The present molecular phylogenetic hypothesis does not support the phylogenetic relationships of genera proposed by Kritsky and Boeger (2003) and Vianna (2007). However, the overall phylogenetic relationships of the genera in both ML and BI trees (Fig. 2) are similar to those proposed previous hypotheses (i.e., Boeger et al., 2020; Přikrylová et al., 2021), the present analysis, which included an 18S rRNA gene sequence of a species of Polyclithrum, revealed an unexpected evolutionary dynamic of morphological features. The disclosed scenario of the present hypothesis is richer in morphological homoplasies – mostly representing evolutionary convergences – than recognized previously. These revealed homoplasies may explain the inconsistencies observed between the morphological and molecular phylogenies obtained herein and those of Kritsky and Boeger (2003) and Vianna (2007).
The current phylogenic hypothesis indicates that Polyclithrum sp. does not have a close phylogenetic relationship with Gyrodactyloides, Macrogyrodactylus, Swingleus, and Fundulotrema, as suggested by Kritsky and Boeger (2003) or, in addition, with Mormyrogyrodactylus and Diechodactylus as proposed by Vianna (2007). However, both hypotheses were based solely on morphological characteristics of specimens belonging to these genera. The present hypothesis supports that species of Macrogyrodactylus (composing a basal clade in the evolution of the family, Fig. 2) - Gyrodactyloides, Mormyrogyrodactylus, and Diechodactylus are distantly related to Polyclithrum, Swingleus, and Fundulotrema. Although more closely related than these genera, the clade of Swingleus + Fundulotrema (clade J, Fig. 2, B) and Polyclithrum are not sister groups, as proposed by Kritsky and Boeger (2003) and Vianna (2007) but, according to the present hypothesis (Fig. 2), these genera have closer relationships with other species of the polyphyletic Gyrodactylus, respectively (Fig. 2).
Incongruences between the morphology-based and molecular-based phylogenies may originate from either or both databases. Incongruences may result from “(i) homoplasy (nondivergent evolutionary change), which is common when a character has a limited number of alternative states and evolves rapidly relative to the phylogenetic events being evaluated, and (ii) incongruence between the phylogeny of the characters and that of the species in which they are measured” (Pamilo & Nei, 1988). For instance, hypotheses generated based on the 18S rDNA (see Boeger et al., 2020; Prikrylova et al., 2021; Přikrylová et al., 2013), representing distinct assumptions and criteria, resulted in differences in the phylogenetic relationship among species of several included genera. That poses a question on the strength of the phylogenetic signal of this fragment for the desired level of analyses - even though the present molecular alignment was not shown to be saturated – and the adequate correspondence between gene and species phylogeny.
Also, the reconstruction of the evolution of morphological characters (Fig. 3) revealed that the incongruences between morphological and molecular phylogenies questions the nature of the homologous series proposed by Kritsky and Boeger (2003), which were also the basis for the hypothesis presented by Vianna (2007). Despite of some striking similarities between expressed morphological states, some represent extensive convergent evolution shared between Polyclithrum and the genera proposed as closely related by Kritsky and Boeger (2003) and Vianna (2007).
Indeed, the use of morphological features alone for phylogenetic reconstruction of groups, such as Gyrodactylidae, are limited by the small number of potential morphological character series. These small databases hinder the testing of a priori hypotheses about the homology of states determined by the Hennig’s Auxiliary Principle (Hennig, 1966; Wiley & Lieberman, 2011). (Hennig, 1966) defined his “auxiliary principle” which can be stated simply as “characters meeting various similarity criteria (relative qualitative identity) are to be considered homologous unless evidence is presented that demonstrates that they are homoplasious”. These hypotheses in homology are subsequently tested through the process of phylogenetic reconstruction (Hennig, 1966). However, although small datasets, such as those of Kritsky and Boeger (2003) and Vianna (2007) can recognize many instances of homoplasy, they are often too small to robustly test initial hypotheses of homology, and may result in unstable scenarios of history of character states and phylogenetic relationships among species. As suggested by Wake (1991), “when homoplasy is rampant, we can expect discordance with phylogenetic analyses based on nonmorphological data sets.”
Phylogenetic conflicts between distinct databases constitute a feature of the process of erection of phylogenetic hypotheses (Dávalos et al., 2012). For these authors, “to overcome phylogenetic conflict, biologists will need to understand the processes that bias all data types, and this need will only grow as phylogenies move from single-system or gene to multiple systems and loci.” These biological processes are relatively well-known for molecular data and they have been accounted for in the reconstruction of phylogenetic hypotheses using models of nucleotide substitutions (e.g. Bergsten, 2005; Masters, 2007; Philippe et al., 2005; Susko & Roger, 2021).
For morphology, however, the situation is far more complex as it may include several possible mechanistic processes (Collin & Miglietta, 2008; Cronk, 2009; Mitsiadis et al., 2006). Some of these processes are likely developmental, associated with the retention of the genetic basis for the development of complex structures (Cronk, 2009). This appears to be the case for the evolutionary pathway of many features in Gyrodactylidae and likely explains the large number of contradictions in the phylogenetic analyses produced independently by various morphological and molecular datasets. It is, however, difficult to account for these processes in the construction of phylogenetic hypotheses as it requires reasonable knowledge on the genetic and structural mechanisms that influence the evolution of morphological features within the focal biological group.
Hall (2007) suggests that it should not be “surprising to find that different environments or selective pressures can trigger the reappearance of similar features in organisms that do not share a recent common ancestor (homoplasy).” However, the expression of specific homoplasies, under these scenarios, may indicate an underlying similarity in the developmental processes involved. Hence, the relative conservative nature of the evolutionary process suggests that specific homoplasies may present a higher probability to occur in closely related species (Arendt & Reznick, 2008; Mueller et al., 2004; Wake, 1991). This is supported by the suggestion of Mueller et al. (2004) and Wake (1991) that extensive homoplasy is expected also in lineages depicting a limited set of putative homologous morphological features, as in the case for Gyrodactylidae (as suggested above).
Hence, morphological homoplasies may be also informative in phylogenetic reconstructions. This perspective differs from previous attempts to handle homoplasies in the construction of phylogenetic hypotheses by down-weighting their informative value (as suggested by Goloboff et al., 2008). However, to manage putative homoplasies in the manner suggested herein would require likely a more comprehensive understanding of gene composition, interaction, and expression within a specific group – this is far from being the case for Gyrodactylidae and many other groups of parasitic species.
While we do not understand the developmental mechanisms that determine the expression of phenotypes, “especially in groups where morphology matrices are limited, a heavy bias towards informative, independent genomic sequence characters will probably dominate phylogenetic hypotheses” (Ragsdale & Baldwin, 2010). The accumulation of genomic data should provide a less biased and more extensive source of data for the construction of phylogenetic hypotheses within Gyrodactylidae. The reconstruction of the historical relationships among the morphological features onto the molecular phylogenetic hypotheses (as in Boeger et al., 2014; 2020), in special using multilocus genetic data, should generate a stronger understanding about the evolution of the species groups and their corresponding morphological features.