MOLECULAR ECOLOGY OF PEROMYSCUS POLIONOTUS 
 
 
 
 
 
Except where reference is made to the work of others, the work described in this 
dissertation is my own or was done in collaboration with my advisory committee.  This 
dissertation does not include proprietary or classified information. 
 
 
 
 
Jeffrey L. Van Zant 
 
 
 
Certificate of Approval: 
 
 
 
 
F. Stephen Dobson     Michael C. Wooten, Chair 
Professor      Professor 
Biological Sciences     Biological Sciences 
 
 
 
 
 
James B. Grand     Craig Guyer 
Associate Professor     Professor 
Forestry and Wildlife Science    Biological Sciences 
 
 
 
     
 
Stephen L. McFarland 
Acting Dean 
    Graduate School 
MOLECULAR ECOLOGY OF PEROMYSCUS POLIONOTUS 
 
Jeffrey L. Van Zant 
 
A Dissertation  
Submitted to  
the Graduate Faculty of 
Auburn University  
in Partial Fulfillment of the  
Requirements for the 
Degree of  
Doctor of Philosophy 
 
Auburn, Alabama 
August 7, 2006 
 
 
MOLECULAR ECOLOGY OF PEROMYSCUS POLIONOTUS 
 
 
 
 
 
 
 
 
Jeffrey L. Van Zant 
 
 
 
Permission is granted to Auburn University to make copies of this dissertation at its 
discretion, upon request of individuals or institutions and at their expense.  The  
author reserves all publication rights. 
 
 
 
 
 
 
Signature of Author 
 
 
 
Date of Graduation 
 iii
 iv
VITA 
 
 Jeffrey L. Van Zant, son of Leon and Patricia (Bridgers) Van Zant, was born 
April 20, 1961, in Roanoke Rapids, North Carolina.  He attended Gulf Coast Community 
College and graduated magna cumme laude with an AA degree in biology in 1994.  He 
entered the University of Montana in 1994 and graduated cumme laude with a degree in 
wildlife biology in December 1996.  He became a doctoral student at Auburn University 
in January 2000.  He married Alison J. Lovell, daughter of Douglas and Sandra Lovell, 
on May 22, 2004.  They have a son, Brice Carden. 
 v
DISSERTATION ABSTARCT 
MOLECULAR ECOLOGY OF PEROMYSCUS POLIONOTUS 
 
 
Jeffrey L. Van Zant 
Doctor of Philosophy, August 7, 2006 
(B. S., University of Montana, 1996) 
(A. A., Gulf Coast Community College, 1994) 
 
337 Typed Pages 
Directed by Michael C. Wooten 
 
Lowe et al. (2004) describes ecological genetics as ?the investigation of the origin 
and maintenance of genetic variation within and between populations, which ultimately 
leads to adaptation and speciation.?  Thus, to gain a more complete understanding of a 
real world species an ecological genetics investigation must look across multiple levels 
and utilize different types of DNA markers as these processes occur on both a regional 
and microgeographic scale.  A regional scale typically involves between population 
processes while the microgeographic level often investigates the same processes but at a 
much finer scale.  Measures of genetic variation that are investigated at both scales 
include genetic diversity and genetic differentiation.  Here, I investigated these two 
 vi
measures of genetic variation on both a regional and microgeographic scale using two 
different DNA markers (mitochondrial and nuclear DNA). 
 First, mitochondrial DNA was used to investigate phylogenetic relationships 
within the species Peromyscus polionotus based on a previously proposed phylogenetic 
model.  The relationships predicted using mtDNA genes cytochrome-b and the control 
region were incongruent with the previously proposed model.  A revised model, then, 
indicates three distinct clades occupy the Gulf Coastal region of Florida and Alabama.  
Divergence among the clades may have occurred approximately 300 000 YBP.  The Gulf 
Coastal region is comprised of two inland clades and a beach clade.  Inland clades appear 
to integrate while beach subspecies are comprised of a single haplotype and do not appear 
to have recently integrated with adjacent inland populations.  In fact, extant beach forms 
represent lineages that are older than the present dune systems they inhabit. 
 Secondly, on a local or microgeographic scale, genetic diversity and 
differentiation was investigated using nuclear microsatellite markers.  It was discovered 
that genotypes are partitioned on a microgeographic scale (< 1000 m).  Genotype 
partitioning has several evolutionary implications and has been recognized as such for the 
majority of the past century.  Partitioning was apparently temporal, deteriorating through 
annual population declines and reorganizing during birth pulses.  Also, genotype 
partitioning reorganized following a hurricane that kept population densities low for four 
years.  Importantly, partitioning of genotypes was demonstrated to slow the rate of 
genetic decay within the population of Alabama beach mice. 
 vii
ACKNOWLEDGMENTS 
 
 
 Funding from the U.S. Fish and Wildlife Service, the Auburn University College 
of Sciences and Mathematics, the Office of the Vice President for Research, Auburn 
University and the Alabama Agricultural Experiment Station supported this research.  I 
express sincere gratitude to Dr. Michael C. Wooten for his guidance and friendship.  He 
was patient yet persuasive and from him I learned much about biology and life.  I also 
express gratitude to my Advisory Committee, Drs. James B. Grand, Craig Guyer, and F. 
Stephen Dobson for their ideas and constructive criticisms.  They all demonstrated a great 
deal of patience as I worked my way through this project.  I am greatly appreciative for 
the guidance I received from all of my mentors, as it was an instrumental part of my 
academic development.  Also, I offer special thanks to Dr. Marie Wooten and members 
of the Wooten lab.  Many undergraduates also gave their time to this project, 
unfortunately, there is not room here to mention them by name but their work is greatly 
appreciated.  I am grateful to have spent these last few years at Auburn University.  Many 
faculty, students and staff have contributed to this time being academically challenging, 
intellectually stimulating and generally enjoyable. 
 Finally, with deep conviction, I offer wife Alison and son Brice unending 
devotion.  They have given me patience, understanding and have affirmed in me the 
greatest joys of life.  Lastly, I dedicate this work to my parents, Leon Caleb and Patricia 
Ann Van Zant because my accomplishments are through their love and support.  
 Style manual or journal used Molecular Ecology 
  
 Computer software used Word  97, SAS 9.1, PAUP 4.0B.10, Powerpoint 97, 
Arlequin 2.0, Genescan 3.7,
 
 viii
 ix
TABLE OF CONTENTS 
 
 
 
LIST OF TABLES.....................................................................................................          x 
 
LIST OF FIGURES...................................................................................................       xiv 
 
GENERAL INTRODUCTION.................................................................................           1 
 
PLEISTOCENE EVENTS IMPACT MITOCHONDRIAL DNA  
LINEAGES AND HISTORICAL DEMOGRAPHY OF PEROMYSCUS 
POLIONOTUS...........................................................................................................          6 
 
 Introduction....................................................................................................          6 
 Methods..........................................................................................................        16 
 Results............................................................................................................        27 
 Discussion......................................................................................................        60 
 
SPATIOTEMPORAL DYNAMICS OF GENE DIVERSITY IN THE  
ALABAMA BEACH MOUSE (PEROMYSCUS POLIONOTUS 
AMMOBATES)...........................................................................................................      154 
  
Introduction....................................................................................................      154 
 Methods..........................................................................................................      164 
 Results............................................................................................................      169 
 Discussion......................................................................................................      181 
 
CUMULATIVE REFERENCES...............................................................................      275 
 
APPENDICES...........................................................................................................      294 
Appendix 1.....................................................................................................      295 
Appendix 2.....................................................................................................      304 
Appendix 3.....................................................................................................      310 
 x
LIST OF TABLES 
 
 
 
Chapter 1          
  
Table 1.     Key to specimens abbreviations relative to species and  
      collection site................................................................................     141 
 
 Table 2.     Measures of diversity for the two major sets of sequence  
      data that were used for phylogenetic analyses..............................     143 
 
 Table 3.     Permutational Chi-squared tests of geographical association  
      of clades and the inferred biological explanation.........................     144 
 
Table 4.     Measures of diversity for the clades estimated by maximum   
likelihood......................................................................................     146 
 
Table 5.     Measures of diversity for model I 5
th
 level clades  
      estimated using nested clade analysis...........................................     147 
 
Table 6.     Measures of diversity for model II 5
th
 level clades  
      estimated using nested clade analysis...........................................     148 
 
 Table 7.    Measures of diversity for model III 5
th
 level clades  
     estimated using nested clade analysis............................................     149              
 
Table 8.     Fu and Li?s D and F tests for Peromyscus polionotus with  
Peromyscus maniculatus as the outgroup haplotype along  
with McDonald-Kreitman test used only with cytochrome-b  
P. maniculatus as the outgroup haplotype....................................     150 
  
Table 9.     Fu and Li?s D and F tests for Peromyscus polionotus clades  
      or groups with other P. polionotus as outgroup haplotypes..........     151 
 
   Table 10.    Fu and Li?s D* and F* and Tajima?s D tests for estimated  
       clades and groups.........................................................................     152
 xi
 
 
 
Table 11.   Coalescent based maximum likelihood estimates of N
e
,  
      theta (?), and exponential growth rate (95% CI) under  
      models of constant size and growth as implemented by  
      Kuhner et al. (1998)......................................................................     153 
    
 Chapter 2 
 
 Table 1.    Characterization of the 11 microsatellite loci used for these 
      analyses including primer sequence, accession number,  
    repeat motif, annealing temperature (T
a
), and the  
    number of observed alleles.............................................................     246 
 
Table 2.    Total captures, new captures per trapping period and 
      the number of individuals from which tissues was  
     obtained for the combined trapping periods..................................     247 
 
 Table 3.    Total captures, new captures per trapping period and 
     the number of individuals from which tissue was  
     obtained on Gazebo grid................................................................     248 
 
 Table 4.    Total captures, new captures per trapping period and 
       the number of individuals from which tissue was 
       obtained on Vet Village grid..........................................................     249 
 
 Table 5.    The total number of alleles observed for each trapping  
     period on each grid and combined.................................................     250 
 
 Table 6.    Summary statistics for the trapping periods of February 
       1995 on Gazebo grid and March 1995 on  
     Vet Village grid..............................................................................     251 
 
 Table 7.    Summary statistics for the trapping periods of April 1995 
       on Gazebo grid and April 1995 on Vet Village grid......................     252 
 
 Table 8.    Summary statistics for the trapping periods of June 
       1995 on Gazebo grid and June 1995 on Vet  
     Village grid....................................................................................     253 
 
 Table 9.    Summary statistics for the trapping periods of February 
       1996 on Gazebo grid and March 1996 on Vet 
       Village grid....................................................................................     254 
 
 xii
Table 10.  Summary statistics for the trapping periods of April 
      1996 on Gazebo grid and April 1996 on Vet 
      Village grid...................................................................................     255  
 
Table 11.  Summary statistics for the trapping periods of June 
     1996 on Gazebo grid and June 1996 on Vet 
     Village grid....................................................................................     256 
  
Table 12.  Summary statistics for the trapping periods of February 
       1997 on Gazebo grid and March 1997 on Vet 
       Village grid....................................................................................     257 
 
 Table 13.  Summary statistics for the trapping periods of April 
       1998 on Gazebo grid and February on Vet 
       Village grid....................................................................................     258 
 
 Table 14.  Summary statistics for the trapping periods of March 
       2000 on Gazebo grid and March 2000 on Vet 
       Village grid....................................................................................     259 
 
 Table 15.  Summary statistics for the trapping periods of November 
       2000 on Gazebo grid and December 2000 on 
       Vet Village grid..............................................................................     260 
 
 Table 16.  Summary statistics for the trapping periods of March  
     2001 on Gazebo grid and March 2001 on Vet 
     Village grid....................................................................................     261 
 
 Table 17.  Summary statistics for the trapping periods of November 
       2001 on Gazebo grid and October 2001 on Vet 
       Village grid....................................................................................     262 
 
 Table 18.  Summary statistics for the trapping periods of March 
       2002 on Gazebo grid and February 2002 on Vet 
       Village grid....................................................................................     263 
 
 Table 19.  Mantel test for Gazebo and Vet Village grids along with  
     the number of samples for each trapping period............................     264 
 
 Table 20.  F
ST
 values based on various models of subdivision  
     within Gazebo and Vet Village grids.............................................     265 
 
  
 
 
 xiii
Table 21.  Summary statistics produced by the combined trapping  
     periods of February/March 1995 along with combined  
     trapping periods of April/April 1995 on Gazebo and  
     Vet Village grids............................................................................     266 
 
 Table 22.  Summary statistics produced by the combined trapping  
     periods of June/June 1995 along with combined  
     trapping periods of February/March 1996 on Gazebo and  
            Vet Village grids............................................................................     267 
 
 Table 23.  Summary statistics produced by the combined trapping  
     periods of April/April 1996 along with combined  
     trapping periods of June/June 1996 on Gazebo and  
            Vet Village grids............................................................................     268 
 
 Table 24.  Summary statistics produced by the combined trapping  
     periods of February/March 1997 along with combined  
     trapping periods of April/February 1998 on Gazebo and  
            Vet Village grids............................................................................     269 
 
 Table 25.  Summary statistics produced by the combined trapping  
     periods of March/March 2000 along with combined  
     trapping periods of November/December 2000 on Gazebo  
     and Vet Village grids.....................................................................     270 
 
 Table 26.  Summary statistics produced by the combined trapping  
     periods of March/March 2001 along with combined  
     trapping periods of November/October 2001 on Gazebo  
            and Vet Village grids.....................................................................     271 
 
 Table 27.  Summary statistics produced by the combined trapping  
     periods of March 2002 on Gazebo grid and February 
     2002 on Vet Village grid................................................................     272 
 
 Table 28.  Mantel tests for the combined trapping periods along with  
     the number of samples for each trapping period............................     273 
 
 Table 29.  Between grid measures of differentiation including F
ST
,  
     R
ST
 and the number of migrants per generation.............................     274 
 
 Table 30.  Population assignment using the jackknife method from 
       the program WHICHRUN (Banks & Eichert 2000).........................     275 
 
 xiv
LIST OF FIGURES 
 
 
             
  Chapter 1.  
 
Figure 1.     Bowen?s (1968) proposed evolutionary sequence of the  
Gulf  Coastal and some Atlantic forms of Peromyscus   
polionotus......................................................................................      75 
 
 Figure 2.    Locations where P. polionotus samples were collected................      77 
 
Figure 3.     Insert region found within P. polionotus and its  
       alignment with P. maniculatus.....................................................      79 
 
 Figure 4.     Maximum likelihood topology using the cytochrome-b  
       region and model HKY +? with outgroup P. leucopus................      81 
 
 Figure 5.     Maximum likelihood topology using the cytochrome-b  
       region and model TVM +? with outgroup P. leucopus................      83 
 
 Figure 6.     Maximum likelihood topology using the cytochrome-b  
       region and model HKY +? with outgroup P. keeni......................      85 
 
 Figure 7.     Maximum likelihood topology using the cytochrome-b  
       region and model GTR +? with outgroup P. keeni.......................      87 
 
 Figure 8.     Maximum likelihood topology using the D-loop and  
       model HKY +I +? with outgroup P. leucopus.............................      89 
 
 Figure 9.     Maximum likelihood topology using the D-loop and  
       model HKY +I with outgroup P. keeni.........................................      91 
 
 Figure 10.   Maximum likelihood topology using the cytochrome-b  
       region and D-loop from model HKY +I +? with outgroup  
       P. leucopus....................................................................................      93 
 
 Figure 11.   Maximum likelihood topology using the cytochrome-b  
       region and D-loop from model HKY +I +? with outgroup  
       P. keeni..........................................................................................      95
 xv
Figure 12.   Weighted maximum parsimony topology using the  
        cytochrome-b region with outgroup P. leucopus.........................      97 
 
 Figure 13.   Maximum parsimony topology using the cytochrome-b  
       region with outgroup P. leucopus.................................................      99 
 
Figure 14.   Weighted maximum parsimony topology using the  
       cytochrome-b region with outgroup P. keeni..............................      101 
 
 Figure 15.   Maximum parsimony topology using the cytochrome-b  
       region with outgroup P. keeni.....................................................      103 
  
Figure 16.   Weighted maximum parsimony topology using the D-loop  
       with outgroup P. leucopus...........................................................     105 
 
 Figure 17.   Maximum parsimony topology using the D-loop with  
         outgroup P. leucopus...................................................................     107 
 
 Figure 18.   Weighted maximum parsimony topology using the D-loop  
         with outgroup P. keeni.................................................................     109 
 
 Figure 19.   Maximum parsimony topology using the D-loop with  
         outgroup P. keeni.........................................................................     111 
 
 Figure 20.   Weighted maximum parsimony topology using the  
        cytochrome-b and D-loop with outgroup P. leucopus................     113 
 
 Figure 21.   Maximum parsimony topology using the cytochrome-b 
         region and D-loop with outgroup P. leucopus.............................     115 
 
 Figure 22.   Weighted maximum parsimony topology using the  
        cytochrome-b region and D-loop with outgroup P. keeni...........     117 
 
 Figure 23.   Maximum parsimony topology using the cytochrome-b  
       region and D-loop with outgroup P. keeni...................................     119 
 
 Figure 24.   Bayesian analysis using the cytochrome-b region with  
       model GTR +? and outgroup P. leucopus...................................     121 
 
 Figure 25.   Bayesian analysis using the cytochrome-b region with  
       model HKY +? and outgroup P. keeni........................................     123 
 
 Figure 26.   Bayesian analysis using the cytochrome-b region with  
       model GTR +I and outgroup P. keeni..........................................     125 
 
 xvi
 Figure 27.   Bayesian analysis using the D-loop with model HKY +I +?  
         and outgroup P. leucopus.............................................................     127 
 
 Figure 28.   Bayesian analysis using the D-loop with model HKY +I +?  
         and outgroup P. keeni..................................................................     129 
 
 Figure 29.   Bayesian analysis using the cytochrome-b region and the  
       D-loop with split models HKY +? (cty-b) and HKY +I +?  
       (cyt-b and D-loop) with outgroup P. leucopus............................     131 
 
 Figure 30.   Bayesian analysis using the cytochrome-b region and the  
       D-loop with split models HKY +? (cyt-b) and HKY +I +?  
       (cyt-b and D-loop) with outgroup P. keeni..................................     133 
 
Figure 31.   Statistical-parsimony network (NCA) of P. polionotus  
       haplotypes....................................................................................     135 
  
Figure 32.   Constant-rate model of time estimates for estimated  
       divergence periods within P. polionotus......................................     137 
 
 Figure 33.   Range map of P. p. allophrys and location of debated  
       haplotype......................................................................................     139 
 
Chapter 2 
 
 Figure 1.     Gazebo and Vet Village grids established during the  
       fall of 2000...................................................................................     196 
 
 Figure 2.     Reduction of both grids after the 1997 trapping periods 
          increased the distance between them to 920 m...........................     198 
  
Figure 3.     Observed and estimated heterozygosity for each trapping  
        period on Gazebo grid.................................................................     200 
  
 Figure 4.    Observed and estimated heterozygosity for each trapping 
         period on Vet Village grid...........................................................     202 
 
 Figure 5.    Gazebo grid correlograms from the trapping period of 
         February 1995..............................................................................     204 
 
 Figure 6.    Gazebo grid correlograms from the trapping period of 
        April 1995.....................................................................................     206 
 
 Figure 7.    Gazebo grid correlograms from the trapping period of 
         June 1995.....................................................................................     208 
 xvii
 
 Figure 8.    Gazebo grid correlograms from the trapping period of 
         February 1996..............................................................................     210 
 
 Figure 9.    Gazebo grid correlograms from the trapping period of 
        April 1996.....................................................................................     212 
 
 Figure 10.   Gazebo grid correlograms from the trapping period of 
          June 1996....................................................................................     214 
 
 Figure 11.   Gazebo grid correlograms from the trapping period of 
         March 2002..................................................................................     216 
 
 Figure 12.   Vet Village grid correlograms from the trapping period  
       of March 1995..............................................................................     218 
 
 Figure 13.   Vet Village grid correlograms from the trapping period 
         of April 1995................................................................................     220 
 
 Figure 14.   Vet Village grid correlograms from the trapping period 
         of June 1995.................................................................................     222 
 
 Figure 15.   Vet Village grid correlograms from the trapping period 
         of March 1996..............................................................................     224 
 
 Figure 16.   Vet Village grid correlograms from the trapping period 
         of April 1996................................................................................     226 
 
 Figure 17.   Vet Village grid correlograms from the trapping period 
         of June 1996.................................................................................     228 
 
 Figure 18.   Vet Village grid correlograms from the trapping period 
         of February 2002..........................................................................     230 
 
 Figure 19.   Correlograms from the combined trapping periods of  
       February 1995 on Gazebo grid and March 1995 on  
       Vet Village grid............................................................................     232 
 
 Figure 20.   Correlograms from the combined trapping periods of  
       April 1995 on Gazebo grid and April 1995 on  
               Vet Village grid............................................................................     234 
 
 Figure 21.   Correlograms from the combined trapping periods of 
         June 1995 on Gazebo grid and June 1995 on 
         Vet Village grid............................................................................     236 
 xviii
 
 Figure 22.   Correlograms from the combined trapping periods of 
         February 1996 on Gazebo grid and March 1996 
         on Vet Village grid.......................................................................     238 
 
 Figure 23.   Correlograms from the combined trapping periods of 
         April 1996 on Gazebo grid and April 1996 on 
         Vet Village grid............................................................................     240 
  
