|
Modern Evolutionary Systematics: A New Introduction August 12, 2009 |
|
MODERN EVOLUTIONARY SYSTEMATICS: A NEW INTRODUCTION Contents: 2. A short note on consistency 4. Phylogenetics
and evolutionary systematics compared 5. Example of extinct paraphyly 8. Phylogenetics and creationism 9. Quotations 10. Other authors 11. Manifesto and additional comments 12. A new framework The original introduction
way somewhat discursive and unfocused; this is a new, more readable version,
with some new, hopefully interesting material. Evolutionary systematics,
or evolutionary taxonomy, is the science of apprehending nature through a
naming system of nested groups, with the species as basic unit of
classification, using the Linnaean system and, to the extent possible, what
is known about evolutionary relationships.
Evolutionary systematics uses both
sister-group relationships and ancestor-descendant relationships as
recommended by This is in contradistinction to the now popular
phylogenetic system of Hennig, which focuses
exclusively on sister-group relationships (splitting of lineages). Phylogenetics now
incorporates powerful analytic tools, including statistical analysis of
molecular data, but the elimination of representations of macroevolution in
phylogenetic trees has led to various problems that have been pointed out in
recent literature. Major examples of this critical literature are gathered
together on this Web site. Is phylogenetics truly
revolutionary as it is commonly presented? If it were, the present author
would be at the first barricade waving a red flag. Modern evolutionary systematics
uses recent developments in analyzing evolutionary relationships but extends
such into evaluations of the results of anagenetic
change leading to macroevolution, and presents both classifications and trees
of life that reflect both sister-group and ancestor-descendant relationships.
All evolutionary systematists
generally agree that recognition of paraphyletic
groups (ancestral groups denied recognition at separate and taxonomically
equal rank to that of their descendants by phylogeneticists)
contain important evolutionary information that should be represented in
classifications. Well-known examples are the sinking or attempted sinking of
Aves into Reptilia, or Cactaceae
into Portulacaceae. Detailing macroevolution in
classification through recognition of paraphyletic
groups is fundamental to evolutionary systematics.
Macroevolution is a real scientific concept supported by plenty of data
showing autophyletic lineages arising from the
midst of paraphyletic lineages. Phylogenetics is methologically
inconsistent. For example, one taxon represented by
two OTUs may appear on a molecular tree
(phylogenetic paraphyly or apparent polyphyly). Given
the structuralist justification for evolutionary
classification, the only explanation must be “convergence.” But a cladogram does not diagnose exactly what different
ancestors the two paraphyletic OTUs
came from. The explanation is inconsistent with the method because it uses a
different method. An alternative evolutionary systematics
explanation is the scientific theory of a deep shared ancestor with the same
diagnosis inclusive of both OTUs, which is
consistent. A morphological cladogram may
be different from a molecular cladogram, and both
may be intuitively convincing, the one clarifying morphological changes, and
the other gene changes. Phylogenetics cannot
explain this inconsistency, but tries to combine both in “total evidence”
analysis, lumping all data together. This combines
inconsistent results into a jumble where strong signals of one process
(mostly non-coding changes in DNA) overwhelm, usually, weak signals of
another, different process (fixation of expressed traits). An alternative
evolutionary systematics over-arching theory is
that taxa basal in a morphological cladogram but terminal in a molecular cladogram
signal the status of that taxon as surviving
ancestor of possibly many lineages (a kind of coelacanth). This last theory
renders the differences between morphological and molecular analyses consistent.
