Dynamic Homology and Phylogenetic Systematics: A Unified Approach Using POY
Año: 2006 p. 365
Phylogenetic systematics demands a general and logically consistent analytical framework. To meet that demand, we propose here a system of testing phylogenetic hierarchies with the broadest possible diversity of evidence. Traditionally, phylogenetic analyses were based entirely on phenotypic evidence derived from such sources as comparative morphology, molecular biology, and ethology, which entailed only a few distinct character types. As genomic data have become more readily available, however, the diversity of character types has increased dramatically to include complex genetic, chromosomal, and even entire genome sequences. Fundamentally, there is no evidentiary distinction to be made among these sources of information. Logical consistency requires that all evidence be treated equivalently. At present, there is no way to test phylogenetic hypotheses with the full diversity of available evidence in a consistent manner, a shortcoming we aim to correct here. The incorporation of genotypic character types into systematics requires a fundamental rethinking of the way phylogenetic hypotheses are tested and inferences made. A central distinction in this book is made between the static homology approach, where data are encoded (or aligned) in a fixed character matrix prior to phylogenetic analysis, and the dynamic homology approach, where constraints on possible transformations arel reduced and characters (transformation series) are inferred a posteriori as a result of phylogenetic analysis. The ideas in this book have been implemented in the computer pro-gram POY, version 3.012, written by Wheeler and colleagues (Wheeler et al. 1996–2003). POY is an open source ( pub/molecular/poy) program written in ML and C. Precompiled binaries are available for Linux x86, MS-Windows, and Mac OS X. The code is fully parallelized (using PVM and MPI) and fault tolerant with multiple program options and algorithms to take advantage of a diver-sity of multiprocessor platforms, including clusters and shared memory multiprocessor (SMP) machines. In the first part of this book, we discuss the theoretical core of phyloge-netic systematics as it relates to POY. Given a minimal set of evolution-ary assumptions, a definable (if large) collection of possible hypotheses is tested with evidence derived from observed biological variation. We are not attempting to be exhaustive in our discussions, but to focus on the issues relevant to POY and its motivation. Readers should go to more general sources (e.g. Schuh 2000) for broader treatments. The second part presents the problems faced by systematic analysis and discusses their solutions in both theoretical and practical, applied terms. Exact solutions will be few and far between, given the computational complexity of the problem. We are limited, largely, to heuristic solutions that take advantage of distributed computing. We discuss the trade-offs associated with each approach, as well as the optimal search strategies given different phylogenetic problems and computer resources. The third part is a users’ guide for the implementation of these ideas in the program POY, including program installation instructions for multi-ple platforms, descriptions of over 250 user-specified commands, and instructions for use. Finally, as may be expected in any field of science confronted with new problems of unanticipated complexity, a diversity of possible solutions exists, and it is only by engaging in open scientific debate that false theo-ries and logically inconsistent methods are eliminated and progress made. Given the infancy of phylogenetic analysis in the genomic era, it should come as no surprise that significant disagreements persist among the authors of this book. Although some amount of compromise is required in collaborative projects, such as this book, and it is impractical and inappropriate to debate our differences of opinion exhaustively, we believe it would be intellectually dishonest to overlook substantive dis-agreements altogether. The reader should look to the primary literature for more fully developed arguments. Nothing is final and irrefutable in science. We believe our open criticism to be a necessary aspect of the An Approach to scientific process that is all too often hidden from the reader in an attempt to present a unified front.