Figure 24.   Correlograms from the combined trapping periods of 
         June 1996 on Gazebo grid and June 1996 on 
         Vet Village grid............................................................................     242 
 
 Figure 25.   Correlograms from the combined trapping periods of 
         March 2002 on Gazebo grid and February 2002 
         on Vet Village grid.......................................................................     244 
 1 
GENERAL INTRODUCTION 
 
 
 An earnest investigation concerning the movement of genes both within and 
between populations probably began with Wright (1931).  Wright and others (Wright 
1943; Kimura 1955; Kimura & Weiss 1964; Levins 1969; Slatkin 1977) developed 
increasingly more complex models that were designed to infer gene flow between 
subpopulations.  Models, however, represent indirect estimates of gene flow.  Direct 
methods of estimating gene flow were, historically, based on observational studies, 
however, more recently, genetic methods of measuring gene flow are becoming more 
popular as molecular markers and methods become more efficient to use.  Gene flow is 
influenced by at least two factors, the physical ability of a species to change location and 
the environment the organism exists in.  For example, species ill adapted for water are 
unlikely to cross major bodies of water except in rare circumstances.  The amount of gene 
flow between groups has predictable outcomes in regard to genetic differentiation of the 
groups (Hartl & Clark 1998).  Gene flow, however, occurs both between and within 
populations.  Understanding the genetic composition of a population requires an accurate 
assessment of genetic diversity and gene flow on both a regional and local scale. 
My primary research interest and, thus, the focus of my dissertation involved 
understanding genetic diversity within and between populations.  The project was marked 
by a unique opportunity to investigate the maintenance of genetic variation within a 
population that functions without any obvious means for partitioning genotypes.  The 
 2
genetic architecture of a population, as well as the dynamics acting to direct and maintain 
this architecture, are part of a population?s response to counteract the persistent decay of 
gene diversity.  Thus, understanding these processes and their outcome(s) has important 
evolutionary and conservation implications.  As the model system for my research, I used 
the old-field mouse (Peromyscus polionotus spp).  This species exhibits a collection of 
features that, I believe, made it an excellent model for the questions examined here.  The 
species is monogamous, forms strong pair bonds, exhibits limited dispersal and occupies 
a temporally dynamic habitat.  In addition, over 10 years of tissue samples and detailed 
ecological data from a single natural population were available.   
First, I examined biogeographic variation at the regional level. Gene flow from 
the mainland of Alabama into the beach populations had long been proposed (Bowen 
1968).   This hypothesized gene flow was important to understand in the evolutionary 
context of my primary study population of Alabama beach mice (Peromyscus polionotus 
ammobates).  Thus, I conducted a biogeographic study of populations spanning the 
northern Gulf Coast.   Despite biological studies dating back to the 1920?s, the 
evolutionary relationships within and among beach and inland subspecies remains an 
open question.  Various phylogenetic hypotheses have been proposed for the Gulf Coast 
populations but no definitive conclusions have been reached.  For example, within beach 
mice, there are varying degrees of color morphs that range from mostly white forms in 
the Santa Rosa beach mouse (Peromyscus polionotus leucocephalus) to the much darker 
morph of the Alabama beach mouse (Peromyscus polionotus ammobates).  Bowen (1968) 
suggested that this dichotomy of pelage color was the result of differing levels of 
historical gene flow.  The Santa Rosa beach mouse has been effectively isolated on Santa 
 3
Rosa Island for at least 5000 years.  In contrast, seemingly, a land bridge existed between 
the habitat of the Alabama beach mouse and the nearby inland populations until modern 
times.  Bowen (1968) hypothesized that this land connection allowed regular gene flow 
into the beach population.  However, results here indicate this that this scenario is 
unlikely.  It appears that two gene regions (cytochrome-b, control region) in the mtDNA 
genome are monomorphic in the Alabama beach mouse population.  These results 
conflict with Bowen?s hypothesis of beach mouse phylogeny because the nearby inland 
population is not monomorphic for these gene regions and sampled variations are 
relatively diverged from the beach forms.  Interestingly, the data here suggest that the 
Alabama beach mouse and the Santa Rosa beach mouse may share a more recent ancestor 
with each other than they do all other forms, which is also very different than predicted 
by Bowen (1968).   In general, beach forms appear to be more substantially diverged 
from inland forms than suggested by Bowen; thus, my data bring into question all 
previous phylogenetic conclusions. 
Peromyscus polionotus is generally thought to be a young species having evolved 
after being isolated in the southeastern United States during a mid to late Pleistocene 
glacial cycle.  New data presented here along with recent fossil evidence indicates that P. 
polionotus is far older than previously thought.  Using these mitochondrial DNA data, P. 
polionotus may have been isolated in the southeastern United States since the late 
Pliocene or early Pleistocene.  Thus determining which form of Peromyscus maniculatus 
that polionotus split from is difficult.  In fact, my data suggest that the forms of 
maniculatus examined here split following a polionotus split from maniculatus.  Thus, 
 4
the subspecies of maniculatus for which I have data are represent younger lineages than 
polionotus. 
Bowen?s model of the evolution of Gulf Coastal forms of P. polionotus ssp 
indicated a recent colonization of the present beach by inland forms.  Thus, present beach 
forms should be relatively undifferentiated from extant inland forms.  However, my data 
suggest a split among three extant clades of polionotus of approximately 300 000 YBP.  
This indicates that present forms of beach mice are significantly older than the 
approximately 10 000 years suggested by Bowen (1968).  My data strongly suggest that 
populations of polionotus populating shorelines and barrier islands of the Gulf Coast are 
older than the dunes they now inhabit.  Therefore, as an alternative to Bowen?s recurrent 
invasion hypothesis, I propose the Shoreline-Tracking Hypothesis.  My data suggest that 
beach forms, already diverged from inland forms, were able to track the shoreline as the 
water of the Gulf of Mexico rose due to melting glaciers. 
The second, and central, focus of my dissertation involved examination of 
variation at the population level.  Once I had developed a historical framework for my 
study population, I examined the fate of genetic variants at the microgeographic level.  
The idea for this portion of my research arose from two observations reported by my 
advisor.  Dr. Wooten noted that, following Hurricane Opal and the resulting collapse in 
population density, individual heterozygosity increased in a population of Alabama beach 
mice.  This result runs completely counter to what one would expect under bottleneck 
conditions, unless some form of significant genotype partitioning existed prior to the 
storm event.  The immediate question that came to me was ?How did this partitioning 
arise??   The simplistic breeding, dispersal and social systems reported for beach mice 
 5
have not traditionally been considered adequate to drive formation of extensive 
population structure.  More interesting, how could such structure exist in a population 
where individuals could easily move across the entire habitat space?   Based on 
preliminary genotype data, Dr. Wooten proposed that beach mice were capable of 
forming ?genetic neighborhoods? and he hypothesized that genetic relatedness within 
these neighborhoods served as reservoirs of genetic variance.  Starting with these ideas, I 
formulated a series of questions that served to guide the following portion of my 
dissertation work.  First, I attempted to document and quantify both spatial and temporal 
genetic structure.  Once patterns of variation were identified, I examined the potential for 
this structure to influence the rate of genetic decay.  Using this approach, I demonstrated 
that genetic partitioning does occur in beach mice populations and that this phenomenon 
serves to slow the rate of decay of genetic variance in these populations.  I believe that 
this is a broad phenomenon that is an important factor underlying the genetic architecture 
of many mammalian populations. 
  
 6 6
Chapter One 
 
Pleistocene Events Impact Mitochondrial DNA Lineages and  
Historical Demography of Peromyscus polionotus 
 
Introduction 
The Pleistocene Epoch, covering approximately the last two million years, had a 
profound effect on the biota of Earth (Hewitt 1996, 2000).  Cyclical glacial-interglacial 
periods caused range alterations (Hewitt 2000, Blondel & Aronson 1999) and exchanges 
of biota.  These climatic events have been explored as a possible cause of numerous 
speciation events (Haffer 1969, Veith et al. 2003, Avise et al. 1998) and extinctions 
(Graham & Lundelius 1984).  One important aspect of these fluctuations was that habitats 
were lost, gained, or repositioned in regions far beyond the physical ice sheet.  Repeated 
rounds of range contraction, dispersal, and vicariance events were likely common.  The 
last major geological event (a withdrawal of glaciers and a corresponding significant 
increase of sea level) produced a Varity of effects on numerous species (Hoffman & 
Blouin 2004, Zamudio & Savage 2003, Arbogast 1999) many of which persist today.  For 
such species, these historical processes often played central roles in defining present 
geographic distribution of genetic lineages (Pielou 1991, Hewitt 1993, Avise 2000). 
  
 7
Complex climatic and geographic processes essentially confirm a complicated 
evolutionary history for species within affected regions.  Disentangling geographic from 
contemporary effects on genetic lineages has long been recognized as essential for 
gaining a more complete understanding of an organism?s evolutionary history.  Classical 
approaches to the study of genetics data in relation to spatial history began with Wahlund 
(1928) and Wright (1931, 1943).  Wright developed F-statistics to quantify the 
subdivision of populations and estimate gene flow under the assumption of drift-
migration equilibrium in an island model.  However, F-statistics were not designed to 
make use of the temporal information inherently found within gene sequences.  
Elucidating past intraspecific cycles of contraction, isolation, and expansion is difficult.  
More recently, expanded use of molecular techniques and the emergence of 
mitochondrial DNA data placed an emphasis on intraspecific variation.  These events 
have fostered the growth of a new discipline termed phylogeography (Avise et al. 1979, 
1987; Avise 2000).  Phylogeography is concerned with ?historical aspects of the 
contemporary spatial distributions of gene lineages.? (Avise 1996).  Analytical methods 
that have aided the growth of phylogeography include, nested clade analysis (Templeton 
et al. 1987, 1992; Templeton & Sing 1993), estimates of historical demographic 
parameters (Emerson et al. 2001), and tests of neutrality (Fu 1997).  Nested clade 
analysis (NCA) was used to test for geographical association of haplotypes by placing 
haplotypes in clades at different hierarchical levels until a final level was reached that 
contains all haplotypes.  NCA then tests for geographical association among haplotype 
groups.  If significant geographical associations are found, a decision key is used to infer 
a biological scenario through which the observed association may have occurred.  
  
 8
Historical demographic parameters, such as effective population size, can be estimated 
using coalescent theory (Kuhner et al. 1995).  Neutrality tests, initially derived for testing 
the neutrality of genes, can also indicate other circumstances such as population growth.  
Together, these analyses can be used to estimate how historical mechanisms have 
influenced extant genetic structure within a population. 
As glaciers advanced and retreated, Pleistocene climatic oscillations repeatedly 
altered vegetation types in the southeastern United States.  Sea level changes as great as 
100 m (Gates 1993) marked the occurrence of glacial cycles that moved shorelines by 
more than a hundred kilometers in some regions of the Gulf of Mexico (Wanless 1989).  
Invariably, organisms inhabiting this region were also affected.  Intraspecific genetic 
structuring of terrestrial vertebrates has been investigated within the region (Avise 1996), 
with the predominate phylogeographic model being turtles (Walker & Avise 1998).  
Phylogeographic studies of mammals from the region have also been completed 
(Ellsworth et al. 1994; Hayes & Harrison 1992; Avise et al. 1983); however, few, if any, 
have combined phylogenetic, NCA, and historical demographic methods in an attempt to 
measure the effects of past geologic events on present genetic distribution.  For a species 
intimately associated with such an unstable environment, application of multiple methods 
in the study of intraspecific variation is important for understanding and determining the 
significant pattern(s) that may be detectable using gene sequences. 
Here, I examine a mammalian species (Peromyscus polionotus) that is restricted 
to the Southeastern United States and that has an evolutionary history closely tied to the 
Florida and Alabama coastline of both the Gulf of Mexico and Atlantic Ocean.  
Peromyscus polionotus is a small fossorial mouse of the southeastern United States 
  
 9
probably derived from a grassland form of Peromyscus maniculatus (Osgood 1909; 
Sumner 1926; Blair 1950; Bowen 1968).  Recent fossil evidence dates the appearance of 
Peromyscus polionotus to the late Pliocene or early Pleistocene Epoch (Ruez 2001) 
representing approximately two million years before present (YBP).  This time frame is 
significantly different from the earlier widely held hypothesis of a late Pleistocene/early 
Holocene split (Hibbard 1968; Webb 1974).  Webb (1974) reported fossils of P. 
polionotus occurring at sites that dated between 20 000-100 000 years and fossils of 
Peromyscus dating to the mid-Pleistocene.  A late Pleistocene split seems to be the 
assumption accepted by most authors when investigating the biogeography of this group 
(Hibbard 1968; Bowen 1968).  However, using restriction enzymes and mtDNA, Avise et 
al. (1983) estimated that P. polionotus and a form of P. maniculatus from southern 
California split 1.5 MaBP but considered this an interim estimate. 
According to Osgood (1909) there are two major morphological variants among 
P. maniculatus.  These consist of a long-tailed, large-eared, large-footed forest type and a 
short-tailed, small-eared, small-footed grassland type.  While today a forest type (P. m. 
nubiterrae) is geographically closer, Osgood (1909) noted that P. polionotus was more 
similar to the two grassland types, P. m. pallescens of Texas and P. m. bairdii of 
Michigan, and most resembled P. m. pallescens.  Since Osgood?s time, two competing 
hypotheses have arisen concerning the origin of P. polionotus.  Blair (1950) postulated 
that P. maniculatus moved east along the Gulf Coast Corridor during the Pleistocene 
glacial periods.  The Gulf of Mexico was much lower than present during glacial periods, 
and the Gulf Coastal Plain connected Florida and Texas, thus allowing the migration of 
many mammals from North and South America into Florida (Webb & Wilkins 1984).   
  
 10
Bowen (1968), however, proposed a different hypothesis for the origin of P. 
polionotus.  He suggested a northern origin for the P. maniculatus stock that gave rise to 
polionotus.  Presently, P. m. bairdii is the grassland form occupying the area most 
consistent with Bowen?s model (Southern Michigan).  Bowen speculated that during a 
pre-Pleistocene dry period, a north-south grassy corridor could have formed allowing 
passage of the ancestral polionotus stock.  Bowen cited the absence of P. polionotus from 
suitable habitat west of Mobile Bay and the Alabama River as evidence against Blair?s 
hypothesis.  Subsequently, however, P. polionotus has been determined to inhabit areas 
west of the Alabama River (M. C. Wooten, personal communication).  This seems to be 
the extent to which this speciation event has been characterized in the literature.  Based 
on the evidence and technology of the time, it was difficult to advance either theory.  
However, since 1968, numerous advances in biogeographic and genetic techniques have 
been made.  Thus, at this time, it may be possible to determine with increased probability 
which P. maniculatus was involved in this speciation event, and if one of the extant 
grassland forms (bairdii, pallescens) is the most recent ancestor of P. polionotus.   
Both the sequence of events leading to the speciation of P. polionotus and the 
radiation of subspecies within the polionotus group remain poorly understood.  Clarifying 
the systematic relationships within P. polionotus requires identification of the ancestor of 
the group, but the geographic distribution and complex biogeographical events make this 
task difficult.  Driven by glacial cycles, both shifting vegetation zones and changing 
shorelines have almost certainly influenced the present biogeographic structure of P. 
polionotus.  Indeed, isolated in the southeastern coastal plain, the polionotus group 
clearly evolved considerable geographic variation throughout its range (Blair 1950; 
  
 11
Bowen 1968; Selander 1970a).  In fact, phenotypic variation has given P. polionotus 
something of a storied history.  Beginning early in the last century and continuing 
through the majority of it, studies of P. polionotus were critical in the growth of current 
concepts such as geographic variation and the adaptiveness of morphological characters.  
Osgood (1909) initiated the historical research of P. polionotus by listing it along with 
four subspecies, three of which previously had species status.  Subsequent splitting of the 
polionotus group followed until 1968 when Bowen, in an attempt to clarify the taxonomy 
of the Gulf Coast forms named five new subspecies.   
Sumner (1926), concentrating primarily on color morphs that ranged from brown 
in the interior populations to almost white in the insular populations used several of the 
subspecies in crossing experiments to analyze the genetic basis of morphological 
characters.  His results were widely cited by many evolutionists (e.g. Mayr 1942, 1954; 
Huxley 1943; Haldane 1948; Ford 1954, 1960).  Sumner (1929) noted an apparent cline 
from Santa Rosa Island extending inland approximately 64 km, where pelage changed 
from pale to dark.  He characterized this cline as gene flow inland from the beaches.  
Huxley (1943) and Haldane (1948) used this apparent gradient to produce some of the 
first analyses of clinal variation.  However, according to Bowen (1968), Sumner?s results 
were misleading largely due to the inadequately studied taxonomy of forms along the 
Gulf Coast.  Bowen argued conversely that gene flow was predominately from interior 
populations into the beach populations.   
Based on the first broad geographic studies of electrophoretic protein 
polymorphism, Selander (1970b) and Selander et al. (1971) also concluded that gene 
flow was most likely from interior to insular populations.  In 1971, Selander et al. 
  
 12
published a comprehensive investigation of protein polymorphism in P. polionotus.  
Their analysis showed significant geographical partitioning of polymorphic loci.  They 
demonstrated that the proportion of polymorphic loci within insular populations along the 
Florida Panhandle (8.8%) was decidedly lower than that of the adjacent mainland 
populations (22.0%).  Moreover, the interior populations of the Florida Peninsula were 
found to have the highest proportion of polymorphic loci (29.0%) while the insular 
populations of the Atlantic coast were lower (22.0%).  Inland populations across Georgia 
and South Carolina were polymorphic at 20% of loci sampled.  Selander (1970b) 
contented that a low proportion of polymorphic loci among insular populations was due 
to their isolation from one another and from interior populations and the amount of gene 
flow that did occur was not sufficient to overcome this.  Their analysis also identified a 
north-south cline of increased polymorphic loci.  One criterion used for hypothesizing a 
center of origin is the area of greatest variation.  Thus, from the Selander et al. (1971) 
data one might postulate that the center of origin for P. polionotus was peninsular Florida 
where there is the greatest amount of protein polymorphism. 
Under Bowen?s evolutionary scenario for P. polionotus (Fig. 1) the beach forms 
are of recent origin (< 10 000 years).  This implies that, at the genetic level, beach forms 
should be very similar to inland forms, especially, according to Bowen, P. p. 
griseobractus.  Peromyscus polionotus griseobractus inhabits what is today the mainland 
area of Eglin AFB.  Bowen believed that griseobractus and a modified form of 
griseobractus played a determining role in the evolution of all beach forms.  However, 
according to Bowen, other inland forms were also involved in the founding and 
subsequent evolution of the beach subspecies.  For example, the beach forms allophrys 
  
 13
and peninsularis were each considered to be derived from P. p. sumneri and a modified 
form of griseobractus.  Peromyscus polionotus ammobates, in Alabama, was 
hypothesized to be a descendent of griseobractus modified by intermittent contact with 
P. p. polionotus.  Bowen hypothesized that P. p. trissyllepsis and P. p. leucocephalus 
were the two most recent beach forms, with leucocephalus being derived from 
griseobractus (pre-peninsularis stock) and trissyllepsis resulting from the admixture of 
griseobractus, ammobates, and leucocephalus. 
Bowen (1968) further hypothesized that P. polionotus had originated from 
maniculatus stock, and that at least two forms of polionotus existed by the Yarmouth 
interglacial stage of the Pleistocene.  He based his phylogeny on the occurrence of relict 
shorelines that were associated with interglacial cycles.  However, these shorelines are 
today thought to be older than previously estimated (Winker & Howard 1977) thus, 
Bowen's use of relict shoreline refugia may be correct but his time periods may be 
significantly incorrect.  Bowen also focused primarily on retreating sea levels to devise 
his phylogeny.  He believed that encroaching sea levels would inundate marshes to the 
north of frontal dunes creating islands.  Beach forms would then be isolated on these 
islands and eventually lost as the islands were submerged.  Thus, with each marine 
encroachment existing beach forms were lost, which, during glacial periods, inland forms 
would again colonize more typical beach habitat.  Differentiation of beach and inland 
forms was hypothesized to have occurred because selection pressures within the beach 
habitat favored pigment reduction. 
In relation to extant beach forms, Bowen?s latest and most significant shoreline 
(Silver Bluff) arose 6000 to 4000 YBP.  This is significant because it was viewed as the 
  
 14
predominant event that allowed the recent colonization of the beach, thus leading to the 
present day beach forms.  However, there is some debate concerning the origin of the 
Silver Bluff shoreline.  The Silver Bluff Sequence may be the result of a significant 
slowing in the rise of the sea level, thus, allowing the accretion rate to increase (Wanless 
et al. 1994).  This suggests that rather than a decrease in sea level an increase in beach 
deposition occurred, which could have a significant effect on the inferred biogeography 
of beach forms.  Whether the sea level has risen and fallen or whether the sea level rise 
slowed remains under debate (Otvos 1995; Donoghue et al. 1998).  Regardless, Bowen?s 
conclusion of a very recent divergence of extant beach forms and their relation to one 
another through a common ancestor (P. p. griseobractus) is a testable hypothesis. 
My original questions concerning P. polionotus were intended to explore 
microgeographic genetic structuring.  As part of this I sequenced a hypervariable region, 
D-loop, of the mitochondrial DNA intending to track maternal lineages within a 
population of the Alabama beach mouse (P. p. ammobates).  It soon became apparent, 
however, that the P. p. ammobates population was probably comprised of a single 
mitochondrial lineage.  I viewed this as significant because this beach population was 
widely believed to have experienced recent, recurring gene flow from the adjacent 
mainland (Howell 1920; Bowen 1968; Selander 1970b) until 100 years ago when the 
intracoastal canal was opened.  In fact, P. p. ammobates has been offered as an example 
of an organism in a unique environment that failed to change phenotypically because of 
sufficient gene flow from the more typical environment of the organism (Howell 1920).  
Given the initial result, it became obvious that I needed to sample and sequence the same 
mtDNA region of the adjacent mainland population, as well as, other polionotus 
  