It is only a theory, but this is
far better as a basis for classification than the apparent deductive
perfection of a phylogenetic cladogram. [Added March 7, 2011.] There are three
patterns that are reconciled by pluralistic evolutionary systematics. These patterns give three different views of
the evolutionary process, and are not equivalent, thus one is not better at
charting evolution than the others. Assuming
that one pattern is fundamental and other patterns and data may be relegated
to that one fundamental pattern is “structuralism.” Structuralism as a
retrogressive force in modern systematics is
discussed at length in this
paper. The
three patterns are: (1)
Classical systematics produces classifications of macroevolutionarily distinguished hierarchical sets of taxa. These are generated through taxa
distinguished by overall similarity of conservative and apparently homologous
traits, and distinction of major evolutionary transformations at the taxon level associated with habitat change and other
criteria. Conservative traits are those stable at different collecting sites
and across different habitats. (2)
Morphological cladograms presenting nested sets of
trait transformations away from an apparent primitive (plesiomorphic)
state usually represented by an outgroup taxon. Macroevolution is represented only by the
exemplars named through classical systematics,
while renaming of such taxa is solely based on
present-day patterns. The transforming traits are, usually, conservative
characters from classical descriptions. (3)
Molecular trees are variously assembled through parsimony analysis (as in
morphology), or maximum likelihood or Markov chain Some comparisons of phylogenetic and evolutionary systematics: Paraphyly is a phylogenetic somewhat disparaging word for what is
generally known as ancestors involved in macroevolution, that is, a label for
a group from which one or more other groups at the same (or higher) taxonomic
rank have apparently evolved. “Para”
implies faulty, wrong, amiss, recrementitious, or
merely similar to the true form. Evolutionary systematists,
however, celebrate that which is presently known as paraphyly. Thus, Aves is an autophyletic
product of the paraphyletic Reptilia,
while Cactaceae is autophyletic
to the paraphyletic Portulacaceae,
and Cinclidotaceae, Ephemeraceae,
and Splachnobryaceae are all autophyletic
to the large, extended paraphyletic group Pottiaceae (see PDF reprints of Zander 2007, 2008). All these taxa
are euphyletic. Thusly orthodox systematics
can use a more robust terminology to match the tendentious terms of parasystematics (phylogenetics,
phenetics, or any system that offers a quick and
easy method to simplify an unsimplifyable field of
research). A natural taxon (I’m sure you will ask) is any group that is
probabilistically the best representation of the expressed traits as an
evolutionary trajectory at that taxonomic rank. When split to fit a molecular
tree, for a split to be a natural taxon it must be
significantly more robustly supported by expressed traits than any other
split in any other way of splitting. This is because some expressed traits (a
combination by chance alone) may support any split that may have appeared in
a molecular cladogram. The traits should include autapomorphic traits and any traits that particularly fit
the organism to a particular environment. Finding some expressed traits by
chance alone that support a taxon in any particular
molecular clade is a problem in statistics called
“multiple comparisons.” A
morphological cladogram is best viewed as a step
towards a natural key, inasmuch as it often conflicts with molecular cladograms, but it does reflect at least present-day
relationships based on expressed traits. Likewise, when a natural taxon
is combined with another natural taxon, the result
may be clearly unsatisfactory inasmuch as important (at that rank)
evolutionary features are not represented in classification, and the
resultant salmagundi is “dysphyletic.” Consider a terminal branch of a molecular cladogram (((((A,B)C)D)E)A)
. . . . An exemplar of taxon A is at the far end
and another exemplar of Taxon A is also at the base
of the branch. A theory could be made that Taxon A
constitutes the taxon from which B, C, D, and E all
evolved. Consider the same branch but the lowermost A lineage is
extinct. The actual evolution remains the same. Consider the same branch with
only the uppermost A lineage extinct. The evolution still remains the
same. In phylogenetics, the
classification would be different for the different scenarios of extinction
of Taxon A.
The aim of evolutionary systematics is to
use other information (e.g. biosystematics, geography) such that the
classification of the two scenarios is the same. A basic question is how many branches in between the two
A lineages are tolerable? If too many, then the two As doubtfully infer the
complete continuity of Taxon A. And, if only one
very odd taxon is nested in a large group, it is
doubtful that there is an extinct lineage of the odd taxon
at the base of that large group, and therefore the odd taxon
evolved directly from the large group. An example in the mosses is the recent
removal of the small family Ephemeraceae to deep
nesting in the large Pottiaceae family. Ephemeraceae probably directly evolved from the Pottiaceae because of the many nodes (= deep nesting)
between the Ephemeraceae and the base of the Pottiaceae cladogram, i.e. the
chance of an extinct Ephemeraceae lineage below the
Pottiaceae is very small. Can the amount of extinct paraphyly
be estimated by the amount of paraphyly in extant
groups? What do you think? I have been
prompted to develop this Web site by several kinds of outrages over the
years, some admittedly theoretic and moot, some terribly obvious and
problematic, particularly: (1)
the elimination in phylogenetic classifications of any aspect of
macroevolution, resulting in synonymy and splitting of what are apparently
well-founded species and higher taxa of organisms
through the phylogenetic classification principle of holophyly
(which reflects nothing in nature); (2)
the assumption of no surviving ancestors, or that one taxon
cannot be in two different molecular lineages, though there is plenty of
evidence that this may result from punctuated equilibrium followed by long
stasis of taxa governed by stabilizing selection -
there may be in fact up to n − 1 terminal taxa
that are ancestral to at least one other terminal taxon.
