropical), less the well-supported sister-group relationship that is consistently recovered between the Malagasy (Paretroplus) and South Asian (Etroplus) etropline lineages (Sparks, 2004a). Recovering the cichlid sister group has also been problematic, which is as much a result of dogmatic thinking (e.g., assuming monophyly of Labroidei and only including labroid lineages in molecular systematic studies) as it is due to the overall morphological similarity of many percomorph lineages (noted by many researchers, e.g., Stiassny, 1981; Johnson and Patterson, 1993; Johnson, 1993). Monophyly of the suborder Labroidei, an assemblage comprising Cichlidae, Pomacentridae, Embiotocidae, Odacidae, Labridae, and Scaridae, has been hypothesized on the basis of several features of the pharyngeal jaw apparatus (Kaufman and Liem, 1982; Stiassny and Jensen, 1987). Monophyly of the suborder has nevertheless been questioned (e.g., Johnson, 1993; Streelman and Karl, 1997), given that some of these traits are reported to occur outside of the labroid lineages, others are lacking in all members of the assemblage, and outside of features of the pharyngeal jaw apparatus, additional corroborative morphological evidence for their monophyly is lacking (Stiassny and Jensen, 1987; Johnson, 1993; Streelman and Karl, 1997). Results of a recent molecular phylogenetic study likewise suggest that Labroidei is not monophyletic (Streelman and Karl, 1997). Only Stiassny (1982) has looked extensively at non-labroid lineages in the context of placing Cichlidae within Perciformes, and no molecular study of cichlid fishes has presented a robust test of their monophyly, usually limiting outgroups to pomacentrids and embiotocids, or has attempted to place the family within Perciformes. Surprisingly, no family level molecular study of cichlids to date has included a nonperciform outgroup, and almost none have sampled outside of ‘‘Labroidei’’. Our primary objective was to use multiple nuclear and mitochondrial genes, with markedly different rates of evolution, to recover a well-supported family level phylogeny for Cichlidae. A broad range of both perciform and non-perciform outgroups were included in order to provide a robust test of ingroup monophyly and place Cichlidae within Perciformes, as well as to attempt to recover the cichlid sister group (Table 1). Within the context of this phylogeny, we addressed explicit macroevolutionary questions regarding the diversification and distribution of cichlids. Our approach was explicitly to provide robust tests of the monophyly of the Neotropical or African assemblages by including all major lineages within each respective continental assemblage. It was not our intention to resolve generic-level relationships within these clades, which is a task well beyond the scope of this analysis. Many comprehensive phylogenetic studies have been published that focus specifically on major lineages within these species-rich clades. However, by establishing a robust phylogenetic framework for Cichlidae, finer-scale studies aimed at sorting out the intrarelationships within these geographic assemblages will 502 J. S. Sparks & Wm. L. Smith / Cladistics 20 (2004) 501–517