 15
subspecies/populations.  It also became apparent from the accumulating data that 
empirical results on the patterns of divergence, timing of divergence events and 
geographical genetic variation were not consistent with Bowen?s (1968) postulated 
evolution of P. polionotus.  Resolution of these issues became a major focus for my 
dissertation because of the direct implications for my primary research model, genetic 
structuring within a P. p. ammobates population.  Before I could fully investigate 
microgeographic structuring and its affects on retarding the loss of genetic variation in a 
semi-isolated population I needed to be confident in my understanding of historical 
dynamics.  My preliminary results suggested little evidence of gene flow into the P. p. 
ammobates population from the adjacent mainland population.  In fact, adjacent interior 
haplotypes appeared to be quite diverged from the beach haplotype.  Therefore, I viewed 
it as critical that the hypotheses put forth by Bowen (1968) and others concerning genetic 
variation within P. polionotus and its geographic placement throughout the Florida 
panhandle region be tested.  Resolution, or clarification, of past premises or disputes is 
likely to be possible using gene sequence data with contemporary analytical methods of 
analyses.  Most importantly, my results should have important implications broader 
questions concerning genetic structure of insular populations. 
My goal, then, was to investigate phylogenetic relationships, gene flow, and 
geographical arraignment of genetic diversity within P. polionotus using Bowen?s (1968) 
hypothesized phylogeny as the initial model.  Bowen focused primarily on subspecies of 
the Florida Panhandle and used crossing experiments, standard body measurements and 
pelage reflectance values to construct his phylogeny.  However, at the time of his 
publication, the crossing experiments were unfinished, thus, they only partially 
  
 16
contributed to his results.  Regardless, he offered the hypothesis that P. maniculatus gave 
rise to P. colemani in turn giving rise to P. polionotus.  Bowen then postulated the 
evolution of other inland forms, as well as, beach forms.  All of the beach subspecies, 
according to Bowen, were derived from inland forms during the Holocene, thus, leading 
to the conclusion that they are less than 10 000 years old.  To test these hypotheses, I 
made use of contemporary gene sequence analyses such as phylogenetic reconstruction, 
nested clade analysis, and estimated historical demographic parameters and population 
trends.  Using these methods, I evaluated genetic variation of P. polionotus within the 
Gulf Coast region to: infer relationships among populations of insular and interior 
populations; test biogeographical scenarios concerning geographical placement of genetic 
variation; infer patterns and degrees of gene flow and finally to estimate approximate 
times when lineages may have split.   
Methods 
Samples 
Tissue samples of Peromyscus polionotus were primarily obtained from toe clips.  
Additional muscle/organ tissue was collected from trap mortalities and/or museum 
specimens.  Peromyscus maniculatus tissue was obtained from tail snips, muscle, and 
organs from wild caught or museum specimens.  Samples of Peromyscus maniculatus 
bairdii {approximate location Ann Arbor, Michigan, BW1 (PMB01), BW2 (PMB02), 
BW3 (PMB03)}, Peromyscus maniculatus sonoriensis {approximate location White 
Mountain Research Station, California, SM4 (PMS04), SM5 (PMS05), SM6 (PMS06)}, 
and Peromyscus polionotus subgriseus {Ocala National Forest, Florida, PO10076 
(MCFL76), PO10077 (MCFL77), PO10078 (MCFL78), PO10299 (MCFL99)} were 
  
 17
obtained from the Peromyscus Genetic Stock Center (University of South Carolina).  
Tissue samples of Peromyscus polionotus niveiventris {Ppn-2917 (NIV17), Ppn-2936 
(NIV36), Ppn-762 (NIV62)} were obtained from Dr. Christopher L. Parkinson 
(University of Central Florida).  Tissue samples from Peromyscus keeni {Kittitas Co, 
Washington, GK6601 (PK01), GK6621 (PK21), GK6622 (PK22), GK6623 (PK23)} and 
P. maniculatus pallescens {Robertson Co, Texas, GK6590 (PMP90), GK6592 (PMP92), 
GK6593 (PMP93), GK6594 (PMP94)} were obtained from Dr. Ira Greenbaum (Texas A 
& M University).  Most tissue samples were preserved in 1.5-ml microcentrifuge tubes 
filled with 100% ethanol.  DNA was isolated from 145 P. polionotus ssp, along with 
three P. m. bairdii, three P. m. sonoriensis, four P. keeni, four P. m. pallescens, six P. 
gossypinus, and three P. leucopus for this study.  A total of seventy-eight individuals 
(Table 1) were used in at least one of the analyses presented here. 
DNA was extracted using QUIAGEN DNeasy? tissue kits (QUIAGEN
?
, 
Valencia, California).  A fragment of approximately 2.7 kb was amplified and 2452 base 
pairs were used for analyses.  This fragment contained partial 12s rRNA (137 bp), partial 
cytochrome-b (1137 bp), and complete D-loop region (971 bp) along with tRNA-
threonine (68 bp), tRNA-proline (66 bp), and tRNA-phenylalanine (68 bp). It was 
amplified using primer CB3R-L (5?-CATATTAAACCCGAATGATATTT-3?) and 
primer 12SAR-H (5?-ATAGTGGGGTATCATATCCCAGTT-3?) (Palumbi et al. 1991).  
Lengths of gene regions within the sequence are approximate as they varied among 
species.  Identification of regions was based on alignment with positions 14 152 - 1024 of 
the Mus musculus mtDNA genome (GenBank accession number, AB042432).  Not all of 
the nucleotides contained within the P. polionotus sequences were assigned using the M. 
  
 18
musculus alignment.  For example, two nucleotides (1206, 1207) either at the end of 
tRNA-Thr or the beginning of tRNA-Pro were not accounted for in the Mus sequence 
inferring that one or both polionotus tRNAs are longer.  To sequence the entire 2.7 kb 
region internal sequencing primers developed within the lab of M. C. Wooten (Auburn 
University) and consisted of Dls-01F (5?-AAGGACTAACCCCCACCATC-3?), and Dle-
01R (5?-ATAAGGCCAGGACCAAACCT-3?), PL276-R (5?-
TAACCCTGCTTGTCCAAATG-3?), PL792-F (5?-TTTGGGGTTTGTCAAGGATA-3?), 
BF1; (5?-AGGACAACCAGTCGAACACCCATT-3?) and BR1; (5?-
TGGCTGGCACGAAATTTACCAACC-3?) were used.  These primers are 
approximately 200 nucleotides internal of CB3R-L and 12SAR-H respectively.  To 
sequence more of cytochrome-b gene (approximately 1 kb) another amplification using 
primer 14 724-F (5?-CGAAGCTTGATATGAAAAACCATCGTTG-3?) (Palumbi et al. 
1991) and primer TD20-R (5?-ACTATCAGGGCAATGGGTGA-3?) was carried out.   
The Polymerase Chain Reaction (PCR) method was used to amplify specific 
regions of DNA to determine the nucleotide sequence.  PCR reactions were conducted 
using QUIAGEN Taq PCR Core Kit and Promega PCR Nucleotide Mix.  Reactions were 
run in a Perkin Elmer GeneAmp PCR System 2400, a Hybaid Omn-E HBTRE02, or a MJ 
Research PTC-200.  Amplification reactions were conducted in 31 ul volumes 
containing, 12.6 - 241.8 ng DNA, 0.32 uM primer, 0.75 mM MgCl
2
, 200-uM dNTPs, 1X 
buffer, 1.25 U Taq.  PCR conditions were: 35 cycles of 94?C (30 s) denaturing, 56? C 
(45 s) annealing, and 72?C (45 s) extension, followed by one 6 min period at 72?C.  
Product was determined to be present or absent by means of UV visualization.  Product 
from successful reactions was cleaned using QUIAGEN QIAquick
?
 PCR Purification Kit 
  
 19
or QUIAGEN QIAquick
?
 Gel Extraction Kit and sequenced by the Auburn Genomics 
and Sequencing facility (Auburn University).  Nucleotide sequence determination was 
completed using an ABI Prism? 3100 Genetic Analyzer (Applied Biosystems
?
, Inc., 
Foster City, California). BIOEDIT v5.0.9 (Hall 1999) software was used to manipulate, 
proof and hand manipulate sequence data.  CLUSTALX v1.81 (Thompson et al. 1997) was 
used to create global alignment files in NEXUS format. 
A total of 145 samples were obtained from 32 locations across the southeastern 
United States (Fig. 2).  A majority of the inland tissue samples were obtained from the 
Auburn University tissue stock collection.  Florida samples were obtained from various 
inland localities including Okaloosa County, Walton County, Jackson County, Liberty 
County, Wakulla County, Suwannee County, and Marion County (Fig. 2).  Inland 
Alabama samples were obtained from South Baldwin County, Lee County, Geneva 
County, and Russell County.  Georgia samples were obtained from Talbot County.  
Beach locations included samples from Florida and Alabama.  Samples were collected 
from Santa Rosa Island, Florida (P. p. leucocephalus), Topsail Hill State Preserve, 
Florida, Grayton Beach State Recreation Area, Florida, and Shell Island, Florida (P. p. 
allophrys).  In addition, tissues from obtained from St. Joseph State Park, Florida (P. p. 
peninsularis), Florida Point, Florida, Perdido Key, Florida, Johnson Beach, Florida, 
Perdido Key, Florida (P. p. trissyllepsis), Cape Canaveral National Air Force Station (P. 
p. niveiventris), and Bon Secour National Wildlife Refuge, Alabama (P. p. ammobates).          
Phylogenetic Hypothesis 
 Phylogenetic topologies were constructed using maximum parsimony (MP) and 
maximum likelihood (ML) optimality criterion in PAUP* v4.0b10 (Swofford 2002).  A 
  
 20
Bayesian analysis was conducted using MRBAYES v3.0 (Huelsenbeck & Ronquist 2001).  
Phylogenetic reconstruction was limited to unique haplotypes and homologous 
haplotypes from different geographic locations.  Within P. polionotus alignments, two 
single base insertions-deletions were removed from 13 samples as they were considered 
to be artifacts from the sequencing process.  However, apparent insertions-deletions 
between species were considered likely to actually have occurred and, thus, were retained 
in the analysis.  The fragment that was sequenced consisted primarily of cytochrome-b 
gene (coding) and the D-loop region (noncoding), separate phylogenies were developed 
using cytochrome-b gene, the D-loop region and the two sequences together.  Each data 
set was analyzed twice, once with P. leucopus and P. gossypinus as outgroups and once 
with P. keeni as the outgroup.  For both the MP and ML analysis, robustness and nodal 
support were evaluated using 1000 bootstrap iterations (Felsenstein 1985).  To search for 
the best tree, a heuristic search was conducted using 100 random additions and TBR 
branch swapping in the parsimony analysis and 10 random additions and TBR branch 
swapping in the maximum-likelihood analysis.  When applicable in Bayesian analysis, 
data sets were analyzed by assigning each partition (gene region) its individual model and 
prior probability distributions.  Partitions were unlinked to conduct partitioned likelihood 
analyses.   
For the MP analysis, weighted and unweighted phylogenies were generated.  
Weighting for the parsimony analysis was based on estimating a transition/transversion 
ratio and 1
st
, 2
nd
, 3
rd
 codon position changes for cytochrome-b.  These were estimated 
using the Hasegawa, Kishino, and Yano model (HKY85: Hasegawa et al. 1985) within 
PAUP v4.0b10.  Likelihood-ratio tests along with AIC scores derived using MODELTEST 
  
 21
v 3.5 (Posada & Crandall 1998) were used to select a nucleotide substitution model for 
ML analysis.  When likelihood-ratio tests and AIC scores were in disagreement, an 
analysis was conducted using each selected model.  Parameter abbreviations in 
MODELTEST are, the number of substitution types (Nst), transition/transversion ratio 
(Tratio), substitution model (Rmat), rate of evolution for variable sites (Rates), gamma 
distribution shape parameter (?), and the proportion of invariable sites (I).  Models for 
the Bayesian analysis were chosen using MRMODELTEST v1.1b (Nylander 2002).  Again, 
if the log likelihood score and the AIC score were in disagreement each model was used 
and a separate analysis was conducted.  For the Bayesian analyses, four chains were run 
for two million generations.  Sampling occurred every 10 generations to produce a 
posterior probability distribution of 200 001 trees.  A burn-in value was determined by 
graphing the log likelihood scores and determining the point at which relative stationarity 
had been achieved.  Once stationarity was reached, a 50% majority-rule consensus tree 
was constructed from the remaining trees. 
Nested Clade Analysis 
 Following the method of Templeton et al. (1995), a nested clade analysis was 
conducted.  Nested clade analysis (NCA) tests the assumption, or null hypothesis, that 
haplotypes or clades of haplotypes are randomly dispersed with respect to space.  
Primarily, NCA allows one to infer historical causes for the present spatial distribution of 
haplotypes (i.e. range expansion, colonization, or vicariance events) through the genetic 
marker under analysis.  Thus, individual effects of recurrent gene flow and historical 
events can be identified and inferences made about their role in the present 
phylogeographic structure.  Based on work by Hudson (1989) using coalescent theory, 
  
 22
this method was designed to resolve population level divergence by calculating the 
overall limits of parsimony and developing a diagram of haplotype relationships.  This is 
accomplished through the construction of a 95% statistical parsimony network of 
haplotypes (Templeton et al. 1992).  Haplotypes are then hierarchically subdivided into 
nested clades using the algorithm and following the rules of Templeton et al. (1987) and 
Templeton & Sing (1993).   
In order to employ this method, it first must be shown that the use of parsimony is 
justified (Sober 1988).  This assumption can be tested by estimating the parameter Theta 
(?) (Watterson 1975) where ? = 4N
e
?, N
e
 is the effective population size and ? is the 
mutation rate per nucleotide.  For the mitochondrial genome ? = 2N
e
?.  For this analysis 
? was estimated using the software DNASP version 4.0 (Rozas et al. 2003) then used to 
solve for H in equation 1 of Templeton et al. (1992).  Program TCS version 1.18 (Clement 
et al. 2000) was used to construct the cladogram and estimate probabilities.  TCS uses the 
algorithm of Templeton et al. (1992).  Clades are nested by moving one mutational step 
inward from the tips uniting haplotypes into one-step clades.  After the tip clades have 
been nested, interior clades are nested as additional one-step clades.  The next round of 
nesting unites two-step clades, again, beginning with the tips.  The nesting continues until 
the final round results in a single clade that encompasses all haplotypes.   
The final hierarchical clades were analyzed in GEODIS (Posada et al. 2000).  The 
GEODIS analysis investigates the association of clades and nested clades in two different 
ways.  First, locations are treated as categorical variables, and a permutational 
contingency analysis is used to test clades against their geographical location.  Secondly, 
a more conclusive test is conducted utilizing geographical distances (Templeton et al. 
  
 23
1995).  Within this analysis two primary parameters are calculated, clade distance (D
c
) 
and nested clade distance (D
n
).  D
c 
is a measure of the geographical spread of a clade and 
D
n
 measures the distribution of a clade relative to clades of the same nesting level.  
GEODIS begins with the null hypothesis that haplotypes are dispersed randomly with 
respect to geographical location.  When the null was rejected (5% level) an inference key 
(http://darwin.uvigo.es/download/GEODISkey_14Jul04.pdf) was used to draw inferences 
concerning population processes that may underlie the observed contemporary patterns.  
Nested clade analysis was used here to test for geographical associations of haplotypes 
using three different models of phylogeny based on TCS parsimony networks.  The first 
model used the statistical parsimony network generated by the program TCS (Model I).  
Secondly, haplotypes from the Florida peninsula were removed prior to generating a 
parsimony network through TCS (Model II).  Finally, a third model was analyzed with the 
parsimony network generated by TCS without the Florida peninsula haplotypes (Model 
III); however, the branch containing haplotypes from central Alabama and Georgia 
(LCAL01, LCAL02, and TCGA01) was manually moved to branch from a node basal to 
the beach clade.  This branch was moved to be consistent with the core phylogeny 
estimated by Bayesian, MP, and ML analyses. 
Neutral evolution and demographic analysis 
For clades or groups identified by phylogenetic analysis, measures of diversity 
were estimated for cytochrome-b, D-loop, and the combined gene regions.  Measures of 
diversity were estimated using DNASP version 4.0 (Rozas et al. 2003) and included the 
number of haplotypes (h), the number of polymorphic sites (S), the total number of 
mutations (Eta), haplotype diversity (Hd), nucleotide diversity (?), theta (?), and the 
  
 24
average number of nucleotide differences (k).  The assumption of neutrality for the 
genetic markers under analysis was tested using several neutrality test statistics.  
However, there is the potential for multiple processes to occur within a population that 
may produce similar outcomes from some neutrality test statistics (i. e. selection, 
population expansion/contraction, population admixture, population subdivision).  
Neutrality tests were conducted using DNASP version 4.0 (Rozas et al. 2003).  These tests 
included Tajima?s D (Tajima 1989), Fu & Li?s D, Fu & Li?s D*, Fu & Li?s F, and Fu & 
Li?s F* (Fu & Li 1993), and McDonald-Kreitman (McDonald & Kreitman 1991).  
Tajima?s D (Tajima 1989) statistic tests for violations of neutrality by comparing two 
measures of ? and does not require an outgroup.  Neutral evolution predicts that both 
estimates of ? would be the same.  Tajima?s D statistic can be either positive or negative.  
Values greater than zero indicate higher average heterozygosity than expected which can 
indicate heterozygote advantage or population contraction.  Values less than zero indicate 
a lower than expected average heterozygosity, as expected to result from events such as 
purifying selection, population expansion, or population admixture. 
While Fu & Li?s tests permit but do not require an outgroup, the analysis was 
conducted both with an outgroup (D, F) and without an outgroup (D*, F*).  Because Fu 
& Li?s tests compare mutations that occurred in the distant past with mutations that 
occurred in the recent past, an outgroup if available is preferred (Fu & Li 1993).  Fu & 
Li?s D and F tests were conducted for each P. polionotus clade using P. maniculatus as 
an outgroup and for selected each P. polionotus clades using a P. polionotus from a 
different clade as an outgroup and for each P. polionotus clade without an outgroup.  A 
final neutrality test, the McDonald-Kreitman test (McDonald & Kreitman 1991) classifies 
  
 25
polymorphic sites in two ways, as differences between species or differences within 
species and whether the change is synonymous or nonsynonymous.  This test does not 
assume that the population has reached equilibrium (i.e. the population may be expanding 
or in stasis).  The McDonald-Kreitman test tests whether the ratio of nonsynonymous to 
synonymous changes between species is equal to that within a species.  Neutrality tests 
were primarily intended to test the neutrality of nucleotide variation.  However, they have 
inherent issues with limiting assumptions, and their outcomes may be open to competing 
interpretations.  For instance, positive and negative values of Tajima?s D statistic can 
simultaneously reflect different types of selection, population expansion, or population 
contraction. 
Analysis of intraspecific demographic parameters was conducted using between 
clade comparisons.  Clades that were analyzed were derived through statistical parsimony 
(TCS), maximum parsimony, maximum likelihood, and Bayesian phylogenetic analysis.  
Maximum likelihood (ML) estimates of the population parameters ? and exponential 
growth rate (g, scaled to reflect the number of mutations) were obtained using the 
program FLUCTUATE (Kuhner et al. 1998).  FLUCTUATE is available from 
http://evolution.genetics.washington.edu/lamarc.html.  FLUCTUATE assumes there is no 
selection or recombination.  It utilizes the Metropolis-Hastings sampler to search the 
space of possible genealogies.  Under the assumption of growth, the program accepts an 
initial genealogy, ?, and g which it then rearranges.  It accepts or rejects the rearranged 
genealogy based on the probability of the data given the genealogy.  FLUCTUATE can 
estimate its own beginning genealogy or have one entered along with an initial ? and g 
values.  In my analysis, a method of 10 short chains with 100 000 steps and 2 long chains 
  
 26
with 200 000 steps while sampling every 20
th
 step was used.  DNASP version 4.0 was used 
for initial estimates of ? and when the population was allowed to grow the initial value of 
g was set at 1.  Genealogies for selected clades were estimated under the assumption of 
growing and stationary population models.  The growth models were compared and 
assessed by the 95% confidence intervals derived using the companion zero-growth 
model. 
Molecular clock 
As an initial estimate of time since P. polionotus and P. maniculatus last shared a 
common ancestor, percent divergence between a representative P. polionotus haplotype 
(P. p. ammobates) and the three P. maniculatus subspecies (pallescens, bairdii, 
sonoriensis) was estimated.  For comparison, percent divergence was estimated for P. p. 
ammobates between a sample of other polionotus, as well as, an estimate between the 
maniculatus subspecies.  Peromyscus keeni was compared with ammobates, sonoriensis, 
and leucopus.  Also, percent divergence was estimated between P. leucopus and P. 
gossypinus.  Likelihood ratio tests (LRT) were conducted to determine the 
appropriateness of adding a molecular clock to the nucleotide substitution models.  For 
each ML data set, a clock assumption was added to the model selected by MODELTEST.  
The log-likelihood values of each data set with and without a clock assumption were 
compared and P-values calculated.  Fossil dates of P. polionotus (Ruez 2001) were used 
to infer absolute times from genetic distances.  Unfortunately, only one fossil calibration 
point was available, it was based on the estimated age of the Florida sinkhole locale 
(Inglis 1C).  Ruez (2001) estimated the age of fossil deposits at the Inglis IC site to be 
between 2.01 - 1.78 Ma.  For inferring time back to specific nodes, branch lengths were 
  
 27
estimated using maximum likelihood.  To infer the age of nodes by a second method, a 
substitution rate (r) was measured using equation 5.10 (Li & Graur 1991) and equation 
5.12 (Nei 1987).  The estimated substitution rate could then be used to produce estimates 
of time to nodes for which there was no fossil evidence. 
RESULTS 
Phylogenetic analysis 
The majority of this project was conducted using portions of mtDNA sequences 
that were 2449 base pairs in length.  However, phylogenetic analysis of the complete 
sequence with Peromyscus leucopus and Peromyscus gossypinus as outgroups was 
conducted using sequences of 2452 base pairs in length.  The additional three nucleotides 
were gained when leucopus and gossypinus were added to the alignment.  Pairwise 
analyses of the D-loop of P. polionotus, P. maniculatus, P. leucopus, P. keeni, and P. 
gossypinus indicate an insert in P. polionotus that is approximately 74 base pairs in 
length.  This insert, based on sequence alignment data with Mus musculus (GenBank 
ABO42432), appears to be located near the two ETAS regions of the D-loop (Sbisa et al. 
1997) and may be a repeat of either ETAS region 1 or ETAS region 2.  For phylogenetic 
analyses of P. polionotus using other Peromyscus as outgroup haplotypes, the large insert 
created alignment problems.  Because the principle goal of this project was to determine 
the relationship within P. polionotus, a consensus sequence of the insert was created from 
the polionotus alignment and placed within outgroup haplotypes at the corresponding 
portion of the sequence.  I feel confident in the placement of the insert within outgroup 
haplotypes due to the strong alignment on either side of the insert (Fig. 3). 
  