Using the word “taxon” rather than “species” or
“population” is due to my expectation that genera and higher taxa do evolve as units through joint species’ reference
to the paragenetic regulatory functions of a shared
particular selective regime (an “envirosome”); (3)
parsimony using morphological traits assumes independent and unique
distributions, yet selection and epigenetic effects commonly link traits such
that they may be fixed during speciation as a unit; (4)
avoidance of weighting of traits reflects the
“automatic classification” philosophy originating with pheneticists
and continued by phylogeneticists as “theory-free”
taxonomy, such that a parsimony analysis of morphology is in reality based on
raw similarity, lacking phyletic weighting -
similar problems occur with molecular traits. If phylogenetics
is truly a theory-free discovery process then there is no hypothesis to test,
since the pattern discovered is a fact. Phylogenetics
is theoretically speaking bankrupt and has always been so. Is phylogenetics
“to big to fail”?; (5)
molecular analysis either assumes that traits are independent or if dependent
then of taxonomic importance anyway, yet convergence of molecular traits with
morphological ones because of similar convergent selection supports a wrong
tree, and convergence of molecular traits with each other because of similar
selection may well support a wrong molecular tree over a poorly supported but
correct morphological tree. Pervasive selection on non-coding DNA traits has
been reported in the literature, while entire coding genes, regulator genes, and promoter regions are
commonly included in molecular analyses; and (6) the common elimination of the environment from
consideration in phylogenetic analysis.
E.g., following the mechanisms for macroevolution of genera and higher
taxa summarized by Gould (2002) in his opus, the
environment acts as a regulator and guide (external chromosome equivalent) of
shared change among species of the higher taxa
through stabilizing and disruptive selection acting on individual species or
on traits shared by all species. This shared feature is not represented in
the analytic processes of phylogenetic analysis, but is part of the expected
judgment involved in classification by evolutionary systematists
familiar with the habitats of organisms of their specialization. A few points summarize my own take on modern
evolutionary systematics: (1)
Alpha taxonomy plus biosystematic and ecological
study yield well-conceived taxa as apparent results
of past evolution, these correspond to what the “exemplars” of phylogenetics represent; (2)
phylogenetic sister-group analysis is powerful and effective, but is crippled
by insistence on classification by holophyly (a
simplifying classification principle corresponding to no thing in nature); (3)
molecular trees demonstrate lineage continuity and isolation but not
necessarily speciation; (4)
macroevolution is being unfortunately eliminated from classification by
phylogenetic insistence on (a) classification by holophyly
and (b) not naming ancestral nodes (because this would create paraphyly); (5)
elimination of macroevolution results in phylogenetic trees without names for
ancestral nodes leading to clades but no caulis,
i.e. a phylogenetic Tree of Life may be totally replaced by a Nested
Parentheses of Life; (6)
taxa may be found in two (or more) different
molecular lineages, commonly as surviving ancestors, because of stasis
associated with punctuated evolution may also follow isolation of two
populations that remain identical at some taxonomic level, followed by
budding evolution, and splitting of lineages is not necessarily accompanied
by speciation; and
(7) stabilizing evolution on morphology and interaction of expressed traits
with the environment may be decoupled from gradual accumulation of changes in
the genome, such as apparently non-coding traits that are used to track
continuity and splitting of lineages.
