Table 2.

Properties of multilocus Nested Clade Analysis versus Approximate Bayesian Computation.

Property	NCA	ABC
Genetic Data Used	Haplotype Trees	Broad Array of Genetic Data
Data Analyzable	Geographic, Phenotypic, Ecological, Interspecific, etc.	Geographic, Ecological, Interspecific, etc.
Nature of Inference	Strong	Weak
Interpretative Criteria	Explicit, A priori, Universal	Explicit, Ad hoc, Case Specific
False Positives	False Positive Rate Estimable; False Positives Reduced via Cross-Validation	False Positive Rate Not Estimable; No Mechanism for Correcting for False Positives
A Priori Phylogeographic Model	No: Allows Discovery of Unexpected	Yes: Inference Limited to Finite Set of a priori Alternatives
Nature of Inferred Phylogeographic Model	Built-up From Simple Components, Each with Explicit Statistical Support	Full Models Specified a priori, Resulting in Confoundment When Model Has Several Components
Mechanics of Inference	Interpretive Key Followed By Phrasing as Null Hypotheses Tested With Likelihood Ratios	Simulation Requiring Multiple Parameters: Model and Parameter Values Confounded
Sampling Error	Incorporates Errors Due to Tree Inference, No. of Locations, No. of Individuals, and Evolutionary Stochasticity	Incorporates Sampling Error and Evolutionary Stochasiticy in Simulations; Ignores Error In Current Generation Statistics
Probability Distributions and Probablities	Simulates Full Sampling Distribution; Likelihood Ratios Based on Full Distributions	Local Probabilities, Obscuring Good-Fitting, Irrelevant, and Over-Determined Models
Sample Size	Handles Large Numbers of Locations and Individuals	Severely Restricted on the Number of Locations Analyzable
Convergence	Well Defined	Unknown
Dimensionality of Tests	Well Defined	Ignored
Final Test Product	Tail Probability of Null Hypothesis	A Non-Comeasurable Fit Metric With No Probabilistic Meaning