 28
Regions of the mtDNA sequence used in this study were as follows, partial 
cytochrome-b gene is 1 - 1137, tRNA-Thr is 1138 - 1205, tRNA-Pro is 1208 - 1273, the 
D-loop region is 1274 - 2244, tRNA-Phe is 2245 - 2312, and 12s rRNA is 2316 - 2449 or 
3452.  Using the complete mtDNA sequence of Mus musculus (GenBank AB042432) the 
corresponding genes or gene regions would be: partial cytochrome-b 14 152 - 15 289, 
tRNA-Thr 15 290 - 15 356, tRNA-Pro 15 357 - 15 423, D-loop 15424 - 16 395, tRNA-
Phe 1 - 68, and 12s rRNA 70 - 1024.  There are two bases between tRNA-Thr and tRNA-
Pro and three bases between tRNA-Phe and 12s rRNA that are not accounted for based 
on the alignment with Mus.  The 74 bp insert appears to be, from base 1428-1502 of my 
fragment to 15 578 - 15 650 in the Mus alignment. 
Phylogenetic topologies were constructed under two outgroup scenarios.  
Dependent upon the haplotype used as the outgroup, either 44 sequences or 42 sequences 
were used for each estimated topology.  Outgroup samples were from P. maniculatus 
bairdii, P. m. sonoriensis, P. keeni, P. m. pallescens, P. gossypinus, and P. leucopus.  The 
data used for constructing topologies consisted of different polionotus haplotypes, 
individual polionotus with the same haplotype but from different geographic locations, 
and outgroup individuals.  The data set using P. gossypinus, and P.  leucopus as outgroup 
haplotypes consisted of  44 sequences comprised of 2452 nucleotides.  For analysis, the 
sequence data was divided into three primary sets, the total sequence (2452 bp) partial 
cytochrome-b gene (1137 bp) and the complete D-loop region (971 bp).  The entire 
sequence was composed of a portion of cytochrome-b, the D-loop region, three transfer 
RNA?s (~204 bp), and ~140 bp of 12s rRNA gene.  The cytochrome-b sequence data set 
consisted of 33 haplotypes, S = 193, ?  = 0.0202, k = 22.99, and Hd = 0.977.  The D-loop 
  
 29
region consisted of, 36 haplotypes, S = 241, ?  = 0.0294, k = 28.37, and Hd = 0.994.  
Diversity analysis of the complete sequence consisted of 39 haplotypes, S = 434, ?  = 
0.0217, k = 52.97, and Hd = 0.994.  Another set of analyses was conducted with P. keeni 
as the outgroup along with individuals of P. m. bairdii, P. m. sonoriensis, and P. m. 
pallescens (Table 2).  The total sequence length was 2449 base pairs with 38 haplotypes, 
S = 267, ?  = 0.0141, k = 34.54, and Hd = 0.993.  Cytochrome-b gene consisted of 31 
haplotypes, S = 121, ? = 0.0126, k = 14.27, and Hd = 0.974.  The D-loop portion of the 
sequence consisted of 34 haplotypes, S = 132, ? = 0.0198, k = 19.21, and Hd = 0.988.  
These two data sets were analyzed using maximum parsimony, maximum likelihood, and 
Bayesian methods. 
The ML model selection results were: cytochrome-b gene with an outgroup of P. 
leucopus and P. gossypinus, the HKY + ? (?: shape parameter of the gamma 
distribution) model was identified by log likelihood score (-1nL = 2994.17) and the TVM 
+ ? model was identified by AIC analysis (AIC = 5987.24).  Estimates of parameters for 
the HKY + ? (Hasegawa et al. 1985) model were, base frequencies A = 0.3282, C = 
0.2807, G = 0.1222, and T = 0.2689, Nst = 2, Tratio = 5.8576, rates = gamma, ? = 
0.0672, and I = 0.0000.  Estimated parameters for the TVM + ? model were, base 
frequencies A = 0.3232, C = 0.2766, G = 0.1261, and T = 0.2740, the substitution rate 
matrix was, [A-C] = 26126.14, [A-G] = 170050.33, [A-T] = 8973.04, [C-G] = 6208.09, 
[C-T] = 170050.33, and [G-T] = 1.0000, rates = gamma, ? = 0.0812, and I = 0.0000.   
For the cytochrome-b region with P. keeni as the outgroup, the selected model by 
log likelihood (-1nL = 2427.31) was HKY + ? and by AIC score (AIC = 4859.24) was 
GTR + I (I: proportion of invariable sites) (Lanave et al. 1984).  The parameters from the 
  
 30
HKY + ? model were, base frequencies A = 0.3242, C = 0.2666, G = 0.1328, and T = 
0.2764, Nst = 2, Tratio = 9.7396, rates = gamma, ? = 0.0131, I = 0.0000, and for the GTR 
+ I model, base frequencies A = 0.3208, C = 0.2677, G = 0.1295, and T = 0.2819, Nst = 
6, rate matrix = [A-C] = 12981416.00, [A-G] = 169401536.00, [A-T] = 3301110.75, [C-
G] = 3204468.50, [C-T] = 112213536.00, and [G-T] = 1.00, rates = equal, and I = 
0.7866.   For the D-loop region with P. leucopus and P. gossypinus as the outgroup, 
MODELTEST selected the model HKY + I + ? from both log likelihood score (-1nL = 
3063.81) and AIC score (AIC = 6139.61).  Parameters for this model were, base 
frequencies A = 0.3375, C = 0.2242, G = 0.1127, and T = 0.3256, Nst = 2, Tratio = 
3.2048, rates = gamma, ? = 0.7973, I = 0.4977.  For the D-loop region with P. keeni as 
the outgroup, MODELTEST selected HKY + I + ? (-1nL = 2455.72) by both log likelihood 
score and AIC score (AIC = 4923.43).  Parameters for these models were, base 
frequencies A = 0.3285, C = 0.2279, G = 0.1152, and T = 0.3284, Nst = 2, Tratio = 
5.1041, rates = gamma, ? = 0.8933, I = 0.6216.   
For the combined cytochrome-b gene and D-loop region with P. keeni as the 
outgroup, MODELTEST selected HKY + I + ? by both log likelihood (-1nL = 5555.24) and 
AIC score (AIC = 11122.48).  Parameters for the model were, base frequencies A = 
0.3345, C = 0.2434, G = 0.1289, and T = 0.2932, Nst = 2, Tratio = 6.5150, rates = 
gamma, ? = 0.9817, I = 0.6719.  Finally, using cytochrome-b gene and the D-loop region 
with an outgroup of P. leucopus and P. gossypinus MODELTEST selected a single model, 
HKY+I+ ? (-1nL = 6822.79, AIC = 13657.59).  Parameters for the model were, base 
frequencies A = 0.3386, C = 0.2475, G = 0.1232, and T = 0.2907, Nst = 2, rates = 
  
 31
gamma, ? = 0.8185, I = 0.5795.  These maximum likelihood models were used to 
construct phylogenies using PAUP.   
MRMODELTEST v1.1b (Nylander 2002) was used to determine appropriate models 
for Bayesian analysis using the software MRBAYES 3.0 (Huelsenbeck & Ronquist 2001).  
Using the sequence containing both cytochrome-b and the D-loop regions 
MRMODELTEST selected the HKY + I + ? model by both log likelihood score (-1nL = 
6822.79) and AIC score (AIC = 13657.59).  Parameters for the model were: base 
frequencies A = 0.3386, C = 0.2475, G = 0.1232, and T = 0.2907, Nst = 2, Tratio = 
4.0029, rates = gamma, ? = 0.8185 and I = 0.5795.  Again, using both cytochrome-b and 
the D-loop but with P. keeni as the outgroup MRMODELTEST selected the HKY + I + ? 
model as the best fitting model by log likelihood (-1nL = 5551.71) and AIC score (AIC = 
11115.41).  The parameters were: base frequencies A = 0.3337, C = 0.2437, G = 0.1291, 
and T = 0.2935, Nst = 2, Tratio = 6.5150, rates = gamma, ? = 0.9816, and I = 0.6717.  
Using the D-loop region with P. leucopus and P. gossypinus as outgroups, 
MRMODELTEST selected the HKY + I + ? model by log likelihood (-1nL = 3063.80) and 
AIC score (AIC = 6139.61).  The parameters were: base frequencies A = 0.3375, C = 
0.2242, G = 0.1127, and T = 0.3256, Nst = 2, Tratio = 3.2048, rates = gamma, ? = 
0.7973, and I = 0.4977.  Using the D-loop region with P. keeni as the outgroup 
MRMODELTEST selected HKY + I + ? model by log likelihood (-1nL = 2455.72) and AIC 
score (AIC = 4923.43).  The parameters were: base frequencies A = 0.3285, C = 0.2279, 
G = 0.1152, and T = .03284, Nst = 2, Tratio = 5.1041, rates = gamma, ? = 0.8933, and I = 
0.6216.  Using cytochrome-b gene and P. leucopus and P. gossypinus as outgroups 
MRMODELTEST selected GTR + ? by log likelihood score (-1nL = 2985.61) and AIC 
  
 32
score (AIC = 5989.23).  The parameters were: base frequencies A = 0.3233, C = 0.2764, 
G = 0.1265, and T = 0.2738, Nst = 6, rate matrix was [A-C] = 5599042.50, [A-G] = 
35834508.00, [A-T] = 1921158.63, [C-G] = 1327221.00, [C-T] = 36681148.00, and [G-
T] = 1.00, rates = gamma, ? = 0.0825, and I =0.0000.  Using cytochrome-b gene with P. 
keeni as the outgroup MRMODELTEST selected the HKY + ? model by log likelihood 
score (-1nL = 2427.31) and GTR + I model by AIC score (AIC = 4859.24).  Parameters 
for the models were, respectively: base frequencies A = 0.3242, C = 0.2666, G = 0.1328, 
and T = 0.2764, Nst = 2, Tratio = 9.7396, rates = gamma, ? = 0.0160, I = 0.0000, base 
frequencies A = 0.3208, C = .02677, G = 0.1295, and T = 0.2819, Nst = 6, rate matrix 
was [A-C] = 35658768.00, [A-G] = 465339968.00, [A-T] = 9067707.00, [C-G] = 
8802202.00, [C-T] = 308245280.00, and [G-T] = 1.0000, rates = equal, and I = 0.7866.   
Twenty-seven topologies were developed that depict the hypothesized 
phylogenies generated from the multiple phylogenetic approaches.  While the topologies 
vary substantially, significant agreement was observed (Fig. 4 - 30).  Consistent among 
the MP and ML trees was the 100% bootstrap support for P. polionotus as a 
monophyletic clade indicating reciprocal monophyly.  Bayesian models also strongly 
supported P. polionotus as a monophyletic clade with 100% posterior probability.  
Twenty-eight topologies were estimated based on maximum likelihood, maximum 
parsimony, Bayesian, and statistical parsimony methods.  Eight maximum likelihood 
topologies were constructed (Fig. 4 - 11).  Maximum likelihood analysis of cytochrome-b 
region (1137 bp), produced four topologies that were constructed from the best-fit models 
selected by MODELTEST v3.5.  The number of models (28) resulted from testing the 
assumption of different outgroups, and, under each outgroup different models may have 
  
 33
been selected by likelihood ratio tests and AIC scores.  Model selection for the 
cytochrome-b gene was the most variable.  Using the ML method, four separate models 
were selected as best-fit models.  With P. leucopus and P. gossypinus as outgroups the 
likelihood ratio tests (LRT) selected the HKY + ? model, but the AIC score best-fit 
model was TVM + ?.  Using P. keeni as the outgroup the LRT best-fit model was the 
HKY + ? while the AIC best-fit model was GTR + ?.  Using these different outgroups 
did not change the log likelihood model selection for the cytochrome-b portion of the 
sequence.  However, with both outgroups the AIC method concluded adding more 
parameters was appropriate and selected a different model for each outgroup.  All 
models, however, were in agreement concerning rate heterogeneity across sites.  Using 
MRMODELTEST for model selection by Bayesian analyses with the cytochrome-b region, 
the only disagreement in model choice occurred when P. keeni was used as the outgroup.  
In this case, LRT selected GTR + I as the best-fit model while the AIC method selected 
the HKY + ? model as most appropriate. 
Model selection for the D-loop region was more congruent.  For the ML method, 
model selection using different outgroups did not change the model (HKY).  However, 
using P. leucopus and P. gossypinus as outgroups added the parameter of rate 
heterogeneity (?) and proportion of invariable sites (I) while using P. keeni as the 
outgroup added only the parameter of I.  For the D-loop region, LRT and AIC were in 
agreement with model selection for each outgroup.  Model selection resulting from 
Bayesian analyses of the D-loop region was in agreement regardless of the outgroup 
(HKY + ? + I).  When combining the cytochrome-b gene and the D-loop region for the 
ML phylogeny estimation, model selection by both LRT and AIC was in agreement 
  
 34
(HKY + ? + I).  Maximum likelihood model selection for the combined genes was not 
changed by either outgroup.  The Bayesian analysis of the complete sequence was 
conducted using the substitution model that was selected for each of the separate regions. 
Using ML methods for phylogenetic estimation from cytochrome-b gene with P. 
gossypinus and P. leucopus as outgroup haplotypes, LRT indicated that the best-fit model 
was HKY + ?.  However, the AIC score identified TVM + ? as the most appropriate 
model.  The HKY + ? substitution model was somewhat more resolved (Fig. 4).  It shows 
the presence of a clade comprised of seven haplotypes (inland clade II) that was not 
present in the TVM + ? model (Fig. 5).  However, the clade (inland clade II) has low 
(52%) bootstrap support value.  Otherwise, the two models produced essentially the same 
topology.  Bootstrap support values were generally low to moderate in both topologies.  
The HKY + ? model indicated the possibility of four major clades emanating from a star 
phylogeny.  The TVM + ? model, however, suggested only three clades but again 
indicated a star phylogeny.  The clades (HKY + ?) were comprised of three inland groups 
and the partial formation of a beach clade.  These clades were defined as inland clade I 
(BCAL05, RCAL01, TCGA03, BCAL04, OCFL03, BCAL01, WCFL04, MCAL01, and 
JCFL01), inland clade II (WCFL02, WCFL03, WCFL01, OCFL05, OCFL04, WCFL02, 
and OCFL06), the beach clade (PEN05, 89CI01, ALL02**, ALL05, and ALL02*).  The 
third inland group was comprised of haplotypes WCFL01, WCFL06, JCFL03, OCFL01, 
and OCFL07. 
Cytochrome-b topologies with P. keeni as the outgroup (Fig. 6, 7) were similar to 
the topologies identified with P. leucopus and P. gossypinus as outgroups.  The model 
selection, HKY + ?, by LRT for the both outgroups produced essentially the same 
  
 35
topology.  Also, model selection by AIC score (TVM + ?, GTR + ?) with 
leucopus/gossypinus and keeni as outgroups, respectively, was in near complete 
agreement on the structure of the topology.  Again, with keeni as the outgroup the HKY + 
? model was slightly more resolved but essentially the same as the HKY + ? model with 
P. leucopus and P. gossypinus as outgroups.  The same clades were present and the 
bootstrap values were only slightly different.  Cytochrome-b established the presence of 
various relationships within P. polionotus, however, the amount of overall divergence 
was low and thus yielded inadequate resolution for solving my original question.  
Cytochrome-b is a moderately evolving gene, thus the lack of a defining topology was 
expected.  However, cytochrome-b gene did establish P. polionotus as a monophyletic 
clade within a group of its sister taxa, and did indicate that P. polionotus is a significantly 
older taxon.  This finding is more in agreement with recent fossil evidence (Ruez 2001) 
than previously hypothesized by biogeographic models. 
Topologies based on data from the D-loop region (971 bp), were more resolved 
(Fig. 8, 9) than for cytochrome-b.  This outcome was expected as the D-loop contains 
hypervariable regions (Vigilant et al. 1991) that were expected to provide greater 
phylogenetic signal at the subspecific level.  Unlike the cytochrome-b region, a single 
best-fit model was identified for both outgroups (by both LRT and AIC scores).  This 
model was HKY.  However, when using keeni as the outgroup, the parameter I 
(proportion of invariable sites) was included.  When gossypinus and leucopus served as 
outgroups two parameters, I and ? (shape parameter of the gamma distribution) were 
components of the HKY model.  The two resulting topologies exhibited the same 
branching configuration with only slight differences in bootstrap values.  Haplotypes 
  
 36
placed within Inland clade I did not change regardless of gene region, model, and 
outgroup indicating strong support for this clade by the ML method.  Inland clade II 
differed slightly between D-loop and cytochrome-b topologies.  The D-loop topology 
placed haplotypes LCFL01 and SCFL01 within inland clade II but with low (model HKY 
+ ? + I 50%; model HKY + I 52%) bootstrap support.  In contrast, the beach clade 
produced by the D-loop analysis was substantially different from the cytochrome-b beach 
clade (either model).  Haplotypes (PEN05, CCFL01, TRI07, AMM01, LEU08, OCFL01, 
WCFL01, WCFL06, OCFL07, and JCFL03) comprised the D-loop beach clade while 
haplotypes (PEN05, PEN02*, ALL02**, ALL05, and ALL02*) comprised the 
cytochrome-b beach clade. 
While there was added clarification from the analyses of the D-loop region, the 
overall topology was not fully resolved.  I attempted to improve the stability of the 
topology by conducting a third round of analysis utilizing the combined gene regions 
(2449 bp).  Again, neither of the outgroup haplotypes affected model selection.  When 
either leucopus/gossypinus or keeni were used as outgroups, the model HKY + ? + I was 
selected by both LRT and AIC scores (Fig. 10, 11).  The complete fragment analyses 
yielded topologies that were most similar to the separate ML analyses of the cytochrome-
b region and D-loop that were devised through the HKY model with leucopus/ 
gossypinus as outgroups.  However, combining the regions produced topologies that were 
more resolved than the analyses of either single region.  Indeed, the beach clade now 
consisted of 16 haplotypes (LCAL01, TCGA01, PEN05, AMM01, LEU08, OCFL01, 
WCFL01, WCFL06, OCFL07, JCFL03, CCFL01, TRI07, PEN02*, ALL02**, ALL05, 
and ALL02*).  Inland haplotypes LCAL01 and TCGA01 (central Alabama and central 
  
 37
Georgia respectively) branched from the basal node within the beach clade.  Also, 
another group of inland haplotypes (OCFL01, WCFL01, WCFL06, OCFL07, and 
JCFL03) was placed within the beach clade.  Conversely, haplotypes found in Gulf 
Coastal beach populations are not found outside of the beach clade.  Inland clade I had 
the same haplotypes as the ML D-loop and cytochrome-b topologies, however in the 
combined analyses it was more strongly supported by bootstrap values.  Inland clade I 
was then supported by all the topologies and was apparently quite robust.  The HKY 
model from the cytochrome-b region supported the possible existence of at least two 
major inland clades.  The D-loop analysis also supported two major inland clades and, 
minus six haplotypes, the beach clade. The D-loop included two new haplotypes within 
inland clade II while significantly altering the beach clade.  The combined data set of 
cytochrome-b and the D-loop identified divergence points for at least three major 
lineages within P. polionotus, two inland clades and a single beach clade comprised of 
haplotypes found only within populations of subspecies along the Gulf Coast of Florida 
and Alabama.  
For the maximum parsimony (MP) analysis, six weighted and six unweighted 
topologies were constructed (Fig. 12 - 23).  Again, cytochrome-b and the D-loop were 
analyzed independently and combined into a single sequence.  Also, where applicable, 
analyses were conducted using weighted characters such as transition/transversion ratio 
and nucleotide substitutions based on codon position bias.  Weighting values for each 
data set were calculated using an HKY85 model of nucleotide substitution (PAUP* v 
4.0b10).  Analyses of cytochrome-b with P. leucopus and P. gossypinus as outgroups 
included 44 individual sequences of 1137 bp each.  Among the 44 sequences were 33 
  
 38
haplotypes with 193 variable sites of which 126 were parsimony informative.  A second 
set of cytochrome-b sequence data with P. keeni as the outgroup consisted of 42 
sequences of 1137 bp each.  The 42 sequences produced 31 haplotypes consisting of 121 
variable sites of which 65 were phylogenetically informative. 
The weighted MP analysis of cytochrome-b gene with P. leucopus and P. 
gossypinus as outgroups (Fig. 12) yielded 28 equally parsimonious trees (1646 steps) 
with a consistency index (CI) of 0.9195, a retention index (RI) of 0.9071 and a rescaled 
consistency index (RC) of 0.8341.  Weighting consisted of a 9.5:1 transition/transversion 
ratio, 1
st
 codon positions were weighted 5:1 and 2
nd
 codon positions 40:1.  The 
unweighted analysis of cytochrome-b with leucopus and gossypinus as outgroups 
revealed 77 equally parsimonious trees (272 steps) with a CI of 0.7684, a RI of 0.8205, 
and a RC of 0.6305.  Using leucopus  and gossypinus as outgroups, the MP weighted and 
unweighted cytochrome-b analyses produced similar results.  Interestingly, the weighted 
topology placed haplotype NIV17 (P. p. niveiventris) branching from a basal node in the 
P. polionotus clade, however, at low (51%) bootstrap support.  Weighting the analysis 
produced an inland clade I with the same haplotypes as estimated by the ML analyses 
with the addition of haplotype LCAL01.  Removing the constraints of the weighted 
parameters (Fig. 13), inland clade I contained only the haplotypes BCAL04, OCFL03, 
BCAL01, WCFL04, MCAL01, and JCFL01.  Inland clade II, however, was comprised of 
haplotypes LCFL01, WCFL02, WCFL03, WCFL01, OCFL05, OCFL04, WCFL02, and 
OCFL07.  Thus, inland clade II differed from the cytochrome-b ML analyses with the 
addition of haplotype LCFL01.  The D-loop ML analyses placed the LCFL01 haplotype 
within inland clade II where it remained throughout the remainder of the ML analyses.   
  