There is considerable evidence from molecular analyses that this is
true, e.g. “cryptic” species, genera and families. Limits to resolution of molecular phylogenetic analyses by potential
of extinct paraphyly: "Phylogenetically informative" may prove somewhat of
an oxymoron. This is because empty precision leads to aleatory
classification. This is how: In parsimony of morphology, traits are not
necessarily tacked onto a taxon as speciation
gradually continues, but an initial linked set may be necessary for selection
into a new environment. Thus, if A and B share three traits that are
selectively linked, and A and C share two traits
that are not (maybe neutral or sequentially added as the environment changes
over time), then A and C probabilistically share the latest ancestor, not A
and B. Although when dealing with masses of shared traits, main clusters of a
parsimony cladogram may be okay or acceptably
approximate, parsimonious decisions about relationships of small groups of OUT's may need additional information, but are for now cladogrammed by chance. In
molecular analyses, any sister group pair may have had an extinct lineage
identical in phenotype to one of the sister groups occurring below the split.
If so, then this is not a sister group relationships but ancestor-descendant
relationship instead. If the extinct lineage identical in phenotype to one of
the sister groups is even farther down in the tree (phylogenetic polyphyly, or if within reasonable partristic
distance then extended paraphyly) then the
molecular tracking of splits in the gene history is further compromised. The results of macroevolution can give
often statistically certain sets of lineages of present-day specimens
(exemplars). But should we use these patterns for classification. Because
macroevolution involving progenitors in stasis shuffle lineages of taxa, even
trying to “fix” the pattern by renaming taxa that
are out of order does not give a classification that reflects evolution well.
Only by going beyond pattern, and
using pattern to help infer process, can
we create a classification that is not often plain wrong. The
evolutionary story has been lost to reductionism in ignoring all information
on evolution not in a database of phylogenetically
informative traits, and to irredeemably faulty methods of analyzing evolution
and assessing classification. Phylogenetic analyses might possibly be saved
as modified with information from chromosome counts, ecology, biogeography, phyletic weighting of traits, and genomics, among other
information. Renaming taxa that occur in two or
more different lineages or lumping paraphyletic
groups with their autophyletic macroevolutionary
products is just ignoring significant evolutionary information to preserve
assumptions that are contrary to reality (e.g. "a taxon
cannot be in two molecular lineages at once") and save the hyperexact Method. Systematic
analysis is based on two steps, evaluation of evolutionary relationships and
interpretation in classification. If both morphological and molecular data
are poorly dealt with in software, the first in that many are selected for in
packages of two or more traits, the second in that extinct molecular paraphyly and extended paraphyly
may be common, then highly precise analysis based on such data is futile. A
method that does not necessarily produce maximum parsimony, like neighbor
joining, may be sufficient to ascertain the gross clusters of extant taxa that reveal evolutionary processes at acceptable
resolution. On-going wrangling over which software yields the shortest
morphological or molecular tree or the most credible Bayesian molecular tree
is doubtless useless for evolutionary detail because the most detailed
results in both the phylogenetic tree and in resulting classifications are aleatory. Extinct
paraphyly is a problem with resolution of sequence
of molecular lineage splitting. The resolution of a molecular tree depends on
distinguishing extended paraphyly, i.e. a
reasonable inference of a deep shared ancestor (evolutionary monophyly) from evolutionary polyphyly
(no reasonable inference of a deep shared ancestor). The question is whether
any particular sister group is or is not the remnant of a paraphyletic
ancestor, which would affect accuracy of mapping of expressed traits or taxa on the molecular tree. Without additional
information like relative age of the groups involved, the best guide is the
extent of paraphyly or extended paraphyly,
by some measure of patristic distance, of related extant natural taxa. Without
other data, a cladogram with 10 percent of the
nodes exhibiting paraphyly in extant taxa may indicate that 10 percent of the ancestral nodes
at any past time were also paraphyletic, and all
present-day nodes are then only 90% credible due to this one problem alone.
Individual lineages that are well-supported by bootstrapping or credible
intervals are in no way immune to this problem. Other data possibly of value
in evolutionary analysis preliminary to classification include various autapomorphic (phylogenetically
uninformative) traits, paleontology, chemistry, ecology, biogeography,
chromosome numbers, and any other information that might throw light on
ancestor-descendant relationships of accepted or natural taxa.
