Conservation Genetics of Whales:  In the late 1800’s over-hunting of whales caused significant population declines in many species.   Genetic tools can be used to
estimate pre-whaling population sizes and such estimates may help inform current management.  I collaborated with my wife, Kristen Ruegg, and Steven Palumbi at Stanford University in inferring historical population sizes of Antarctic minke whales [article], and am currently assisting on a project with humpback whales.  In the process I extended the LAMARC program’s Bayesian estimation capabilities [tech report abstract].
 

Southwest Fisheries Science Center, Santa Cruz, CA

and          Adjunct Prof, AMS Department, UCSC

Eric C. Anderson

I am broadly interested in statistical inference from genetic data.  In the statistical realm, my work has focused on Monte Carlo methods for efficient calculation of likelihoods from complex genetic models, Bayesian methods, computations on the coalescent process, application of hidden Markov models and graphical models to genetics, and inference in finite mixture models.  I have applied these methods in a variety of projects, a handful of which are listed below. 

Genetic Stock Identification (GSI) of Pacific Salmon:  Pacific salmon spawn in separate freshwater populations, but are usually harvested in oceanic fisheries.  To un
derstand the impacts of ocean fisheries on particular freshwater stocks of salmon, it is necessary to identify the stock or population of origin of the harvested fish.  I collaborate with a large consortium developing genetic markers for GSI of salmon from Russia to California.  For this project I have developed nearly unbiased methods for assessing the power of a set of markers for GSI (with collaborators Robin Waples and Steven Kalinowski) [article], and I have developed unbiased means of assessing the power for GSI of a subset of genetic loci selected specifically for the purposes of GSI [article].  I have also written software for an efficient, scriptable implementation of conditional maximum likelihood for GSI.
 
Parentage Based Tagging of Pacific Salmon:  Much Pacific salmon management has been based on recoveries of coded wire tags (CWTs) implanted in juvenile hatchery-raised salmon.  The tags bear codes which allow fish recovered with tags to be identified to hatchery of origin and to cohort.  This CWT program faces numerous challenges, but efforts
to upgrade the system using genetic data have been hampered by the fact that the standard GSI approach does not yield the cohort data necessary for stock forecasting.  My colleague at NMFS, Carlos Garza, proposed a system based on large scale parentage inference—parentage based tagging (PBT)—which would provide all the data that the CWTs provide (and much more!). We estimate that such a PBT system would be economically competitive with the CWT program at today’s genotyping costs, and in the coming decade, with continual declines in genotyping costs, it will become considerably cheaper.  It will also yield scientifically valuable pedigree information at salmon hatcheries.  For this project I developed importance sampling methods for computing the power for PBT of a set of single nucleotide polymorphism (SNP) markers [article] and have written software capable of performing parentage inference on a massive scale [software and report].
 
The Natal Origin of Polyploid Green Sturgeon:  Green sturgeon spawn in a handful of rivers on the West Coast, and they exhibit large, non-spawning congregations in river estuaries.  The origin of the fish in these congregations was not well known.  I collaborated with Josh Israel (then of the Genomics
Variation Lab, UC Davis) to develop GSI methods and assessment procedures that were appropriate to green sturgeon, given the tetrasomic nature of much of their genome, and the absence of information about the exact mode of inheritance of their tetrasomic markers.  This provided conclusive evidence regarding the natal origin of threatened green sturgeon in estuarine congregations [article].
 
Inference of Species Hybrids Using Genetic Data:  The admixture of species with different
allele frequencies creates predictable patterns of genetic variation in admixed individuals.  One way of characterizing that variation is though a mixture of different genealogical classes (F1, F2, backcrosses of various degrees, etc).  With Elizabeth Thompson, I developed such a model for inferring the hybrid category of individuals using genetic data [article].  I originally applied the method to rainbow trout-cutthroat trout hybrids, but since the release of my software, NewHybrids, it has been applied in numerous studies on many different species.  I have also extended the method to be useful for dominant markers like AFLPs [article].
 
Efficient Computation for Coalescent-derived Likelihoods in the Absence of Mutation:  A number of contemporary population-genetic inference problems can be addressed by using the coalescent process to formulate the likelihood.  In problems that deal with relatively
short time spans (a few to tens of generations) it is reasonable to assume that few, if any, mutations will have occurred in the relevant study period, and that mutation may be ignored.  Under such an assumption, I developed an importance sampling method for computing coalescent-based likelihoods that is orders of magnitude faster than the previously available MCMC-based methods.  I first applied these techniques to temporal estimates of the effective size of a population [article] and later, in collaboration with Monty Slatkin, to the problem of estimating the number of founders in recently colonized population [article].
 
Haplotype Blocks in the Human Genome:  Collaborator John Novembre (when he was
a grad student in the Slatkin Lab) and I developed a minimum-description length approach to characterizing and inferring haplotype blocks in the human genome [article].  In later work I collaborated with Monty Slatkin to explore the genetic and demographic mechanisms that could be behind the formation of haplotype blocks [article].
 
Inferring the Ancestral Origin of Lake Washington Sockeye Salmon:  I wrote my Master’s Thesis [eric_anderson_ms_thesis.pdf] on the statistical issues involved in using genetic data to infer the ancestral origin of Lake
Washington sockeye salmon.  At that time, insufficient genetic data were available to say much (and I was basically hopeless at producing genetic data—that is what drove me to doing population genetic theory!)  However, several years later, Ingrid Spies at the University of Washington produced a very good microsatellite data set to bring to bear to the problem.  I collaborated with her on the analysis of those data, finding evidence of a lineage of fish in Lake Washington that was distinct from possible introduced populations [article].