 39
Within the weighted and unweighted MP analyses of cytochrome-b using P. 
leucopus and P. gossypinus as outgroups, the beach clade fragmented in the same manner 
as the ML cytochrome-b analyses (Fig. 12, 13).  Like the ML analyses, this analysis 
produced multiple, single long branches and placed haplotypes PEN05, PEN02*, 
ALL02**, ALL05, ALL02* in one clade and haplotypes WCFL01, WCFL06, JCFL03, 
OCFL01, and OCFL07 in a second clade.  The previous two groups of haplotypes were 
placed in the beach clade in the ML analysis combining the cytochrome-b and D-loop 
region.  Inland clade II contained the same haplotypes in both MP cytochrome-b 
analyses.  Bootstrap values were essentially the same between the analyses for the two 
beach groups but were somewhat higher in the weighted analysis for inland clade II.  
Results from the two analyses, weighted and unweighted, were essentially the same with 
the exception of the removal of haplotypes LCAL01, BCAL05, RCAL01, and TCGA03 
from inland clade I in the unweighted tree.  While bootstrap values were very similar, 
(CI, RI, and RC) were higher in the weighted analysis.  The weighted analysis produced 
49 fewer equally parsimonious trees but was 1374 steps longer. 
Maximum parsimony analysis of cytochrome-b using P. keeni as the outgroup 
(Fig 14, 15) included 42 sequences of 1137 bp.  Among the 42 sequences were 31 
haplotypes with 121 variable sites of which 65 were parsimony informative.  The 
weighted analysis resulted in 14 equally parsimonious trees (805 steps) with a CI of 
0.9553, a RI of 0.9439, and a RC of 0.9017.  Weighting consisted of 
transition/transversion ratio of 18:1, 1
st
 codon position changes 5:1, and 2
nd
 codon 
position changes 37:1.  Removing the weight assumptions produced 38 equally 
parsimonious trees (149 steps) with a CI of 0.8389, a RI of 0.8863, and an RC of 0.7435.  
  
 40
The unweighted tree with P. keeni as the outgroup placed fewer of the haplotypes into 
clades.  Inland clade II and the beach clade were identical between the analyses, however, 
inland clade I differs by four haplotypes (BCAL05, LCAL01, RCAL01, TCGA03) which 
were absent from inland clade I in the unweighted analysis.  With keeni as an outgroup 
for the cytochrome-b topologies, as opposed to leucopus and gossypinus, the topologies 
were not quite as resolved for the unweighted tree but were highly congruent between the 
weighted trees.  However, bootstrap values for cytochrome-b with outgroup keeni were 
similar for congruencies that do occur. 
Two topologies (weighted and unweighted) were generated for the D-loop region 
using P. leucopus and P. gossypinus as outgroups (Fig. 16, 17).  The data set was 
comprised of sequences from 45 individuals (971 bp).  Again, an artificial insert (74 bp) 
was placed in the D-loop region of outgroup haplotypes.  Among the 45 samples were 37 
haplotypes that consisted of 208 variable sites of which 136 were informative.  Because 
the D-loop is not a coding region, only the transition/transversion ratio was weighted 
(6:1).  The weighted analysis consisted of 17 equally parsimonious trees (745 steps) a CI 
of 0.8631, a RI of 0.8922, and a RC of 0.7700.  The unweighted topology with P. 
leucopus and P. gossypinus as outgroups consisted of the same number of haplotypes and 
variable sites but the number of parsimony informative sites was reduced over the 
weighted topology to 132 from 136.  The unweighted analysis consisted of 22 equally 
parsimonious trees (311 steps) with a CI of 0.7653, a RI of 0.8433, and an RC of 0.6454.  
While the two trees are highly congruent, there were some differences.  Haplotype 
composition of inland clade I was not changed between weighted and unweighted 
analyses and contained haplotypes BCAL05, BCAL01, MCAL01, BCAL04, JCFL01, 
  
 41
OCFL03, WCFL04, RCAL01, TCGA03.  Inland clade II differed between the two 
topologies by a single haplotype (LCFL01).  Otherwise they consisted of the same 
sequences (OCFL04, OCFL06, WCFL01, WCFL02, WCFL02, OCFL05, WCFL03).  
Both analyses identified a beach clade consisting of haplotypes LCAL01, TCGA01, 
CCFL01, TRI07, PEN05, AMM01, LEU08, OCFL01, WCFL01, WCFL06, OCFL07, 
ALL05, ALL02*, PEN02*, ALL02**, and JCFL03.  The weighted analysis grouped four 
haplotypes (SCFL01, MCFL76, WCFL07, WCFL09) that roughly comprise a Florida 
peninsula clade.  The haplotypes WCFL07 and WCFL09 were collected near St. Marks, 
(south of Tallahassee, Florida), so the two haplotypes were geographically removed from 
the true Florida peninsula.  Interestingly, the unweighted analysis placed the haplotype 
SCFL01 branching from a basal node in the P. polionotus group. 
The D-loop analyses with P. keeni as the outgroup included 43 sequences with 
971 bp (Fig. 18, 19).  These sequences constitute 35 haplotypes with 132 variable sites 
where 81 were parsimony informative.  The estimated transition/transversion ratio was 
10:1.  The weighted data set produced 32 equally parsimonious trees (482 steps) with a 
CI of 0.8465, a RI of 0.9094, and a RC of 0.7698.  The unweighted data set yielded 520 
equally parsimonious trees (184 steps) with a CI of 0.7500, a RI of 0.8682, and a RC of 
0.6511.  Both D-loop data sets supported the existence of the two hypothesized inland 
clades containing the same haplotypes as hypothesized by the prior (outgroup = P. 
leucopus) MP D-loop analyses.  Using keeni as the outgroup, both the weighted and 
unweighted D-loop topologies supported inland clade I comprised of haplotypes 
(BCAL05, RCAL01, TCGA03, BCAL01, MCAL01, BCAL04, JCFL01, OCFL03, 
WCFL04).  Again, only the weighted topology placed haplotype LCFL01 in inland clade 
  
 42
II, otherwise, inland clade II was comprised of the same haplotypes as the unweighted 
topology.  These two analyses differed, however, in placement of haplotypes within the 
beach clade.  The weighted topology placed haplotypes LCAL01, TCGA01, CCFL01, 
TRI07, PEN05, AMM01, LEU08, OCFL01, WCFL01, WCFL06, OCFL07, ALL05, 
ALL02*, PEN02*, ALL02**, and JCFL03 within the beach clade.  However, the 
unweighted topology did not include haplotypes LCAL01, TCGA01, PEN02*, ALL02**, 
ALL05, and ALL02* within the beach clade.  The weighted topology roughly supported 
the same peninsula clade.  Neither the weighted nor the unweighted topology using keeni 
as the outgroup proposes a haplotype branching from a basal node among polionotus as 
the unweighted D-loop topology using leucopus and gossypinus as outgroups did.  The 
D-loop topologies, despite the outgroup haplotype, were congruent when comparing 
weighted and weighted or unweighted and unweighted.  The outgroup used to root the 
topology appeared to make little difference.   
Combining the D-loop and cytochrome-b along with the three tRNAs and the 137 
bp of 12s rRNA and using P. leucopus and P. gossypinus as outgroups, there were 44 
sequences 2452 bp long (Fig. 20, 21).  These data represented 39 haplotypes with 426 
variable sites of which 265 were parsimony informative.  The analysis produced 93 
equally parsimonious trees (1668 steps) with a CI of 0.8639, a RI of 0.8718, and a RC of 
0.7532.  The cytochrome-b weighting scheme with P. leucopus and P. gossypinus as 
outgroup haplotypes consisted of 7:1 transition/transversion ratio, 1
st
 positions weighted 
5:1, and 2
nd
 positions weighted 40:1.  The unweighted analysis produced 26 equally 
parsimonious trees (614 steps) with a CI of 0.7687, a RI of 0.8273, and a RC of 0.6359.  
As with the previous topologies there was strong bootstrap support (100%) for P. 
  
 43
polionotus as a monophyletic clade.  The weighted and unweighted topologies indicated 
an association among haplotypes BCAL05, RCAL01, TCGA03, BCAL01, MCAL01, 
OCFL03, WCFL04, BCAL04, JCFL01 to form inland clade I.  The weighted topology 
placed haplotype LCAL01 and connected it to the basal node of the clade.   
Both topologies created using the complete sequence represented inland clade II 
with haplotypes LCFL01, OCFL04, WCFL01, WCFL02, OCFL06, WCFL02, OCFL05, 
and WCFL03, as had many of the previous topologies.  Also, both topologies contained 
beach clades with haplotypes AMM01, LEU08, WCFL01, WCFL06, OCFL01, OCFL07, 
JCFL03, PEN05, CCFL01, TRI07, PEN02*, ALL02**, ALL05, and ALL02*.  However, 
the unweighted topology placed haplotypes LCAL01 and TCGA01 branching from a 
basal node in the beach clade.  The unweighted topology did not group the haplotypes 
from the Florida peninsula, but the weighted topology did contain the peninsula clade 
(SCFL01, MCFL76, WCFL07, WCFL09).  The weighted topology placed haplotype 
NIV17 branching at the basal node within the polionotus group.  However, the 
unweighted topology placed haplotype SCFL01 branching from the basal node in the 
polionotus group. 
Using P. keeni as the outgroup and weighting the data set produced 62 equally 
parsimonious trees (1002 steps) with a CI of 0.8752, a RI of 0.9092, and a RC of 0.7958.  
This data set was weighted as follows, transition/transversion ratio = 12:1, 1
st
 positions 
weighted 5:1, and 2
nd
 positions weighted 37:1.  Removing the weighting assumptions 
from the data produced 440 equally parsimonious trees (354 steps) with a CI of 0.7853, a 
RI of 0.8645, and a RC of 0.6789.  The topologies (weighted and unweighted) were 
similar to their counterparts with leucopus as the outgroup (Fig. 22, 23).  In fact, inland 
  
 44
clades I and II contained exactly the same haplotypes but with bootstrap support values 
that differed slightly.  However, with keeni as the outgroup neither topology placed 
haplotypes LCAL01 and TCGA01 branching from the basal node in the beach clade as 
had the unweighted topology with leucopus as the outgroup.  The weighted data set 
identified a Florida peninsula clade with haplotypes SCFL01, MCFL76, WCFL07, and 
WCFL09 while the unweighted data set does not have this clade.  The weighted data set 
places NIV17 branching from a basal node among polionotus while the unweighted data 
did not. 
Bayesian analyses were carried out with cytochrome-b, D-loop and the both 
regions combined (Fig. 24 - 30).  With P. keeni as the outgroup, LRT and AIC scores 
indicated different best fit models for the data, thus, including the data set with leucopus 
and gossypinus as outgroups there are three cytochrome-b phylogenies from the Bayesian 
analysis.  The first cytochrome-b topology had P. leucopus as the outgroup and model 
GTR + ? was selected by both log likelihood and AIC scores (Fig. 24).  This tree also 
gave strong posterior probability support (95%) for inland clade I.  Like the HKY + ? 
model with P. keeni as the outgroup, the GTR + ? model also placed haplotype LCFL01 
within inland clade II and moderately supported (69%) the clade.  This topology also 
placed haplotypes TCGA01 and LCAL01 within the beach clade but at similar support 
(54%) as the GTR + I model with P. keeni as the outgroup.  With P. keeni as the 
outgroup, models GTR + I (AIC score) and HKY + ? (LRT) were selected.  The topology 
produced using the HKY + ? model was a 50% majority-rule consensus of 191 840 trees 
(Fig. 25).  The topology produced using model GTR + I was a 50% majority-rule 
consensus of 193 162 trees (Fig. 26).  Both topologies gave strong posterior probability 
  
 45
support (100%) for P. polionotus as a monophyletic clade.  Also, both topologies 
supported the existence of three major clades.  The three clades (inland clade I, inland 
clade II, beach clade) were similar to the clades produced by maximum parsimony and 
maximum likelihood cytochrome-b phylogenies.   
The two Bayesian cytochrome-b analyses were not, however, in complete 
agreement.  Each inland clade I contained the same haplotypes, however, the HKY + ? 
model had slightly higher posterior probabilities.  The HKY + ? model placed haplotype 
LCFL01 branching from the basal node in inland clade II but the GTR + I model did not 
place the LCFL01 haplotype in either clade.  Inland clade II was much more strongly 
supported within the HKY + ? model (83%) than within the GTR + I model (50%).  Both 
models indicated the existence of a beach clade that is comprised of the same haplotypes.  
However, both gave low posterior probability support for the beach clade (HKY + ?; 
56%: GTR + I; 51%).  The GTR + I model also placed haplotypes TCGA01 and LCAL01 
branching from the basal node in the beach clade while within the HKY + ? model all 
haplotypes branch from a single node.  
 Both log likelihood and AIC scores identified a single model for the D-loop.  
Using either outgroup (P. leucopus or P. keeni) both AIC and log likelihood scores 
selected HKY+ ?+ I as the best-fit model.  The D-loop Bayesian analysis with P. 
leucopus and P. gossypinus as outgroups produced a topology that is a 50% majority-rule 
of 191 313 trees (Fig. 27).  The D-loop Bayesian analysis with P. keeni as the outgroup 
resulted in a 50% majority-rule of 191 737 trees (Fig. 28).  The clades and node 
placement between the two topologies were essentially the same.  However, using P. 
leucopus as the outgroup produced slightly stronger support for inland clade I (71% Vs 
  
 46
62%) and for inland clade II (92% Vs 73%).  The posterior probability support for the 
beach clade remained essentially unchanged (65% Vs 64%).  Using P. keeni as the 
outgroup, the model KHY + ? + I supported inland clade I with the typical haplotypes 
although somewhat more resolved, but, uniquely placed inland clade I branching from the 
basal node in the polionotus group.  Both of the Bayesian D-loop topologies placed 
haplotype LCFL01 within inland clade II and both also placed haplotype SCFL01 within 
inland clade II.  The beach clade within these two topologies also contained haplotypes 
TCGA01 and LCAL01, however the beach clade was somewhat more resolved than the 
Bayesian cytochrome-b phylogenies.   
The Bayesian topology combining cytochrome-b and the D-loop into a single 
fragment with P. leucopus and P. gossypinus as outgroups resulted in a 50% majority-
rule consensus of 190 150 trees (Fig. 29).  Using keeni for the outgroup, the 50% 
majority-rule represented 194 997 trees (Fig. 30).  Combining cytochrome-b and the D-
loop regions for the Bayesian analysis produced two topologies that are essentially the 
same.  Inland clade I contained haplotypes consistent with other methods (BCAL05, 
RCAL01, TCGA03, BCAL01, MCAL01, OCFL03, WCFL04, BCAL04, JCFL01) as did 
inland clade II (SCFL01, LCFL01, OCFL04, WCFL01, WCFL02, OCFL06, WCFL02, 
OCFL05, and WCFL03).  The beach clade as hypothesized by each topology was 
structurally the same.  Posterior probability support was high and essentially the same 
between the topologies, however, the topology with P. leucopus and P. gossypinus as 
outgroups had slightly stronger support.  The only haplotype that was not in agreement 
between the topologies was NIV17.  With P. leucopus and P. gossypinus as outgroups, 
haplotype NIV17 branched form the basal node within the Florida peninsula clade.  
  
 47
However, with keeni as the outgroup, NIV17 was the only haplotype on a single branch.  
Posterior probabilities (PP) were generally higher for a given analysis than are bootstrap 
values (Huelsenbeck et al. 2002).  Thus, posterior probability support is typically only 
considered strong at approximately 95%. 
The total sequence analyses of maximum likelihood, maximum parsimony, and 
the Bayesian methods for phylogenetic analyses were largely in agreement concerning 
placement of haplotypes into clades.  Comparing each region separately and combining 
the two regions for analyses using various methods also identified strikingly similar 
haplotype grouping.  Minor differences among the analyses primarily involved the 
placement of single haplotypes.  For instance, the unweighted MP analysis with keeni as 
the outgroup placed SCFL01 branching from the basal node in inland clade II.  Haplotype 
SCFL01 was not placed in this position by any other MP analysis.  However, this 
haplotype was placed in inland clade II in some of the ML analyses but with low 
bootstrap values suggesting that support for this placement was weak.  Maximum 
parsimony analyses tended to leave haplotypes SCFL01, LCAL01, TCGA01, WCFL07, 
WCFL09, and NIV17 unresolved in a larger sense, but all of the analyses associate 
WCFL07 with WCFL08 and, separately, LCAL01 with TCGA01.  These data supported 
the presence of three and possibly four lineages.  The area of the fourth lineage (Florida 
peninsula), however, was poorly sampled because the intent of this study was to look for 
any similarities between beach forms from the Gulf Coast and Atlantic coasts of Florida. 
The total sequence analyses indicated something of an unexpected outcome with 
respect to the relationship among the three P. maniculatus subspecies and their 
relationship to P. polionotus.  Maximum likelihood analyses involving P. m. pallescens, 
  
 48
P. m. bairdii, and P. m. sonoriensis from cytochrome-b using P. leucopus as the outgroup 
and HKY + ?, TVM+ ? models of molecular evolution was unresolved.  Concerning the 
three maniculatus subspecies, changing the outgroup to P. keeni using the HKY + ?, and 
GTR + ? models again left the relationship unresolved.  However, ML analyses of the D-
loop with leucopus/gossypinus and keeni as outgroups with substitution models HKY + I 
+ ? and HKY + I, respectively, indicated that bairdii split prior to the 
pallescens/sonoriensis split.  Combining both gene regions and conducting a ML 
analyses produced two outcomes.  With P. leucopus and P. gossypinus as outgroups and 
using the HKY + I + ? model, the relationship between the three maniculatus subspecies 
was, again, unresolved.  However, using P. keeni as the outgroup with the HKY + I + ? 
model indicated that bairdii split prior to the pallescens/sonoriensis split.   
Maximum parsimony (MP) strict-consensus trees were also used to investigate the 
relationship between the three maniculatus subspecies.  Weighted and unweighted MP 
analyses of cytochrome-b using leucopus as the outgroup was unresolved.  However, the 
same analyses with keeni as the outgroup indicated that sonoriensis split prior to the 
pallescens/bairdii split.  Weighted and unweighted MP analyses of the D-loop with 
leucopus as the outgroup indicated bairdii split prior to the pallescens/sonoriensis split.  
Also, weighted and unweighted MP analyses of the D-loop with keeni as the outgroup 
suggested that bairdii split prior to a pallescens/sonoriensis split.  However, combining 
the two gene regions in a weighted and unweighted analysis with leucopus as the 
outgroup leaves the relationship between the three unresolved.  Weighted and unweighted 
MP analyses of the combined regions with P. keeni as the outgroup were also unresolved.   
  