Also relevant here is recent work on irreversible traits (Bridgham
et al. 2009). See
also “Short Essays on Macroevolution in
Classification” and “Mathematics, Science and
Phylogenetics Compared: An Essay.” It is easy to criticize
but harder to offer a workable alternative. Extracted from Structuralism in Phylogenetic Systematics are these elements of a pluralist
alternative to phylogenetic systematics: (1) Classical alpha
taxonomy uses hard-won informal genetic algorithms as a heuristically based expert system
focused on both similarities and distinctions of individual specimens and
traits to generate general clustering by similarity and therefore by
theoretical evolutionary descent. (2) Numerical taxonomy
including morphological parsimony and phenetic
analysis generates dendrograms that accurately
place taxa most dissimilar from an outgoup highest in the tree. Because evaluation of traits
is methodologically clearly specified, the cladogram
is a powerful way to present basic information from which theories of
evolution are generated. It is limited by rejection of all information not
relevant to sister-group clustering. The cladogram
can be used, however, to create a “natural” key (one with embedded
evolutionary theory of descent with modification) with the help of additional
information about autapomorphies. (3) Molecular cladograms provide good estimates of genetic continuity,
subject to the caution that, given that any taxon
may split molecularly without speciation, any one OTU taxon
could be at all or many nodes, thus additional information must be used to
estimate the named caulistic element of the tree.
Phylogenetic paraphyly and polyphyly
(together heterophyly) imply a theoretical
progenitor-descendant diachronic relationship (i.e., involving nameable nodes
on the tree) corresponding to paraphyly-apophyly
arrangements on the cladogram. The progenitor taxon is diagnosable at the taxon
level that includes all exemplars derived from the previously unnamable nodes
involved in heterophyly. (4) Cross-tree heterophyly is basically a comparison of morphological
and molecular trees of the same taxa. A taxon low in the morphological tree (“primitive”) may be
considered the theoretical progenitor of all lineages between the position of
that taxon on the morphological tree and the
position of that same taxon on the molecular tree. (5) Application of Dollo’s Rule in the context of biosystematics allows
additional information from other fields (e.g., biogeography, chemistry,
cytology, ecology, evo-devo, fossils, morphometrics, paleontology, population genetics, and
other biosystematic indicators of descent with modification
of taxa) to help gauge the direction of diachronic,
caulistic evolution. A paper detailing the
method, with examples, is in preparation. Included in that paper is an
explanation of the value of Linnaean classification: “Once
an evolutionary evaluation of a group is performed, that evaluation is then
represented in a formal classification. Both evolutionary and phylogenetic systematics use Linnaean-based classification. This
classification is not particularly appropriate for representing either
sequential macroevolution or cladistic branching as
it does not directly represent processes in nature, but it provides a neutral
ground for representing a diversity (it is “well-hooked”) of observed
evolutionary relationships or processes, and does not purposefully reject
names that cause paraphyly. It uses both nesting
under higher ranks and contiguity (that is, closeness in lists) to simply
signal evolutionary relationships.” Additional
material: In systematics, the supreme question of our time is whether
to abandon science. Structuralism has its attractions, e.g., precision,
statistical certainty, and solely deductive methods of analysis, but it is
not science because it rejects testable inductive theory that conciliates
instead of relegates facts, and its models involve hidden causes, and
unnamable and unobservable entities. There are practices associated with phylogenetics that must be rejected for a new systematics to truly conciliate taxonomic methods that yield
disparate results, especially the following: (1) Shared ancestors are not named at the
same rank as their derivative exemplars in phylogenetics
because this would result in paraphyletic groups,
and nodes are simply place holders for the next higher inclusive rank. Thus,
if followed to an extreme, an alterative view is possible that species do not
disappear at all but their lines are bunched as skeins into the “shared
ancestor” of a higher rank. There is no evidence in a phylogenetic
classification, or a cladogram, or in the
evolutionary analysis that generated the cladogram,
against a species surviving a speciation event or even of immutability of
species. (2) Mapping of traits on cladograms is commonly referred to as instances
