Strong character incongruence and character choice in phylogeny of sea stars of the Asterinidae

Historically, characters from early animal development have been a potentially rich source of phylogenetic information, but many traits associated with the gametes and larval stages of animals with complex life cycles are widely suspected to have evolved frequent convergent similarities. Such convergences will confound true phylogenetic relationships. We compared phylogenetic inferences based on early life history traits with those from mitochondrial DNA sequences for sea stars in the genera Asterina, Cryptasterina, and Patiriella (Valvatida: Asterinidae). Analysis of these two character sets produced phylogenies that shared few clades. We quantified the degree of homoplasy in each character set when mapped onto the phylogeny inferred from the alternative characters. The incongruence between early life history and nucleotide characters implies more homoplasy in the life history character set. We suggest that the early life history traits in this case are most likely to be misleading as phylogenetic characters because simple adaptive models predict convergence in early life histories. We show that adding early life history characters may slightly improve a phylogeny based on nucleotide sequences, but adding nucleotide characters may be critically important to improving inferences from phylogenies based on early life history characters. Additional key words: total evidence, modes of development, life history, mtDNA, Asteroidea The choice of characters for use in phylogenetic analyses is an important and contentious issue in animal phylogenetics. Since Haeckel (1866), phenotypic characters from the earliest parts of the life cycle (such as the morphological forms of planktonic larvae) have been viewed as a potentially rich source of phylogenetic information. However, even in the early 20th century, there was growing skepticism of the utility of early life history characters for tracing evolutionary relationships (e.g., Conklin 1928; Gould 1977). The subsequent growth of larval ecology as a research discipline led to renewed interest in the phylogenetic history of embryological and larval diversity, with a sharp focus on the study of early life history characters expressed during sexual reproduction or during larval development before metamorphosis to the adult form (McHugh & Rouse 1998). Molecular phylogenetic studies of taxa that include species with diverse early a Author for correspondence. E-mail: [email protected] b Present address: Department of Biological Sciences, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada life histories have suggested the evolution of convergent similarities in early life history characters such as egg size, type of parental care, offspring dispersal ability, larval morphology, and fertilization ecology (Hart 2000). Many of these characters reflect functional adaptations of a complex life cycle in which individuals sequentially inhabit planktonic and benthic habitats: the similarities in character states could represent shared ancestral states as a result of constraints on the evolution of early development, shared derived states that evolved prior to speciation events, or convergences that evolved in response to selection (Strathmann 1985; Haszprunar et al. 1995; Wray 1995, 1996; Raff 1996; Smith 1997). The extent of convergence—and the soundness of homology arguments for early life history character states—is often difficult to establish (Strathmann & Eernisse 1994; Haszprunar et al. 1995; Wray 1996; Rouse 1999, 2000). For these reasons, many invertebrate zoologists (see McHugh & Rouse 1998; Hart 2000) have recently chosen molecular characters over characters drawn from reproductive phenotypes or the morphology of embryos and larvae for the evaluation of evolutionary relationships. Name /inbi/123_406 10/19/2004 11:54AM Plate # 0-Composite pg 344 # 2 344 Hart, Johnson, Addison, & Byrne File # 06em Allen Press • DTPro System GALLEY 344 In cases where several types of phylogenetic characters are available, or different characters can be used to address complementary problems, a potentially powerful approach uses all available molecular and phenotypic characters in a total evidence analysis (Hillis 1987; Kluge 1989; Shaffer et al. 1991; Eernisse et al. 1992; Wheeler et al. 1993; de Queiroz 1996; Moritz & Hillis 1996; Levasseur & Lapointe 2001). Nucleotide characters are especially useful where phenotypic homologies are not clear (Raff et al. 1994; Aguinaldo et al. 1997; McHugh 1997) or few shared derived phenotypic states can be observed (Avise et al. 1994). Molecular and phenotypic data often give results that are congruent or at least not strongly conflicting (Eernisse et al. 1992; Rumbak et al. 1994; Lafay et al. 1995; Littlewood & Smith 1995; Paterson et al. 1995; Reid et al. 1996), and combining these character types may allow the inclusion of fossil or other taxa for which molecular data are unavailable (e.g., Wiens 1998). The total evidence approach has sometimes been cautiously extended to include phenotypic characters from marine invertebrate gametes, embryos, and larvae (Reid 1990; Rouse & Fitzhugh 1994; Smith et al. 1995; Reid et al. 1996; Wray 1996). In spite of the general suspicion that early life history characters of marine invertebrates might be misleading sources of phylogenetic information, no quantitative analysis has shown the extent of conflict—or consistency—between these characters and phylogenetic information from nucleotides or other molecular characters. Here we give a simple quantitative analysis of congruence between early life history characters and mitochondrial DNA (mtDNA) sequences for asterinid sea stars (including new sequence data for five species). Many asterinid species share apparent derived forms of reproduction, larval morphology, and brood protection that may reflect either homology or convergence (Byrne & Cerra 1996; Hart et al. 1997; Byrne et al. 1999). We use this quantitative analysis to illustrate the extreme conflict between early life history and nucleotide characters, the risks associated with phylogenies based on such characters, and the scant benefits of including certain early life history characters in total evidence analyses.