 49
Finally, Bayesian 50% majority-rule consensus trees were used to investigate the 
relationship between the three maniculatus subspecies.  Analysis of cytochrome-b using 
model GTR + ? with leucopus as the outgroup shows that pallescens split prior to the 
bairdii/sonoriensis split.  However, Bayesian analysis of cytochrome-b using model 
HKY + ? and GTR + I with keeni as the outgroup was unresolved.  Analyses of the D-
loop using the model HKY + ? + I with leucopus/gossypinus then keeni as the outgroup 
indicated a bairdii split prior to a pallescens/sonoriensis split.  Combining the two gene 
regions within a Bayesian analysis using split models (cytochrome-b, HKY + ?; D-loop, 
HKY + ? + I) with leucopus then keeni as outgroups and a 50% majority-rule consensus 
tree support the conclusion that bairdii split prior to a pallescens/sonoriensis split. 
Nested Clade Analysis 
Nested clade analysis were based on the parsimony network developed using an 
algorithm developed by Templeton et al. (1992).  Prior to the nested clade analysis, fifty-
five sequences consisting of 2449 nucleotides were analyzed in DNASP v4.0.  This 
analysis identified 35 unique haplotypes.  Theta (?) estimated to be 0.01116, was then 
used to solve H in equation 1 of Templeton et al. (1992) (H = 0.0290).  At the 5% level, 
this value indicated that the use of parsimony was justified for these data.  For these data 
and assumptions, using a 95% statistical parsimony analysis, parsimony was excepted for 
up to 21 mutational steps (P
21
 = 0.9536).  However, the TCS generated parsimony 
cladogram contains ambiguities or two unresolved loops (Fig. 31).  One single loop 
involves what is primarily a Florida peninsula clade.  However, two haplotypes, one from 
central Alabama (LCAL01) and one from central Georgia (TCGA01), were placed within 
  
 50
this clade by the analysis.  The other loop involved placement of a single Gulf Coast 
haplotype.   
To resolve these ambiguities, three criterion based on predictions made by 
coalescence theory were used to discern more likely alternative solutions (Pfenninger & 
Posada 2002).  Based primarily on geographic considerations, alternative parsimony 
connections labeled 1A, 1B, 1C, and 2 (Fig. 31) were broken.  Loop 1A connects the 
Florida peninsula clade nearer to the basal haplotypes of the network.  Loops 1B and 1C 
connect the disputed haplotypes nearer the tips of the network.  Connections 1B and 1C 
were broken based on criteria 2 and 3 (Pfenninger & Posada 2002).  The clade formed by 
haplotypes (MCFL76, MCFL77, WCFL08, WCFL09, WCFL10, NIV36, NIV17, 
LCAL01, LCAL01, and TCGA01) is unlikely to have evolved from within the Gulf 
Coastal beach clade.  Leaving the clade connected to branch 1A as opposed to branch 1B 
or 1C seemed the most logical choice as it was more likely to have originated interior in 
the network rather than the interior of the beach clade.  Also, given the amount of 
substitutions within the peninsular clade it was more likely to be an older clade.  Thus, 
connections 1B and 1C that place a highly divergent clade within a less divergent clade 
from a different region seemed unlikely.  The second loop (Connection 2) was broken 
based on criterion 3.  The haplotype PEN02* was unlikely to have evolved within a clade 
from a different region.  Therefore, connection 2 was broken on the assumption that this 
haplotype was most likely to have evolved within the clade from the same geographic 
region. 
To be consistent with other analyses in this project, NCA was conducted both 
with and without haplotypes from the Florida peninsula.  Also, with Florida peninsula 
  
 51
haplotypes removed, the analysis was conducted with haplotypes LCAL01 and TCGA01 
placed in the network as suggested by statistical parsimony and branching manually 
placed from the basal node of the beach group as suggested by ML, MP, and Bayesian 
analyses.  Thus, there were three separate NCA analyses conducted.  One with haplotype 
placement as estimated by TCS (Model I), one with Florida peninsula haplotypes removed 
and again estimated by TCS (Model II), and one without Florida peninsula haplotypes and 
inland haplotypes LCAL01 and TCGA01 manually connected to a basal node of the 
beach clade (Model III).  Within these three hypothetical topologies there were no 
significant associations between haplotypes and geographic distribution until the 2
nd
 level 
(Table 3). 
There was substantial agreement among the three models concerning several 
clades that showed a significant association of haplotypes within the geologic landscape.  
Clades that were the same between models may have different clade numbers; therefore, 
superscribing was used to indicate identical clades between models.  Individually, model 
I indicated that clades 2-2
1
, 3-3
2
, 4-1
3
, 4-4
4
, 4-5
5
, 4-6
6
, 5-2, 5-3, and the total clade were 
significant.  Biological explanations (Templeton 1998) include restricted gene flow with 
isolation by distance (2-2
1
, 5-2), past fragmentation and/or long distance colonization (4-
1
3
), contiguous range expansion (4-5
5
), allopatric fragmentation (4-6
6
), restricted gene 
flow with dispersal and with some long distance dispersal (total clade).  Clades 3-3
2
 and 
4-4
4
 had inconclusive outcomes while clade 5-3 had no significant clade distances. 
Model II, with Florida peninsula haplotypes removed and the LCAL01, LCAL02 
and TCGA01 haplotypes with the inland group (per statistical parsimony), indicated that 
clades 2-2
1
, 3-2
2
, 4-1
3
, 4-3
4
, 4-4
5
, 4-5
6
, 5-1, 5-2, and the total clade were significant.  The 
  
 52
chain of inference indicated restricted gene flow with isolation by distance (2-2
1
), past 
fragmentation and/or long distance colonization (4-1
3
), contiguous range expansion (4-
3
4
), and allopatric fragmentation (4-5
6
).  Clades 3-2
2
 and 4-3
4
 had inconclusive outcomes 
and clades 5-1, 5-2, and the total clade had no significant clade distances.  Model III, with 
the Florida peninsula haplotypes removed and LCAL01, LCAL02, and TCGA01 
manipulated to branch from a basal node of the beach clade indicated that clades 2-2
1
, 3-
2
2
, 4-1
3
, 4-2
4
, 4-4
5
, 4-5
6
, 5-2, and the total clade were significant.  The chain of inference 
identified a patter consistent with restricted gene flow with isolation by distance (2-2
1
), 
past fragmentation and/or long distance colonization (4-1
3
), contiguous range expansion 
(4-4
5
), and allopatric isolation (4-5
6
).  Clades 3-2
2
, 4-2
4
 had no conclusive outcome while 
clades 5-2 and the total clade had no significant clade distances.   
Neutral evolution and demographic analysis 
Overall, the phylogenetic analyses suggested the presence of at least three major 
clades within P. polionotus.  Measures of genetic diversity within and between clades 
include: the number of polymorphic sites (S), the total number of mutations (Eta), 
haplotype diversity (Hd), nucleotide diversity (?), average number of nucleotide 
differences (k), number of haplotypes (h), and theta (?).  As expected, haplotype and 
nucleotide diversity decreased within groups when multiple samples were identical.  The 
effect was most noticeable when sample size is small.  The average number of nucleotide 
differences is much less influenced by the addition of identical samples and, within these 
data, may be a more reliable indicator of sequence variation among samples or groups.  
Table 2 gives estimates of diversity of both the data set containing P. leucopus and P. 
gossypinus as outgroups and the data set with P. keeni as the outgroup.  An obvious 
  
 53
decline in diversity measures occurred when P. leucopus and P. gossypinus were 
removed.  The group containing P. keeni and the three maniculatus subspecies is 
consistent with variation likely expected among closely related species.  Both groups 
indicate more variation within the D-loop portion of the sequence.  This was expected 
due to regions of the D-loop that are hypervariable. 
Table 4 gives measures of diversity for clades derived from maximum likelihood 
phylogenetic analyses.  Tables 5 - 7 show measures of diversity for 5
th
 level clades for 
NCS models 1 - 3 respectively.  The statistical parsimony network generated for the 
nested clade analysis (NCA) was similar to the phylogenetic analyses but did uniquely 
place some haplotypes.  Haplotype diversity (Hd) and nucleotide diversity (?) were 
variable among the all clades and gene regions.  Measures of diversity for cytochrome-b 
and the D-loop separately and combined produced generally predictable results.  The D-
loop region when compared with cytochrome-b gene region tended to be more diverse 
among all groups.  Interestingly, a comparison of the beach clade and the two inland 
clades revealed that the D-loop ? in the beach clade was significantly higher than either 
inland clade, however, inland clade I has a higher ? within cytochrome-b than the beach 
clade.  Nucleotide diversity within the D-loop of inland clade I was only slightly higher 
than ? within cytochrome-b while in all other clades/groups it was approximately twice 
as high.  For the complete fragment, the average number of pairwise differences was 
twice as high in the beach clade (12.28) as in inland clade II (6.47) and almost 1.5 times 
as high as in inland clade I (8.73).  However, for cytochrome-b, the average number of 
pairwise differences for the beach clade was 3.70, inland clade I 3.83, and for inland 
clade II 1.03.  Estimates of diversity for the beach clade were also calculated without the 
  
 54
haplotypes LCAL01, LCAL02, and TCGA01.  These estimates for the total sequence, 
cytochrome-b, and the D-loop are 10.74, 2.98, and 7.16 respectively.  Measures of 
diversity (Table 4) were also calculated for the entire polionotus group, the polionotus 
group (complete) without the Florida peninsula haplotypes (complete*).   
The models analyzed using NCA resulted in seven 5
th
 level clades.  Measures of 
diversity were estimated for each clade (Tables 5 - 7).  Not unexpectedly, the greatest 
amount of nucleotide diversity for cytochrome-b, D-loop, and the complete fragment was 
found within clade 5-2 of model one.  This clade includes the haplotypes from the Florida 
peninsula.  Florida peninsula haplotypes were removed in the analyses of models two and 
three.  The haplotypes LCAL01, LCAL02, and TCGA01 are placed within the clade 
containing inland forms by statistical parsimony analysis resulting in model II.  To be 
consistent with the phylogenetic analyses these haplotypes were placed on a branch 
extending from a basal node of the beach clade.  In model II, measures of diversity are 
approximately 1.5 times higher in clade 5-1, which contains inland haplotypes along with 
LCAL01, LCAL02, and TCGA01.  Model three was similar to model two, however, the 
haplotypes LCAL01, LCAL02, and TCGA01 are within clade 5-2 containing beach 
haplotypes.  In model three clade 5-1 measures of diversity are higher, however, this 
model tends to decrease the difference. 
Neutrality tests of cytochrome-b and the D-loop were conducted separately and as 
a complete sequence.  Interpretations of the neutrality tests are influenced by the 
measuring statistic and by DNA region.  If a gene under selection is linked to a neutral 
site the value of D could be affected (Tajima 1989).  Neutrality statistics were calculated 
using DNASP v4.0 (Rozas et al. 2003) and were calculated using the number of 
  
 55
segregating sites.  Clades for demographic analyses were chosen based on the outcome of 
the ML, Bayesian, and MP phylogenetic analyses along with the nested clade analysis.  
These clades were analyzed using various neutrality tests and are comprised of the 
following haplotypes, inland clade I, BCAL05, RCAL01, TCGA03, BCAL04, JCFL01, 
BCAL01, MCAL01, OCFL03, and WCFL04; inland clade II, SCFL01, LCFL01, 
OCFL04, WCFL01, WCFL02, OCFL06, WCFL01, OCFL05, WCFL03; beach clade, 
LCAL01, TCGA01, PEN05, AMM01, LEU08, OCFL01, WCFL01, WCFL06, OCFL07, 
JCFL03, CCFL01, TRI07, PEN02*, ALL02**, ALL05, and ALL02*.  Because 
haplotypes LCAL01 and TCGA01 were found in central Alabama and central Georgia 
respectively, analyses of the beach clade was conducted both with and without these 
haplotypes (clade beach and clade beach* respectively).  As expected, outcomes of 
neutrality analyses varied by gene region and statistic.  The Cytochrome-b region tended 
to produce more negative values with the d-loop region being less negative. 
Fu & Li?s D and Fu & Li?s F tests were conducted twice, once using P. 
maniculatus as an outgroup (Table 8) and once using polionotus haplotypes from other 
clades as outgroups (Table 9).  With maniculatus as the outgroup, none of the tests 
produced significant results.  However, again, all values were negative with one 
exception, Fu & Li?s D test for the D-loop region of inland clade II.  Fu & Li?s D and F 
tests on polionotus groups with polionotus haplotypes as the outgroup were also not 
significant, however, testing the beach clade with inland clade I as the outgroup did 
approach significance.  McDonald-Kreitman neutrality tests were conducted for 
cytochrome-b region only (Table 8).  These tests were conducted using maniculatus for 
between species ratio estimates and clades or groups of polionotus as populations for 
  
 56
nonsynonymous to synonymous ratio estimate.  The results of Fu & Li?s D* and Fu & 
Li?s F* tests of polionotus clades were, like the prior analysis, negative values (Table 10).  
While the tests were not significant, the tests using all available polionotus haplotypes 
and all regions (minus Fu & Li?s D* for the D-loop region) produced near significant 
values (0.1 > P > 0.05).  Without the four haplotypes from the Florida peninsula clade Fu 
& Li?s F* and D* still approached significance when testing cytochrome-b (0.10 > P > 
0.05).  Tajima?s D was not found to be significant in any analyses (Table 10).  Use of the 
complete P. polionotus data set produced the most negative value of Tajima?s D.  
However, Tajima?s D statistic for clade beach produced the most negative values relative 
to all other clades minus the complete polionotus analysis.  There are multiple ways to 
interpret neutrality statistics.  Nevertheless, along with other evidence, the neutrality tests 
here suggest that polionotus, or clades of polionotus, have undergone demographic 
expansion.   
Six clades or groups of haplotypes were analyzed using program FLUCTUATE 
(Table 11).  All available polionotus haplotypes were analyzed (complete).  Florida 
peninsula haplotypes were removed and analyzed (complete*).  Inland clades I and II 
were analyzed along with clades beach and beach* (haplotypes LCA01 and TCGA01 
removed).  Each clade or group that contained the Florida peninsula haplotypes was 
analyzed both with and without these haplotypes.  Also, the beach clade was analyzed 
with and without haplotypes LCAL01 and TCGA01.  An asterisk (*) was used to indicate 
groups or clades where haplotypes have been removed.  The results of the analyses were 
not significantly changed by removal of the haplotypes.  To determine N
e
, a substitution 
rate of 1.1 X 10
-8
 substitutions per site per year was calculated from the combined data.  
  
 57
The analysis of the complete or complete* set of P. polionotus haplotypes indicated that 
the N
e
 estimate without growth (2 657 062) was significantly different from the N
e
 of the 
growth model (5 120 272).  Effective population size predictions for the beach (constant 
768 109, growth 2 674 778) and beach* clades (constant 549 775, growth 1 729 500) 
were also significantly different.  Other clades analyzed using FLUCTUATE included 
inland clade I (constant 326 013; growth 505 056) and inland clade II (constant 164 478; 
growth 158 444), clade 5-1 (constant 830 388; growth 1 176 722) and 5-1* (constant 527 
952; growth 724 000).  These clades were not significantly different in their constant or 
growth estimates of N
e
 (values).  Inland clade II was the only clade with an estimated 
negative growth rate (values). 
Molecular clock 
Comparisons of percent divergence of the 1137 bp cytochrome-b gene between P. 
p. ammobates (AMM01) and P. m. sonoriensis, P. m. bairdii, and P. m. pallescens 
revealed 52 differences (4.57%), 49 differences (4.31%), and 45 differences (3.96%) 
respectively.  Peromyscus polionotus ammobates was also compared to P. leucopus (111 
differences; 9.76%) and P. gossypinus  (113 differences; 9.94%).  Estimates of percent 
divergence between the subspecies of maniculatus were, bairdii and pallescens with 13 
differences for 1.14%, pallescens and sonoriensis with 18 differences for 1.58%, and 
bairdii and sonoriensis with 17 differences for 1.50%.  A comparison between P. 
leucopus and P. gossypinus produced 57 differences for 5.01% divergence estimate.  
Comparison between P. m. bairdii and P. leucopus and P. m. sonoriensis and P. 
gossypinus revealed 102 differences for 8.97% divergence and 108 differences for 9.50% 
divergence respectively.  Comparisons between P. keeni and the following, P. p. 
  
 58
ammobates, P. m. sonoriensis, and P. leucopus revealed 56 differences for 4.93% 
divergence, 54 differences for 4.75% divergence and 102 differences for 8.97% 
divergence respectively.  Finally, divergence within P. polionotus was estimated using a 
haplotype from the beach clade (P. p. ammobates, AMM01) and a haplotype from each 
of the hypothesized clades, (inland clade I, inland clade II, and the Florida peninsula), as 
well as, a haplotype from the beach clade.  Haplotypes used for comparisons were 
AMM01 to, NIV17, JCFL01, WCFL02, and PEN05 producing estimates of 5 differences 
for 0.44%, 8 differences for 0.70%, 5 differences for 0.44%, and 2 differences for 0.18% 
respectively.   
Likelihood ratio tests (Felsenstein 1981) were conducted to determine if 
substitution rates were consistent among lineages within different phylogenies.  
Likelihood ratio tests (LRT) tests (2??) were conducted for all four nucleotide 
substitution models used in the analysis of cytochrome-b, the two models used to conduct 
the D-loop analyses, and the two models used to conduct the analysis with the complete 
sequence.  For cytochrome-b with P. keeni as the outgroup under the HKY + ? model, 
log likelihood values with and without a clock assumption were 2427.31 and 2442.32 
respectively.  Thus, the LRT test [2(2442.32 - 2427.31) = 30.02, df = 40, and P = 0.8620] 
indicated that the two models are not significantly different.  Cytochrome-b data with 
keeni as the outgroup under the GTR + I model with and without the clock produced log 
likelihood values of 2420.62 and 2435.66 respectively.  Here 2(2435.66 - 2420.62) = 
30.08, df = 40, and P = 0.8729 indicating the two models are not significantly different.  
Cytochrome-b with P. leucopus as the outgroup under the TVM + ? model with and 
without a clock assumption produced log likelihood values of 2985.62 and 3001.73 
  
 59
respectively, then, 2(3001.73 - 2985.62) = 32.22, df = 42, and P = 0.8620.  Cytochrome-b 
with leucopus as the outgroup under the HKY + ? model with and without the clock 
assumption produced log likelihood values of 2994.17 and 3010.22 respectively, then 
2(3010.22 - 2994.17) = 32.10, df = 42, and P = 0.8653.   
The D-loop data set with leucopus as the outgroup under the HKY + I + ? model 
with and without the assumption of a clock produced log likelihood values of 3063.81 
and 3091.98 respectively, then, 2(3091.98 - 3063.81) = 56.35, df = 43, and P = 0.0833.  
The D-loop with keeni as the outgroup under the HKY + I model with and without the 
clock assumption produced log likelihood values of 2455.72 and 2478.70 respectively, 
then, 2(2478.70 - 2455.72) = 45.96, df = 41, and P = 0.2742.  Combining cytochrome-b 
and the D-loop regions and using the HKY+ I + ? model with leucopus as the outgroup 
produced log likelihood values of 6822.79 and 6845.90 respectively, then, 2(6845.90 - 
6822.79) = 46.21, df = 42, and P = 0.3025.  The combined data set with keeni as the 
outgroup with the HKY + I + ? model with and without the clock assumption produced 
log likelihood values of 5555.24 and 5572.78 respectively, then, 2(5572.78 - 5555.24) = 
35.08, df = 40, and P = 0.6911.  Based on log likelihood ratio tests at the 5% level the 
molecular clock assumption was not rejected under any scenario. 
Analysis of the complete sequence with P. keeni as the outgroup produced an 
estimated substitution rate of 1.8 X 10
-8
 substitutions per site per year.  This rate was 
obtained by adding branch lengths from the maniculatus/polionotus split (t1; Fig. 32) to 
each tip then dividing by the estimated age of the sinkhole.  Estimated substitution rates 
ranged from 1.3 X 10
-8
 to 2.4 X 10
-8
.  However, the two predominate estimates were 1.4 
X 10
-8
 and 2.2 X 10
-8
.  Thus, the two most frequent estimates were averaged obtaining 
  
 60
1.8 X 10
-8
.  This rate was then used to convert other distances into time (Fig. 32).  Times 
to nodes were estimated by calculating the distance from a node to the branch tip for each 
branch that extended from that node.  These distance values were converted into time 
estimates to allow calculation of summary statistics.  For example, the node involving the 
divergence of extant forms of P. polionotus (t1; Fig. 32) was estimated at 306 000 ? 48 
000 YBP.  For comparison, time to nodes involving the three maniculatus subspecies 
were also estimated under the assumption of the same substitution rate.  The split of 
bairdii from the other two maniculatus subspecies was estimated at 430 000 ? 144 534 
YBP.  The constant-rate model tends to overestimate small distances (Yang 1996).  Thus, 
when possible, conservative estimates of parameters were used in an attempt to minimize 
overestimating times.  Using the equation 5.12 from Nei (1987) r = K/(2T), a substitution 
rate of 1.1 X 10
-8
 per site per year was obtained.  This estimate was similar to the 
previous estimate but was not used in favor of the more conservative estimated 
substitution rate of 1.8 X 10
-8
. 
Discussion 
Divergence of polionotus/maniculatus  
The multiple phylogenetic analyses used here indicate that P. polionotus forms a 
reciprocally monophyletic clade with respect to P. maniculatus.  These data also indicate 
a much older maniculatus/polionotus split than was anticipated.  While Blair (1950) and 
Bowen (1968) offered opposing theories concerning the ancestral lineage of P. 
polionotus, both inferred a relatively recent divergence (< 500 000 YBP) of polionotus 
from a maniculatus stock.  However, based on the results produced in this study it is 
unlikely that the ancestral lineage of polionotus is one of the three maniculatus 
  
