Flowers in phylogenetic and evolutionary studies

The role of flowers in evolutionary biology has changed in the past 20 years, as the major foci are constantly changing with new approaches and better understanding of evolutionary processes. The revolution of molecular phylogenetics and molecular developmental genetics produced a trend in flower studies away from phylogenetics and towards evolution. In turn, the discovery of many well-preserved Cretaceous fossil flowers led to a new trend in flower studies towards phylogenetics, because fossil flowers do not provide DNA. The following three current fields of flower structural studies may be distinguished:

1. Comparative morphological analysis of flowers – Many new major angiosperm clades have been recognized by molecular phylogenetic studies since Chase et al. (1993), as surveyed in APG (1998, 2009), Stevens (2001 onwards) and Soltis et al. (2005). These new clades need now to be critically studied comparatively in their structure and biology as they are largely unknown (e.g. Endress and Matthews, 2006; Endress, 2010a).

2. Morphology for phylogenetic studies – Flowers were generally used for phylogenetic studies in the era before the molecular revolution. In the past 20 years, phylogenetics has concentrated on molecular approaches, which yield more results in a shorter time than morphology. However, morphological phylogenetic analyses are still performed and yield interesting results, either alone or in combination with molecular analyses (at higher systematic levels, e.g. Nandi et al., 1998; Doyle and Endress, 2000, or lower levels, e.g. Carillo-Reyes et al., 2008; Sweeney, 2008). There has been a pessimistic attitude towards the use of morphological features in phylogenetics because of too much homoplasy (e.g. Givinish and Sytsma, 1997; Patterson and Givnish, 2002; Givnish, 2003; Scotland et al., 2003) and difficulties in scoring structural characters (Stevens, 2000). This is true if superficial structural features that are easy to spot are used (e.g. tepals large and showy versus small and inconspicuous, or fruits capsules versus berries, or storage organs rhizomes versus bulbs). However, morphology encompasses much more than such features. It can be expected that as our knowledge of flowers increases, there will be a resurgence in morphological phylogenetic analyses. In addition, the more fossil flowers become available, the more important morphological phylogenetic analyses will become (e.g. Friis et al., 2009; Doyle and Endress, 2010). There are not only many more fossil flowers available than 20 years ago, but there are also new techniques to reconstruct their morphology: the use of microtome section series (Schönenberger, 2005) and tomography (Friis et al., 2009). The search for and the detection of new structural patterns of interest is a continuing challenge. Characters and character states ‘cannot be defined but need to be discussed,’ as Wagner (2005) put it, meaning that definitions need to be constantly evaluated and updated to fit the current knowledge with each change in the phylogenetic framework. New knowledge on phylogeny (and evolution) continuously creates a new basis for discussion. Of course, if morphological characters are used for phylogenetic studies, this also means the necessity of repeated reciprocal illumination (see also Kelly and Stevenson, 2005). ‘Tree-thinking’ has been encouraged in evolutionary studies (O’Hara, 1988; Donoghue and Sanderson, 1992). This is of course also relevant for the focus on structural features, including the construction of morphological matrices for phylogenetic studies. The more detailed a tree under reconstruction already is and the more detailed our knowledge about the distribution of traits on this tree is, the better we can judge the quality of characters and character states to be scored.

3. Morphology for evolutionary studies – The new phylogenetic results can now be used to study the evolution of flowers on a much more solid basis than was possible before. A general result is that many features are more evolutionarily flexible than previously assumed. A number of examples are surveyed in this study. Rarely is a character more stable than previously assumed at macrosystematic level; such an exception are features of ovules (Endress, 2003, 2005a, 2010). However, such flexibility is not randomly distributed through the larger clades. Given features are more concentrated (but not universal) in a certain clade than in another one. Why is this so? Answers can be expected from better knowledge of the genetic systems that operate in the development of such features (e.g. Borowsky, 2008; Melzer et al., 2008). Thus, homoplasy in structure is pervasive, much more common than earlier imagined and is a fascinating aspect of flower evolution (e.g. Cantino, 1985; Endress, 1996). For more evolutionary aspects of flower morphology, see Endress (1994, 2003, 2005b, 2006).