Phylogenetic terms Flashcards
Model
A representation of a process, rendered in mathematics.
In Bayesian systematics, a model typically describes the process of evolution leading to the data.
Model Assumptions
Factors about the model that are assumed to be true.
- An equal-weight parsimony analysis assumes changes between two character states are equally likely.
- In a Bayesian model, assumptions are written down into parameters, or mathematical facets of the model.
Observed Data
The data that have been collected by the researcher and will be used to infer the phylogeny.
In the case of morphological data, these will be the morphological characters collected, whether from extinct or extant organisms.
Discrete data
Data that can be broken into distinct and non-overlapping classes.
A common example of this data type is presence/absence data.
- Data with two classes are referred to as binary
- Data with more classes are referred to as multistate
Random variable
A variable whose value is the result of a random draw.
In most Bayesian models, the value of a given parameter is a random variable.
For example, the value of a particular branch length on a phylogeny is a random variable, which may be drawn from a distribution.
Continuous data
Data which cannot be broken into distinct and non-overlapping classes, and may take the value of any real number.
Examples include
- Geometric morphometric measurements
- Weights
- Lengths
Exchangeabilities
The rate at which one character state is expected to transition to another.
The exchangeabilities may be represented by one model parameter (in the case of the Mk model) or more (in the case of other, more complex phylogenetic models).
Prior distribution
A statistical distribution that describes the researcher’s prior beliefs or other outside information about the distribution of a model parameter
This allows the researcher to specify reasonable values for a parameter to take. A weak prior can be easily overcome by the data. A strong prior will require stronger signal in the data to be overcome.
Equilibrium character state frequencies
The frequencies of the character states in the dataset if the evolutionary process is allowed to run infinitely long.
In practice, the expected rate of a particular change between two character states is the product of the equilibrium character frequency and the exchangeability.
Q-matrix
A matrix defining the exchangeabilities and equilibrium character frequencies for a model at a given instant in evolutionary time.
The Q-Matrix will have a number of rows and columns equal to the number of character states of the data.
Posterior distribution
The posterior distribution is a distribution of plausible values for a parameter or set of parameters given the data and the prior distribution.
The posterior distribution is proportional to the model likelihood times the prior distribution.
Markov Chain Monte Carlo
An algorithm by which new values are proposed for model parameters, and evaluated.
In this procedure, initial values are scored under a model, then changed. If the changed parameter values improve on the old ones, they are used to seed the next step of estimation.
Brownian motion
A model of morphological change in which the value of a continuous character (X), is expected to change in proportion to an evolutionary rate (σ).
σ is expected to be normally distributed, with a variance that increases with time, such that more evolutionary change may be expected with time.
Model selection
A set of statistical approaches designed to determine whether an increase in the number of parameters of a model is justified given its increased ability to model variation in the data.
The addition of a parameter that does not increase the explanatory power of the model will not be supported by model selection. The exact degree of increase in explanatory power required to add a parameter will vary by model selection criteria.
Stem cetaceans
Extinct cetacean lineages that evolved prior to the origin of crown Cetacea.
This paraphyletic group includes all extinct lineages more closely related to living whales than to Hippopotamus, their closest living relative.
The phylogenetic grouping of living whales and Hippopotamus has been termed Whippomorpha, although there were likely extinct lineages more closely related to whales than hippos, such as Raoellidae, which are Eocene age artiodactyls from South Asia.