 61
subspecies included in this analysis (bairdii, pallescens, sonoriensis).  Instead, the most 
likely scenario is a maniculatus/polionotus split that preceded the divergence of the 
maniculatus subspecies sonoriensis, pallescens, and bairdii.  Avise et al. (1983) analyzed 
mtDNA from 82 maniculatus and polionotus individuals collected across the United 
States portion of their range and concluded that their data best fit a model with polionotus 
as a monophyletic group with respect to maniculatus and maniculatus as a paraphyletic 
group with respect to polionotus.  Here, however, results from this study indicate both 
polionotus and maniculatus as monophyletic groups with respect to one another.  
However, there were only three subspecies of maniculatus represented within my 
analysis.  From these data, it could not be accurately determined if the ancestral 
polionotus stock entered the Southeastern US through a northern (Bowen 1968) or 
western route (Blair 1950).  The phylogenies produced from different methods do not 
agree concerning a western or northern route for pre-polionotus stock.  Avise et al. 
(1983) conducted a restriction enzyme analysis of mtDNA that demonstrated the 
polionotus divergence from maniculatus might be more difficult to elucidate than had 
previously been thought.  In fact, their data suggested polionotus might be most similar to 
a maniculatus form presently found in Southern California.  If the earlier divergence of 
maniculatus/polionotus stock is accurate, it may explain why the speciation event is 
difficult to characterize.  
I hypothesized that the ancestral lineage of P. polionotus became isolated in the 
Southeastern US approximately two million years ago.  Under this scenario, isolation of 
the ancestral form of polionotus may have occurred near the time of the 
Pliocene/Pleistocene border as temperature changes shifted vegetation zones.  Again, this 
  
 62
represents a longer period than most previous researches had hypothesized.  However, 
Avise et al. (1983) used a maniculatus form from Southern California and estimated a 
polionotus/maniculatus split at 1.5 million years.  My data support their time and even 
conclude that it was likely to have occurred earlier.  This conclusion is supported by 
recent fossil evidence of (Ruez 2001) dating polionotus to at least the late Pliocene or 
early Pleistocene.  The degree of divergence of mtDNA between maniculatus and 
polionotus indicated an early split as well.  While realizing that 2% divergence per 
million years for mammalian mitochondrial DNA (Brown et al. 1979) may be a limited 
estimate, it does seemingly fit these data.  Percent divergence between polionotus and the 
three subspecies of maniculatus used within this analysis estimate a separation date of 
approximately two MYB. 
Phylogenetic analysis 
The three phylogenetic methods utilized in my study support the presence of at 
least three major clades with extant P. polionotus.  These clades form polytomy in most 
topologies.  In the present geologic state of the region there is not an obvious 
geographical feature that physically separates the clades.  Molecular data suggest the 
three clades last shared a common ancestor approximately 300 000 YBP.  Assuming a 
polionotus/maniculatus split two million years ago as fossil and molecular data suggest, it 
is intriguing that the divergence of extant forms would date to only 300 000 YBP.  Such a 
date would suggest that extant lineages of polionotus began diverging during a pre-
Illinoisan glacial period of the Pleistocene Epoch.  Interestingly, the initial divergence 
within the beach clade (243 000 YBP) appears to have occurred prior to divergence 
within the two inland clades (Fig. 32).  However, the estimates of initial divergence 
  
 63
within the three primary clades are all within the same confidence intervals.  Interval t5 
within the beach clade and interval t6 within inland clade II are approximately the same 
age and interval t7 within inland clade I occurs approximately 63 000 years ago. 
It is always difficult to determine the precise cause(s) that lead to divergence and 
cladogenesis within a taxonomic assemblage.  Factors that have influenced between and 
within clade variation since divergence are also difficult to elucidate.  However, climatic 
oscillations have clearly played a role within P. polionotus.  If P. polionotus as a species 
has been present in the southeastern United States for the entire Pleistocene, the extant 
major lineage splits must be relatively recent.  Thus, indicating either a long period of 
relative stasis or the loss of lineages that may have diverged prior to the last 300 000 
years.  A possible explanation for this lineage pattern is the cyclical nature of climatic 
phenomenon and its associated changes on the landscape.  The pattern of lineage 
divergence observed within P. polionotus here is similar to other species whose range is 
limited to the southeastern United States and likely to have been driven by episodic 
climatic events that cyclically isolated lineages (Avise 2000).  However, these same 
events may also isolate lineages in substandard habitat eventually leading to their 
extinction.  Thus, while climatic cycles may be a principle factor isolating lineages 
leading to their divergence, levels of variation in these cycles (frequency, magnitude, 
timing) may also act to remove variant lineages thereby reducing variation and 
divergence within the species (i. e. lineage sorting).  Traditionally, effects of Pleistocene 
climatic oscillations were generally characterized at the species level.  However, 
contemporary genetics analyses have provided insight into genetic structure of species 
that had not been possible before.  Climatic factors that influence species do so through 
  
 64
their affects on genetic lineages that comprise the species.  Methods of phylogeographic 
analysis offer the ability to evaluate historical effects and their relationship to extant 
lineages.  This is an important advance because the present lineages that comprise a 
species and their distribution have both a historical and contemporary component. 
Haplotypes included in this analysis were primarily from the Florida panhandle 
region and constitute three primary lineages that form two inland clades and a beach 
clade.  The geographic range of the three clades is overlapping.  In fact, the Florida 
panhandle appears to be an area where haplotypes from the three clades integrate.  The 
Freeport, Florida site, centrally located in the panhandle, was the only location where 
haplotypes from each clade were found.  However, haplotypes from two different clades 
were found in several of the same inland locations.  Repeatedly, from inland sites, 
haplotypes from different clades were found within the same burrow.  Also, in some 
locations haplotypes collected in1988 were still present after 13 years.  There are two 
likely causes for this area of geographic concurrence, either natural dispersal and/or 
human mediated dispersal.  Over the last 200 years, cultivation of fields and installation 
of railroad tracks, roads and power lines has almost certainly influenced the dispersal 
patterns of mainland P. polionotus.  In fact, disturbed sites such as these are principle 
locations for obtaining samples and indeed account for the common name ?oldfield 
mouse? (Hall 1981).  Unfortunately, it is difficult to determine the degree to which 
dispersal of haplotypes was natural or human mediated.  Within beach subspecies, 
however, populations appear to be comprised of a single haplotype.  In fact, the following 
beach subspecies, ammobates, trissyllepsis, leucocephalus, and peninsularis are each 
comprised of a single different haplotype.  An exception to this was a possible single 
  
 65
nucleotide substitution within the ammobates population.  However, this substitution was 
present in only one individual and could easily be a sequencing artifact. 
The subspecies allophrys was comprised of at least two and possibly three 
haplotypes.  However, these haplotypes are found in populations that are now isolated 
from one another due to human development of the beaches.  The P. p. allophrys 
subspecies appears to be the only beach subspecies that posses multiple, extant 
haplotypes.  Historically, it is likely that some level of gene flow did occur between the 
two, and possibly three, allophrys populations.  The historical range of allophrys 
extended from Destin, Florida (Moreno Point) east to Shell Island, Florida and possibly 
as far east as Port St. Joe, Florida (Fig. 33).  This length of beach was likely occupied by 
polionotus (allophrys) populations that were semi-isolated by drainages.  The third 
haplotype placed within the P. p. allophrys subspecies group by multiple phylogenetic 
methods is from an extinct population (Crooked Island, FL).  Crooked Island, Florida 
constitutes a portion of the described range of P. p. peninsularis.  Bowen, (1968) 
delineated the subspecies distribution within the Gulf Coastal region based on breeding 
experiments and pelage reflectance values.  His designation of the Crooked Island 
population, now a section of Tyndall Air Force Base, Florida, as peninsularis instead of 
allophrys now appears to have been incorrect.  Regardless, during the mid 1990?s P. p. 
peninsularis was determined to be extirpated from the Crooked Island, Florida portion of 
its range.  The United States Fish and Wildlife Service (USFWS) determined 
translocation of individuals from the remaining peninsularis population on Cape San 
Blas, Florida to Crooked Island, Florida was an appropriate action.  Thus, two 
translocations were conducted, and, within a few years, the Crooked Island population 
  
 66
multiplied and dispersed across the length of the area.  The translocation effort appeared 
to be a success.  However, prior to extirpation, tissue from individuals of the original 
Crooked Island population had been obtained.  Therefore, that population was 
represented in these analyses.  Unexpectedly, each of the three methods of phylogenetic 
analyses placed the original Crooked Island haplotypes within the allophrys group rather 
than with the peninsularis haplotype (Fig. 11, 23, 30).  In fact, my analyses indicated that 
the Crooked Island haplotype shares a more recent ancestor with the Topsail Hill 
allophrys haplotype than the Topsail Hill haplotype shares with the Shell Island allophrys 
haplotype.  This example highlights the necessity of genetic analyses prior to 
translocation of individuals between populations. 
A possible fourth clade comprised of Florida peninsula haplotypes along with two 
haplotypes from near Crawfordsville, Florida (south of Tallahassee, Florida) does not 
hold up in the majority of Bayesian, ML and MP analyses but was present in the 
statistical parsimony network.  The Florida peninsula was not intended to be an area 
extensively represented within this study.  It was intended to provide Atlantic Coastal 
haplotypes for comparison to Gulf Coastal haplotypes, and, possibly, an outgroup for 
Gulf Coastal haplotypes.  The samples from the Florida peninsula are quite diverged 
relative to the divergence within polionotus.  Selander et al. (1971) did find the Florida 
peninsula to posses the greatest amount of protein polymorphism.  They also showed that 
insular populations of the Atlantic coast possessed the same amount of protein 
polymorphism as inland populations of the Florida panhandle and more than inland 
populations of Georgia and South Carolina.  This is compatible with the findings here 
that show the P. p. niveiventris, a beach subspecies from the Cape Canaveral Air Force 
  
 67
Station, with at least two haplotypes within the population.  Peromyscus polionotus 
niveiventris represents a large population that may be relatively undisturbed by human 
activities. 
Nested clade analysis 
The statistical parsimony network developed through NCA uniquely placed some 
haplotypes.  Principally, however, it supported the phylogenies developed through 
Bayesian, ML, and MP methods.  However, the results of NCA are based on a single 
network that was only one of multiple possibilities.  The application of NCA to my data 
did allow the testing of hypotheses relative to geographic placement of haplotypes to help 
clarify the evolutionary history of P. polionotus.  There were significant geographic 
associations of haplotypes within each model (Table 3).  Some significant associations 
produced inconclusive outcomes and some P-values were significant but had no 
significant clade distances.  Within all three models the first significant pattern of 
haplotype distribution is restricted gene flow with isolation by distance (clade 2-2
1
).  
Haplotypes within this clade are found in the Florida panhandle, central and southern 
Alabama.  Clade 2-2
1
 was composed of haplotypes from inland clade I (MCAL01, 
BCAL01, BCAL03, OCFL03, WCFL04, WCFL05, JCFL01, JCFL02, and BCAL04).  
The second significant association common to the three models (clade 4-1
3
) was the 
inference of past fragmentation and/or long distance colonization involving the Florida 
panhandle and a single haplotype from the northern portion of the Florida peninsula.  
Another significant association (model I, clade 4-5
5
; model II and III clade 4-4
5
) infers 
contiguous range expansion as the biological explanation for haplotype placement.  Clade 
4-5
5
 from model I and the corresponding clade from models II and III, 4-4
5
 contain three 
  
 68
of the five Gulf Coast beach mice subspecies (peninsularis, allophrys, leucocephalus) 
and a single haplotype found in the central Florida panhandle.  The final significant 
association common to the three models (model I clade 4-6
6
; model II and III clade 4-5
6
) 
implies allopatric fragmentation as the biological explanation and involves two of the 
beach subspecies and a group of haplotypes found within the Florida panhandle.  Model I 
has two other significant patterns that produce biological scenarios for their causes.  First, 
restricted gene flow with isolation by distance involving the Florida peninsula, central 
Alabama, and central Georgia (clade 5-2).  The placement of haplotypes LCAL01, 
LCAL02, and TCGA01 within the Florida peninsula clade was unique to the statistical 
parsimony network.  Model I was also significant at the total clade level.  The biological 
scenario inferred for the geographical association of haplotypes is restricted gene flow 
with dispersal with some long distance dispersal.  Model II and III do not have significant 
clade distance at the total clade level. 
With respect to inland clade II (clade 4-1
3
), application of Templeton?s criterion  
was most consistent with the hypothesis that dispersal has been into the Florida 
panhandle from the Florida peninsula.  The dispersal direction for inland clade I is more 
difficult to determine.  Branching patterns indicate two possible dispersal patterns of 
haplotypes from inland clade I.  Dispersal was either from central Georgia or central 
Alabama into the Florida panhandle or within the south Alabama Florida panhandle 
region into central Alabama and central Georgia.  The beach clade (model I, clade 5-3) 
supports dispersal was from central Alabama and central Georgia into the Florida 
panhandle.  Although, removing haplotypes LCAL01, LCAL02 and TCGA01 from the 
beach clade forms a dispersal pattern that more closely reflects a beach clade that 
  
 69
originated in the present range of P. p. allophrys.  However, P. p. allophrys appears to be 
older than the dune system it currently occupies, thus, a model suggesting the beach clade 
originated from that location is inconsistent.  This demonstrates the difficulty of 
determining dispersal patterns when suitable habitat may be ephemeral due to climatic 
oscillations that cause haplotypes to make geographical shifts. 
Neutral evolution and demographic analyses 
Analysis of genetic variation within and between the three hypothesized clades of 
P. polionotus revealed the highest level of mtDNA variation as being within the beach 
clade.  However, variation of cytochrome-b was approximately the same or slightly 
higher within inland clade I.  Selander et al. (1971) found insular or beach populations to 
be monomorphic for most of the proteins they analyzed while inland populations tended 
to be heterozygous.  However, they did find the situation reversed for some loci.  Using 
mtDNA, the pattern of variation was essentially the same with beach populations being 
monomorphic and inland populations being heterozygous.  However, if beach 
populations form a unique clade, variation within the clade is greater or at least equal to 
inland populations of the Florida panhandle region.  Under this scenario the collective 
beach clade has the most variation but, viewed as single populations, the beach 
populations individually contain the least amount of variation.  Each beach population 
appears to contain a single haplotype but when all the beach populations clade together, 
along with some inland haplotypes, it actually forms a diverse, growing population 
segment that may be the most dynamic segment of polionotus within the Florida 
panhandle region.  In fact, if the entire population of the Florida panhandle region is 
growing as these data suggest, it appears to be predominately the beach clade driving the 
  
 70
expansion.  Historical demographic analyses of each clade indicate the beach clade has 
been growing while the two inland clades appear to have been stationary over time.  The 
FLUCTUATE growth rate (g) and theta (?) values tend to be biased upward (Kuhner et al. 
1998).  However, given this, the estimates of effective population size are seemingly out 
of proportion with the amount of available habitat and the manner in which beach mice 
utilize available habitat.  Even if the beach clade has not been growing the estimated 
effective population size is very large (Table 11).  In fact, it is unlikely that the amount of 
historically available beach habitat could support such numbers. 
Estimated divergence times 
The mtDNA data consistently support the conclusion that three extant clades 
diverged approximately 300 000 YBP.  Within clade divergence, however, does not 
appear to have occurred for approximately another 50 000 years, an effect possibly due to 
lineage sorting.  Branching patterns of the topologies suggest that approximately 300 000 
YBP, following an initial dispersal event, the three clades diverged in allopatry possibly 
occupying three refugia in southeastern North America.  Dispersal might have been into 
the Florida peninsula from the panhandle or more interior regions of Alabama or Georgia.  
Contrarily, however, dispersal possibly occurred from the Florida panhandle into the 
interior and along the Gulf Coast.  However dispersal occurred, it appears that extant 
genetic variation within P. polionotus arose from a single lineage over the last 300 000 
years.  The data are, however, inconclusive with respect to the location of a possible 
refugia.  According to Delcourt & Delcourt (1981), at the last glacial maximum (~18 000 
YBP) the Florida peninsula was primarily sand dune, scrub habitat.  This habitat type 
could have served as refugia for P. polionotus during an earlier glacial maximum 
  
 71
approximately 300 000 YBP (Petit et al. 1999).  Such a hypothesis is consistent with the 
findings of Selander et al. (1971) who reported protein polymorphism was highest in the 
Florida peninsula.  Cain (1944) listed thirteen criteria that could be used in conjunction to 
indicate a center of origin.  One criterion was the greatest amount of differentiation (i. e. 
number of species), however, taken here as the greatest amount of genetic variation, 
which, within P. polionotus is found in the Florida peninsula. 
Comparisons among a priori hypotheses 
Peromyscus polionotus was generally believed to be a young species having 
evolved after being isolated in the southeastern United States during a mid to late 
Pleistocene glacial cycle.  However, the data presented here along with data from Avise 
et al. (1983) and recent fossil evidence (Ruez 2001) supports a conclusion that P. 
polionotus is far older and has likely been isolated in the southeastern United States since 
the late Pliocene or early Pleistocene.  Correlated with this contention is the difficulty of 
determining which form of maniculatus that polionotus might have evolved from.  
Regardless of the maniculatus/polionotus split, the divergence among extent clades of 
polionotus probably occurred around 300 000 YBP.  This is a relatively recent split in 
comparison to the estimated maniculatus/polionotus split and may be due to glacial 
cycles and how they effect a small mammal within a geographically limited region and 
that is restricted to open, sandy habitat.  Organisms in the southeastern United States, 
limited in southward movement by the Gulf of Mexico, may have experienced a 
compression or packing effect controlled by the duration, rate and period of glacial 
cycles.  In other words, organisms might have been isolated into single populations from 
which a solitary lineage emerged as the source of present genetic variation. 
  
 72
Bowen?s model (1968) (the recurrent invasion hypothesis), contends that extant 
beach forms are the result of recent colonization events (< 10 000 YBP).  As the water 
level of the Gulf of Mexico began to stabilize, coastal dunes and barrier islands were 
created.  According to Bowen (1968), inland forms colonized these dune systems and 
then differentiated due to distinct selection pressures associated with the beach habitat.  
Bowen postulated both recent and repeated invasion of the dune systems by inland forms 
and contented that gene flow occurred in both a north/south and east/west pattern.  
However, the data presented here were not consistent with Bowen?s model of gene flow 
involving the Gulf Coastal region of Florida and Alabama.  In fact, my data strongly 
support a conclusion that beach subspecies are not the result of recent and repeated 
colonization events of the beach by inland forms.  My analyses indicate that present day 
polionotus populating shorelines and barrier islands of the Gulf Coast form a 
monophyletic clade that is significantly older that the dunes they now inhabit.  Thus, 
beach forms must have existed prior to the formation of current dune systems and 
shorelines.  Based on the new findings, I propose an alternative hypothesis to explain the 
origin of the Gulf coastal forms.  I refer to this hypothesis as the Shoreline-Tracking 
Hypothesis.  Under this hypothesis, beach forms, already diverged from inland forms, 
were able to track the shoreline as the water of the Gulf of Mexico rose due to melting 
glaciers.  This hypothesis explains why haplotypes are older than their habitat and infers 
that forms evolving under the selection pressures of a beach habitat were in existence 
prior to the formation of the present coastal strand and barrier islands. 
A few inland haplotypes were placed in the beach clade, which suggests that gene 
flow, while limited, occurred.  However, Bowen (1969) and Sumner (1929) disagreed 
  
 73
over the direction of gene flow.  Placement of inland haplotypes within the beach clade 
(Fig. 21, Fig. 29), by some analyses, is consistent with the route of gene flow as Sumner 
hypothesized.  However, while it does appear that gene flow has occurred between inland 
and beach clades it was probably not a recent event (< 10 000 YBP).  On a larger 
geographic scale, nested clade analysis indicated that gene flow among major clades has 
likely been both into and out of the Florida panhandle.  Given the limitations of my 
geographic collection sites, I was unable to further explore gene flow on a range wide 
geographic scale. 
Conclusions 
The application of multiple types of analyses that are designed to aid our 
understanding of evolutionary events revealed unexpected results for P. polionotus.  In 
doing so, they provided a greater understanding of how the effects of Pleistocene glacial 
cycles have effected the evolution of a small, regionally isolated mammal that is 
restricted to a specific habitat type.  Applying methods of phylogenetic analyses, nested 
clade analysis, neutrality tests and methods designed to estimate divergence times among 
diverse forms revealed these new insights.  Using these analyses, I have shown that the 
evolutionary history of P. polionotus has in fact been quite complex.  These data indicate 
P. polionotus as a monophyletic group that has probably been evolving in the 
southeastern United States for the entire Pleistocene.  Also, there appear to be three major 
clades represented in the Gulf Coastal region of Florida and Alabama.  Genetic data 
indicated that the clades diverged prior to the last glacial maximum (~18 000 YBP) and 
possibly as far back as the prior glacial maximum (~120 000 YBP).  Presently the clades 
are not geographically isolated from one another and in fact do appear to integrate.  
  