of evolutionary change, yet traits do not evolve, species do, an extreme
reductionism. Thus, trait changes mapped on cladograms
may be used to infer only microevolution and not macroevolution (one taxon evolving from another at the same rank or lower)
yielding a microevolutionary rather than macroevolutionary classification. Modeling descent with
modification of taxa is avoided. Mapping of
morphological traits or biogeographic distributions
on cladograms is an attempt to transform
evolutionary analysis from hypothesis and theory to lemma and theorem, i.e.,
from both deduction and induction to deduction only. (3) The rejection of naming tree nodes is
said to be due to the fact that then all branches of a tree would need to be
collapsed because a taxon cannot directly evolve from
another of the same rank, according the th
phylogenetic principle of holophyly. Classification
by holophyly (strict phylogenetic monophyly) is artificial and leads to degenerate (as a
return to absolutism) non-evolutionary classifications. Holophyly
has no ontological basis as a process in nature, that is, it is not refutable
and so is not a scientific hypothesis. It is ostensibly used for simplifying
taxonomy, but in doing so requires one to lump and split taxa
that in any way appear to represent macroevolution in classification. Holophyly clearly eliminates representation of
ancestor-descendant evolution in classification. Thus, nodes cannot be
assigned scientific names (other than a general and trivial attribution as
ancestors belonging to the general group of all taxa
distal on the tree). Not naming
ancestors gives them an ineffable, metempirical,
and recondite substance. It leads to faith-based taxonomy. Curiously, these three points (apparent
immutability of species, microevolution acceptable but not macroevolution,
and ineffable or mystical generation of species) are quite those of
“scientific creationism” or phylogenetic baraminology,
which uses phylogenetic software to group diachronic skeins of lineages, each
of a single immutable species. This does not instill confidence in phylogenetics. *** An Entirely Relevant Quotation: "I think," she said carefully,
"that perhaps too many people want things to be simple when they are not
and cannot be. Encouraging that desire is seductive and rewarding, but also
dangerous." He looked away a little, as if inspecting
something far in the distance over her left shoulder. "I think power has
always been like that," he said, his voice low. — Iain M. Banks, Against a Dark Background,
Orbit: Hachette Book Group, 1993, p. 378. *** More relevant quotations: According
to Dewey (1957), Darwinian logic and empiricism is opposed to and replaces
the ancient but long-regnant Greek philosophy of a transcendent first cause
with progressive organization to a final perfect form, where pre-Darwinian
science was “compelled to aim at realities lying behind and beyond the
processes of nature . . . .” Nowadays, Dewey averred, “in the twilight of
intellectual transition” there are new, intellectually atavistic, absolutist
technical philosophies that abstract “some aspect of the existing course of
events in order to reduplicate it as a petrified eternal principle to explain
the very changes of which it is the formalization.” -- Dewey, J. (1957) The
influence of Darwinism on philosophy. In: Likewise,
Jaynes (1990: 441), in explaining the attractions
of modern “scientific mythologies,” pointed out that “this totality [of
scientific method] is obtained not by actually explaining everything, but by
an encasement of its activity, a severe and absolute restriction of
attention, such that everything that is not explained is not in view.” -- Jaynes,
J. (1990) The origin of consciousness
in the breakdown of the bicameral mind. Houghton-Mifflin, Mariner Books,
Edition 2000, In discussing how “But if
thought corrupts language, language can also corrupt thought,” Orwell (1950)
warned against complacent acceptance of jargon and clichés: “This invasion of
one’s mind by ready-made phrases (lay
the foundations, achieve a radical transformation) can only be prevented
if one is constantly on guard against them, and every such phrase anaesthetises a portion of one’s brain.” G. Orwell. 1950. Politics and the English
Language. In: G. H. Muller. 1982. The McGraw-Hill Reader. *** Other evolutionary systematists have different priorities and illuminations,
and may challenge the above. The reader is invited to examine the
here-collected PDF reprints of other authors describing and advocating modern
evolutionary systematics. Additional discussions relevant to evolutionary systematics A Manifesto for the Restoration
of Evolutionary Systematics December 12, 2010 Манифест
(the above in Russian) Cladistic
equivalence of nested parentheses and cladogram
April 18, 2011 |
|
|