Sym139: Geometric morphometric in plant
systematics – 29 July
Limits and potentiality of geometric morphometrics
in plant systematics
Passalacqua, NG1
1University of Calabria, Italy
Geometric morphometrics methods started to be
developed in the late 1980s. During the 1990s, it showed
a remarkable qualitative and quantitative development,
thanks to the increased understanding of the theoretical
basis that underlie the methodologies, the development of
protocols of analysis, and the consequent increase in
published works that referred to these methods,
especially in Zoology and Anthropology. During this
time progress in botanical applications has been marginal
indicating a reluctance to accept new methods for
morphometric analysis. While we cannot exclude a
certain cultural inertia by botanists, it is undoubtedly true
that the study material of botanists, i.e. plants, is usually
less suitable for the application of geometric
morphometric techniques. However, in recent years,
experiments carried out by botanists gave a glimpse of
how these techniques can offer considerable support to
systematic studies in Botany. This presentation will first
analyze the main difficulties related to the use of
geometric morphometric techniques with plants, with
particular reference to the method of landmarks; limits
we need to be aware of for a correct application of these
methodologies and interpretation of results. The
presentation will then move on to the observation of
situations where geometric morphometric techniques
showed significant potential for exploration and/or
interpretation in plant Systematics.
------------------------------
The use of geometric morphometrics in evolutionary
studies
Rohlf, FJ1
1Stony Brook University, New York, USA
A discussion of some of the statistical properties of shape
data provided by geometric morphometric studies as well
as some of its limitations for use in taxonomic and
evolutionary studies will be presented. While in many
ways there is nothing special about the multivariate
analysis of such multivariate data, such data does have to
be treated properly. An important property of such data is
that it makes possible new ways of visualizing the results
of multivariate analyses.
-----------------
Integrating morphometric and phylogenetic analyses:
from phenetic systematics to phylogenetic
morphometrics
De Luna, E1, Chew, T1
1Biodiversidad y Sistemática, INECOL, AC, Mexico
The role of morphometric analyses in plant systematics
has stirred mixed reactions. Applications such as
discriminant analyses for taxon identification, or fitting
morphometric data to a phylogeny to estimate ancestral
shapes and changes along nodes on the tree do not seem
controversial. What remains unsettled is if morphometric
analyses should help and how in decisions about
characters, definition of character states, and for
taxonomic grouping. Morphometric studies can be a
strong basis for phylogenetic reconstructions. A sound
integration of morphometric data in phylogenetic
analyses is by the use of multivariate methods not for the
circumscription of taxa, as in a phenetic approach, but for
questions about character homology and monophyly. A
double path is now open for the integration of
morphometrics into phylogenetic analyses depending on
what is considered as 'cladistic character' and how to
analyze continuous patterns of variation in morphological
variables. The first question is what counts as character
data: whole shape, separate modules, partial warps,
relative warps? The second question is whether variation
must be partitioned and coded as discrete states or is
analyzed as continuous data. We argue for a need to
move from phenetic systematics to phylogenetic
morphometrics. We can now integrate very nicely
morphometric data into into phylogenetic reconstructions
of monophyletic groups by using continuous characters
and landmark data without coding or transformation.
Empirical tests show that the use of relative warp scores,
either means or ranges as continuous characters, are to be
preferred for phylogenetic analyses based on criteria of
concordance with non morphometric trees, levels of
homoplasy, and clade resolution/support.
----------------
Reconstructing ancestral floral shape using geometric
morphometrics: a test of the pollinator shift
hypothesis
Van Der Niet, T1,2, Zollikofer, CPE3, Ponce de Leon,
MS3, Johnson, SD1, Linder, P2
1School of Biological and Conservation Sciences,
University of KwaZulu-Natal, Pietermaritzburg, South
Africa; 2Institute of Systematic Botany, University of
Zurich, Switzerland; 3Anthropological Institute and
Museum, University of Zurich, Switzerland
Variation in floral shape has been used for centuries as a
diagnostic tool in plant systematics. The absence of
adequate methods to quantify floral shape, however, has
inhibited research into the underlying evolutionary
processes that generate this variation. The recent
developments of a rigorous statistical framework for
comparative analyses of shape, and the ability to produce
3D reconstructions of flowers with complex shapes, have
opened the door for addressing a wide variety of
evolutionary questions involving floral shape. We
explored the role of pollinators in the evolution of floral
shape variation in the orchid genus Satyrium, the species
of which have remarkably complex flowers, which are
characterized by specialized interactions with a wide
variety of insect and bird pollinators. We produced 3D
reconstructions of flowers using microCT scanning and
obtained 40 homologous landmarks and semi-landmarks.
A principal components analysis of a data set in which
size was removed by translation, rotation and scaling,
showed that species with the same pollinator have highly
similar shapes. To remove potentially confounding
phylogenetic effects, we first reconstructed ancestral
shapes onto an ultrametric tree by applying squared
parsimony to the procrust coordinates. We then
calculated the rate of evolution of shape by calculating
the Euclidean distance between reconstructed shapes at
the base and tip of individual branches, and by dividing
this by the branch length. Finally, we reconstructed
pollinator shifts onto the branches of the phylogenetic
tree. We found that the rate of morphological evolution is
significantly elevated in branches where there is an
unambiguous optimisation of pollinator shifts, compared
to branches where there is unambiguously no pollinator
shift, thus confirming the long-standing notion that
pollinators have played an important role in the
diversification of angiosperm floral shape. Although our
analysis is intuitively appealing, it assumes that, in the
absence of selection, evolution of quantitative traits
occurs according to a model of Brownian motion. Our
initial finding of strong similarity among species with the
same pollinator, however, suggested that floral shape is a
highly adaptive trait. This begs the question whether it is
valid to use a model of essentially neutral evolution for
the reconstruction of ancestral floral shape. Closer
examination of ancestral shapes showed that in cases of a
pollinator shift between sister taxa, the shape of the
immediate ancestor was intermediate as expected from
Brownian motion. These intermediate shapes sometimes
fell outside any of the well-defined, pollinator-specific
clusters in the principal components analysis. This means
either that these 'hybrid' shapes may be unfit, or that they
represent adaptations to unknown pollinators. Either of
these possibilities poses a challenge to accurate
evolutionary analyses. The development of alternative
models of evolution that allow for incorporation of
adaptive peaks, such as Ornstein-Uhlenbeck models for
multivariate traits, would provide a solution to this
problem. Nevertheless, we believe that our novel
approach of analyzing geometric morphometric data in a
phylogenetic and ecological context holds great promise
and represents a first step towards quantitative
approaches to addressing long-standing evolutionary
hypotheses.
No hay comentarios:
Publicar un comentario