Analysis of Morphological Character Evolution
in Neotropical Polystichum
Introduction: here I report the detailed results of a pilot analysis of morphological character variation in continental neotropical Polystichum. In this work I sought to interpret the course of evolution in morphological characters in neotropical Polystichum by determining the most parsimonious set of changes, given that our three-sequence molecular phylogeny of the same study set is taken as the closest approach to the true phylogeny. In other words, morphological change was not included in the construction of the phylogeny I used for the experiment.
Methods: I scored a set of 38 morphological characters (List of Characters) for thirteen Neotropical species of the fern Polystichum. The data for most of these characters was measured; a few were presence-absence. Color requires further explanation. I score color of scales against color chips from a National Bureaur of Standard color-chip set. These have a Munsell color-system three-dimensional value, which is summarized in a single NBS chip number. For this analysis I just used the single number; for the full analysis I will use the three Munsell-dimension numbers.
I first did an exploratory principal-components analysis of the study set to survey for 1) morphological characters that were effective in separating the taxa and 2) groupings of taxa that were consistent or not with the molecular clades.
Then I scored the morphology of plants of all the neotropical species in the three-sequence phylogeny. These I entered into a Morphocode file. I ran Morphocode, which did the artificial delineation of discontinuous character states (Link to database with original and converted character states). The yield was a nexus file, which I then subjected to an MP analysis using PAUP*, which resulted in a completely unresolved consensus tree. I opened this nexus file in MacClade and manipulated a bush tree into our 3-gene phylogeny, with the result that the morphometric characters were optimized onto the molecular tree. From my MacClade printout of the morphological characters forced onto the three-sequence molecular phylogeny I scanned for prominent characters. Petiole-scale hue seemed a useful model for addressing the impact of the program on continuous data, so I put the original Munsell code numbers on the tree. Then, from my intuitive understanding of ambiguities in color coding and the similarities of the different Munsell numbers, I assigned color names to the nodes, along with bicolorous-concolorous condition.
Results: The principal-component analysis revealed that the Mayan (indusiate) and Andean (exindusiate) clades did indeed cluster in different parts of the array. (Principal Components Analysis Plot) Within these clusters, the positioning of species was in general consistent with the results of the cladistic analysis of molecular data. However, the elevationally egregious P. platyphyllym was distant from its molecular allies and peripheral in the array.
From the Morphocode analysis, petiole-scale hue in continental-neotropical Polystichum has evolved as follows (See Tree). The ancestors of the lineage had pale, concolorous scales (which P. lonchitis and the West-Indian lime-dwellers share with many north-temperate polystichums). Once on the continent, pigment becomes more intense. First, the scales become cinnamon-rufous; at the point P. speciosissimum diverges and pursues pigment intensity to the extreme among tropical American species. The common ancestor of the remaining species darkened—but with the center darkening more, yielding a bicolorous scale. In both the Mayan (indusiate) and Andean (exindusiate) lineages the edge continued to darken, yielding a dark concolorous scale (the darkening could have happened in the ancestor, eliminating the transient appearance of bicolorous scales—perhaps the program is set to delay transformations). This ancestral scale is present or virtually unaltered in four of the eight species in these two clades. In the Mayan center, the scale edge becomes pale in the lineage including P. muricatum and P. turrialbae (this is where the cilia get short and eventually disappear and the plants get small). Bicolorous scales also appear in the Andes in P. montevidense. The hypertransforming P. platyphyllum returns to a deep ancestral condition with paler, but still concolorous scales. The scale transformations are plotted on the Fig. 2 of the grant proposal.
Discussion: Most prominent is the stability of concolorous scales – its transformation but twice to bicolorous, and the general trend to intensifying and darkening. It is tempting to evoke the higher ultraviolet contribution at high elevations as a selective force favoring the development of these pigments—and of the intense darkening of these scales as the factor that favored the success of the main montane lineage relative to the species-poor speciosissimum lineage. The loss of these pigments in the deep-forest turrialbae lineage is consistent with this hypothesis.
The low-elevation eccentric P. platyphyllum deserves special attention. Again, the petiole scales are paler, suggesting an economy with the loss of intense radiation typical of higher elevations. This species is not only low-elevation, but it is a rheophyte (along with the so-far unsampled P. dubium, which I expect is a close ally). Perhaps the whole set of transformations that make this species such an eccentric relate to the transition to the streambank environment: it thrives on the little disturbed earthen banks above watercourses. The bulbils, unusual among continental neotropical species (PP. bulbiferum needs attention here) may also relate to this habitat specialization.