 74
However isolated coastal populations appear to consist of a single mitochondrial 
haplotype, however, these populations form a clade that posses the greatest amount of 
genetic variation based on these data.  The three clades established here consist of two 
inland clades and a beach clade.  While more than one geographic model was possible, 
my results were most consistent the idea that inland clade I and the beach clade 
originated within the central Alabama/Georgia region while inland clade II originated 
within the Florida peninsula.  Very importantly, beach forms that comprise the beach 
clade are clearly far older than had previously been hypothesized.  This is significant 
because it refutes Bowen?s (1968) hypothesis that extant beach forms recently (< 10 000 
YBP) diverged from inland colonizers.  Insular populations of P. polionotus comprise a 
unique and complex clade that forms an important segment of the regional population.  In 
fact, extant beach forms represent lineages that have been evolving in a far different 
pattern than previously described.  I hypothesize that beach lineages are older than the 
present dune system and formerly inhabited an area that is currently covered by the Gulf 
of Mexico.  These conclusions and insights were gained through complementary genetic 
analyses that illustrate how genetic data can be used to disentangle past and present 
population processes to gain some understanding of current genetic structure within a 
population. 
 75
Fig. 1  Bowen?s (1968) proposed evolutionary sequence of Gulf Coastal and Atlantic 
forms of Peromyscus polionotus.  Broken diagonal lines indicate major barriers that 
likely limited gene flow.  Bowen?s model hypothesized recent, multiple invasions of the 
beach by inland forms and gene flow between beach populations. 
76
 77
Fig. 2  Tissue samples of Peromyscus polionotus for these analyses were collected in 
Florida, Georgia and Alabama.  Inland individuals are identified by the county and state 
they were obtained.  The first three letters of their subspecific designation identify Beach 
subspecies. 
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
###
#
#
#
#
NIV17,
 NIV36
TCGA01, TCG
A
03
LCAL01, LCAL02, MCAL01
RCAL01
SCFL
01,
 SCFL
02
MCFL01, MCFL02
BCAL03, BCA
L04
WCFL01, WCFL02
AMM01, AMM02, AMM03
TRI01, TRI07, TRI08
LEU08, LEU15
BCAL05, BCA
L01, BCAL02
OCFL01, OCFL02
OCFL04, OCFL05, OCFL03,
OCFL06, OCFL07
WCFL03, WCFL04, WCFL05, WCFL06
WCFL01, WCFL02 WCFL
03
JCFL01, JCFL02, JCFL03
CCFL01
LCFL01, LCFL02
WCFL07, WCFL08, WCFL
09
ALL02**
ALL02*, ALL0
4*
ALL04, AL
L05
PEN02*, PEN03*
PE
N04,
 PE
N05
78
 79
Fig. 3  Diagram depicting the section of mtDNA that was sequenced and placement of 
the insert found in Peromyscus polionotus but not found in the other Peromyscus used in 
this study.  
F
Cytochrom
e
-b
T
P
D-loop
12s rRNA
15  424 D-loop start
Insert ~15 578 -
15 650
16  395 D-loop end
80
 81
Fig. 4  Maximum likelihood topology using the cytochrome-b region and model HKY +? 
with outgroup Peromyscus leucopus.  Model selection was based on log likelihood score.  
Numbers above branches are bootstrap values. 
10
PMAN103 102
PP36
PI21
AL01
SR08
CR01
PK07
GA18
BT01
NB01
PO76
PPN17
AU01
ST05
89CI02
TH02
CH05
GB02
61
63
63
BE01
FR05
MR03
BL01
EG02
60
FA02
FR01
FA01
CT02
CT01
BE02
EG01
52
SC01
TL02
98
SD01
BC04
CR03
BC01
FR02
TU01
MR01
86
EN01
GA30
86
66
100
BW02
SM04
PMP93
82
73
100
BCAL05
PL103
PG36
100
PK02
AMM01
LEU08
CCFL01
TRI07
TCGA01
LCFL01
SCFL01
MCFL76
NIV17
LCAL01
PEN05
PEN02*
ALL02**
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
WCFL10
WCFL13
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
PMB02
PMS04
P93
73
63
61
100
63
60
52
98
66
86
86
82
ALL05
ALL02*
82
 83
Fig. 5  Maximum likelihood topology using the cytochrome-b region and model TVM +? 
with outgroup Peromyscus leucopus.  Model selection was based on AIC score.  
Numbers above branches are bootstrap values. 
10
PMAN103 102
PP36
PI21
AL01
SR08
CR01
PK07
GA18
FA02
FR01
FA01
CT02
CT01
BE02
EG01
BT01
NB01
PO76
PPN17
AU01
ST05
89CI02
TH02
CH05
GB02
60
64
64
BE01
FR05
MR03
BL01
EG02
62
SC01
TL02
99
SD01
BC04
CR03
BC01
FR02
TU01
MR01
88
EN01
GA30
87
64
100
BW02
SM04
PMP93
81
65
100
PL103
PG36
PK02
AMM01
LEU08
CCFL01
TRI07
TCGA01
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
LCFL01
SCFL01
MCFL76
NIV17
LCAL01
PEN05
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
WCFL10
WCFL13
BCAL05
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
PMB02
PMS04
PMP93
100
100
64
64
60
62
9965
64
88
87
81
84
 85
Fig. 6  Maximum likelihood topology using the cytochrome-b region and model HKY +? 
with outgroup Peromyscus keeni.  Model selection was based on log likelihood score.  
Numbers above branches are bootstrap values. 
 
10
PI21
BW02
SM04
PMP93
95
AL01
SR08
CR01
PK07
GA18
BT01
NB01
PO76
PPN17
AU01
ST05
89CI02
TH02
CH05
GB02
61
63
63
BE01
FR05
MR03
BL01
EG02
62
FA02
FR01
FA01
CT02
CT01
BE02
EG01
53
SC01
TL02
98
SD01
BC04
CR03
BC01
FR02
TU01
MR01
87
EN01
GA30
86
68
99
PK02
PMB02
PMS04
P93
AMM01
LEU08
CCFL01
TRI07
TCGA01
LCFL01
JCFL01
MCFL76
NIV17
LCAL01
PEN05
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
WCFL10
WCFL13
BCAL05
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
95
63
63
61
99
62
53
98
68
87
86
86
 87
Fig. 7  Maximum likelihood topology using the cytochrome-b region and model GTR +? 
with outgroup Peromyscus keeni.  Model selection was based on AIC score.  Numbers 
above branches are bootstrap values. 
10
PI21
BW02
SM04
PMP93
94
AL01
SR08
CR01
PK07
GA18
FA02
FR01
FA01
CT02
CT01
BE02
EG01
BT01
NB01
PO76
PPN17
AU01
ST05
89CI02
TH02
CH05
GB02
60
62
62
BE01
FR05
MR03
BL01
EG02
60
SC01
TL02
98
SD01
BC04
CR03
BC01
FR02
TU01
MR01
85
EN01
GA30
86
66
100
PK02
PMB02
PMS04
P93
MM01
LEU08
CCFL01
TRI07
TCGA01
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
LCFL01
SCFL01
MCFL76
NIV17
LCAL01
PEN05
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
WCFL10
WCFL13
BCAL05
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
94
100
62
62
60
60
98
66
85
86
88
 89
Fig. 8  Maximum likelihood topology using the D-loop and model HKY +I +? with 
outgroup Peromyscus leucopus.  Model selection was based on both log likelihood and 
AIC scores.  Numbers above branches are bootstrap values. 
10
PMAN103 102
PP36
PI21
CH05
GB02
PO76
BT01
NB01
CT01
EG01
FA01
FA02
BE02
CT02
FR01
64
98
50
SD01
BC01
TU01
BC04
MR01
CR03
FR02
86
58
EN01
GA30
71
60
AU01
GA18
60
89CI02
TH02
57
CR01
PK07
ST05
AL01
SR08
82
BL01
BE01
FR05
63
64
EG02
MR03
61
96
52
61
SC01
TL02
99
PPN17
PPN36
99
100
BW02
PMP93
SM04
69
100
58
100
PL103
PG36
PK02
ALL05
ALL02*
MCFL76
LCFL01
SCFL01
OCFL04
OCFL06
WCFL04
WCFL05
WCFL02
OCFL05
WCFL06
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
LCAL01
TCGA01
PEN02*
ALL02**
CCFL01
TRI07
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
WCFL10
WCFL13
NIV17
NIV36
PMB02
PMP93
PMS04
100
50
98
64
60
58
86
71
100
60
57
58
61
52
82
96
64
63
61
99
99
100
69
90
 91
Fig. 9  Maximum likelihood topology using the D-loop and model HKY +I with 
outgroup Peromyscus keeni.  Model selection was based on both log likelihood and AIC 
scores.  Numbers above branches are bootstrap values. 
10
PI21
CH05
GB02
PO76
BT01
NB01
CT01
EG01
FA01
FA02
BE02
CT02
FR01
65
98
52
SD01
BC01
TU01
BC04
MR01
CR03
FR02
87
59
EN01
GA30
71
54
AU01
GA18
63
89CI02
TH02
58
CR01
PK07
ST05
AL01
SR08
81
BL01
BE01
FR05
58
60
EG02
MR03
64
97
50
61
SC01
TL02
99
PPN17
PPN36
98
100
BW02
PMP93
SM04
64
100
PK02
ALL05
ALL02*
MCFL76
LCFL01
SCFL01
OCFL04
OCFL06
WCFL04
WCFL05
WCFL02
OCFL05
WCFL06
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
LCAL01
TCGA01
PEN02*
ALL02**
CCFL01
TRI07
PEN05
MM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
WCFL10
WCFL13
NIV17
NIV36
PMB02
P93
PMS04
52
98
65
100
54
59
87
71
63
58
61
50
81
97
60
58
64
99
98
100
64
92
 93
Fig. 10  Maximum likelihood topology using the cytochrome-b region and D-loop from 
model HKY +I +? with outgroup Peromyscus leucopus.  Model selection was based on 
both log likelihood and AIC scores.  Numbers above branches are bootstrap values. 
10
PMAN103 102
PP36
PI21
PO76
PPN17
SD01
BC04
MR01
BC01
TU01
CR03
FR02
86
55
92
EN01
GA30
89
89
ST05
AL01
SR08
92
BL01
BE01
FR05
62
61
EG02
MR03
62
99
CR01
PK07
65
89CI02
TH02
62
CH05
GB02
63
84
76
AU01
GA18
58
56
SC01
TL02
100
NB01
BT01
CT01
FA01
FA02
EG01
BE02
CT02
FR01
63
100
75
53
100
PMP93
SM04
BW02
100
86
100
PL103
PG36
PK02
MCFL76
NIV17
BCAL05
BCAL04
JCFL07
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
CCFL01
TRI07
PEN02*
ALL02**
ALL05
ALL02*
LCAL01
TCGA01
WCFL10
WCFL13
SCFL01
LCFL01
OCFL04
WCFL04
WCFL05
OCFL06
WCFL02
OCFL05
WCFL06
PMP93
PMS03
PMB02
89
92
55
86
89
100
100
56
76
92
99
61
62
62
65
84
62
63
58
100
53
75
100
63
100
86
94
 95
Fig. 11  Maximum likelihood topology using the cytochrome-b region and D-loop from 
model HKY +I +? with outgroup Peromyscus keeni.  Model selection was based on both 
log likelihood and AIC scores.  Numbers above branches are bootstrap values. 
10
PI21
PO76
PPN17
SD01
BC04
MR01
BC01
TU01
CR03
FR02
85
53
91
EN01
GA30
91
86
ST05
AL01
SR08
90
BL01
BE01
FR05
63
64
EG02
MR03
62
98
CR01
PK07
64
89CI02
TH02
60
CH05
GB02
62
83
74
AU01
GA18
60
56
SC01
TL02
100
NB01
BT01
CT01
FA01
FA02
EG01
BE02
CT02
FR01
62
99
71
51
100
BW02
PMP93
SM04
60
100
PK02
MCFL76
NIV17
BCAL05
BCAL04
JCFL01
BCAL01
MCAL01
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
CCFL01
TRI07
PEN02*
ALL02**
ALL05
ALL02*
LCAL01
TCGA01
WCFL10
WCFL13
SCFL01
LCFL01
OCFL04
WCFL04
WCFL05
OCFL06
WCFL02
OCFL05
WCFL06
PMB02
PMP93
PMS04
86
91
53
85
91
100
90
56
74
98
64
63
62
64
83
60
62
60
100
51
71
99
62
100
60
96
 97
Fig. 12  Weighted parsimony topology using the cytochrome-b region with outgroup 
Peromyscus leucopus.  The topology represents a strict consensus tree of 28 equally 
parsimonious topologies (CI = 0.9195, RI = 0.9071, RC = 0.8341, Tratio = 9.5:1, 1st 
position weight = 5:1, 2nd position weight = 40:1, Tree Length = 1646).  Numbers above 
branches are bootstrap values. 
PMAN103 102
PP36
PI21
PPN17
AL01
SR08
CR01
PK07
GA18
NB01
PO76
ST05
89CI02
TH02
CH05
GB02
BE01
FR05
MR03
BL01
EG02
BT01
FA02
FR01
FA01
CT02
CT01
BE02
EG01
SC01
TL02
SD01
AU01
BC04
CR03
BC01
FR02
TU01
MR01
EN01
GA30
BW02
SM04
PMP93
63
61
62
61
99
55
87
85
97
PL103
PG36
PK02
NIV17
MM01
LEU08
CCFL01
TRI07
TCGA01
JCFL01
MCFL76
PEN05
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
LCFL01
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
WCFL10
WCFL12
BCAL05
LCAL01
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
PMB02
PMS02
P93
100
87
100
51
61
55
98
 99
Fig. 13  Unweighted parsimony topology using the cytochrome-b region with outgroup 
Peromyscus leucopus.  The topology represents a strict consensus tree of 77 equally 
parsimonious topologies (CI = 0.7684, RI = 0.8205 RC = 0.6305, tree length = 272).  
Numbers above branches are bootstrap values. 
PMAN103 102
PP36
PI21
AL01
SR08
CR01
PK07
GA18
NB01
PO76
PPN17
SD01
AU01
ST05
89CI02
TH02
CH05
GB02
BE01
FR05
MR03
BL01
EG02
BT01
FA02
FR01
FA01
CT02
CT01
BE02
EG01
SC01
TL02
BC04
CR03
BC01
FR02
TU01
MR01
EN01
GA30
BW02
SM04
PMP93
95
100
86
50
62
61
64
99
PL103
PG36
PK02
AMM01
LEU08
CCFL01
TRI07
TCGA01
SCFL01
MCFL76
NIV17
BCAL05
LCAL01
PEN05
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
LCFL01
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
WCFL10
WCFL13
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
PMB02
PMS04
PMP93
100
67
61
52
100
 101
Fig. 14  Weighted parsimony topology using the cytochrome-b region with outgroup 
Peromyscus keeni.  The topology represents a strict consensus tree of 14 equally 
parsimonious topologies (CI = 0.9553, RI = 0.9439, RC = 0.9017, Tratio = 18:1, 1st 
position weight = 5:1, 2nd position weight = 37:1, tree length = 805).  Numbers above 
branches are bootstrap values. 
PI21
SM04
BW02
PMP93
PPN17
AL01
SR08
CR01
PK07
GA18
NB01
PO76
ST05
89CI02
TH02
CH05
GB02
BE01
FR05
MR03
BL01
EG02
BT01
FA02
FR01
FA01
CT02
CT01
BE02
EG01
SC01
TL02
SD01
AU01
BC04
CR03
BC01
FR02
TU01
MR01
EN01
GA30
99
100
64
62
60
61
98
56
86
87
52
59
62
55
102
 103
Fig. 15  Unweighted parsimony topology using the cytochrome-b region with outgroup 
Peromyscus keeni.  The topology represents a strict consensus tree of 38 equally 
parsimonious topologies (CI = 0.8389, RI = 0.8863, RC = 0.7435, Tree Length = 149).  
Numbers above branches are bootstrap values. 
PI21
SM04
BW02
PMP93
AL01
SR08
CR01
PK07
GA18
NB01
PO76
PPN17
SD01
AU01
ST05
89CI02
TH02
CH05
GB02
BE01
FR05
MR03
BL01
EG02
BT01
FA02
FR01
FA01
CT02
CT01
BE02
EG01
SC01
TL02
BC04
CR03
BC01
FR02
TU01
MR01
EN01
GA30
100
100
64
62
61
52
98
87
86
PK02
PMS04
PMB02
P93
AMM04
LEU08
CCFL01
TRI07
TCGA01
SCFL01
MCFL76
BCAL05
LCAL01
PEN05
NIV17
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
LCFL01
WCFL06
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
WCFL10
WCFL12
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
60
59
51
104
 105
Fig. 16  Weighted parsimony topology using the D-loop with outgroup Peromyscus 
leucopus.  The topology represents a strict consensus tree of 17 equally parsimonious 
topologies (CI = 0.8631, RI = 0.8922, RC = 0.7700, Tratio = 6:1, tree length = 745).  
Numbers above branches are bootstrap values. 
PMAN103 102
PP36
BT01
CT01
EG01
FA01
FA02
BE02
CT02
FR01
NB01
PO76
SC01
TL02
SD01
BC01
TU01
BC04
MR01
CR03
FR02
EN01
GA30
AU01
GA18
CH05
GB02
89CI02
TH02
CR01
PK07
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
PPN17
PPN36
PI21
BW02
PMP93
SM04
100
100
64
62
63
58
100
61
64
85
69
56
67
62
100
PL103
PG36
LCFL01
OCFL04
OCFL06
WCFL04
WCFL05
WCFL02
OCFL05
WCFL06
SCFL01
MCFL76
WCFL10
WCFL13
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
LCAL01
TCGA01
ALL05
ALL02*
PEN02*
ALL02**
CCFL01
TRI07
PEN05
MM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
NIV17
NIV36
PK02
PMB02
PMP
PMS04
99
98
64
62
100
57
106
 107
Fig. 17  Unweighted parsimony topology using the D-loop with outgroup Peromyscus 
leucopus.  The topology represents a strict consensus tree of 22 equally parsimonious 
topologies (CI = 0.7653, RI = 0.8433, RC = 0.6454, tree length = 311).  Numbers above 
branches are bootstrap values. 
PMAN103 102
PP36
PI21
NB01
BT01
PO76
CT01
EG01
FA01
FA02
BE02
CT02
FR01
SD01
BC01
TU01
BC04
MR01
CR03
FR02
EN01
GA30
AU01
GA18
CH05
GB02
89CI02
TH02
CR01
PK07
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
SC01
TL02
PPN17
PPN36
BW02
PMP93
SM04
99
99
99
97
61
79
52
54
60
65
67
88
68
65
98
PL103
PG36
PK02
SCFL01
LCFL01
MCFL76
OCFL04
OCFL06
WCFL04
WCFL05
WCFL02
OCFL05
WCFL06
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
LCAL01
TCGA01
ALL05
ALL02*
PEN02*
ALL02**
CCFL01
TRI07
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
WCFL10
WCFL13
NIV17
NIV36
PMB02
PMP93
PMS04
100
100
64
66
62
108
 109
Fig. 18  Weighted parsimony topology using the D-loop with outgroup Peromyscus 
keeni.  The topology represents a strict consensus of 32 equally parsimonious topologies 
(CI = 0.8465, RI = 0.9094, RC = 0.7698, Tratio = 10:1, Tree Length = 482).  Numbers 
above branches are bootstrap values. 
PI21
BT01
CT01
EG01
FA01
FA02
BE02
CT02
FR01
NB01
PO76
SC01
TL02
SD01
BC01
TU01
BC04
MR01
CR03
FR02
EN01
GA30
AU01
GA18
CH05
GB02
89CI02
TH02
CR01
PK07
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
PPN17
PPN36
BW02
PMP93
SM04
100
100
99
98
63
63
61
55
57
69
86
69
63
54
100
64
64
PK02
LCFL01
OCFL04
OCFL06
WCFL04
WCFL05
WCFL02
OCFL05
WCFL06
SCFL01
MCFL76
WCFL10
WCFL13
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL02
RCAL01
TCGA03
LCAL01
TCGA01
ALL05
ALL02*
PEN02*
ALL02**
CCFL01
TRI07
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
NIV17
NIV36
PMB02
PMP93
PMS04
98
61
55
110
 111
Fig. 19  Unweighted maximum parsimony topology using the D-loop with outgroup 
Peromyscus keeni.  The topology represents a strict consensus tree of 520 equally 
parsimonious topologies (CI = 0.7500, RI = 0.8682, RC = 0.6511, tree length = 184).  
Numbers above branches are bootstrap values. 
PI21
BT01
NB01
AU01
GA18
89CI02
TH02
CH05
GB02
PO76
CT01
EG01
FA01
FA02
BE02
CT02
FR01
SD01
BC01
TU01
BC04
MR01
CR03
FR02
EN01
GA30
ST05
CR01
PK07
AL01
SR08
BL01
BE01
FR05
EG02
MR03
SC01
TL02
PPN17
PPN36
BW02
PMP93
SM04
100
100
99
99
62
97
63
7759
67
85
72
68
62
98
PK02
LCFL01
SCFL01
LCAL01
TCGA01
PEN02*
ALL02**
ALL05
ALL02*
MCFL76
OCFL04
OCFL06
WCFL04
WCFL05
WCFL02
OCFL05
WCFL06
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
CCFL01
TRI07
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
WCFL10
WCFL13
NIV17
NIV36
PMB02
P93
PMS04
60
68
112
 113
Fig. 20  Weighted parsimony topology using the cytochrome-b and D-loop with outgroup 
Peromyscus leucopus.  The topology represents a strict consensus tree of 93 equally 
parsimonious topologies (CI = 0.8639, RI = 0.8718, RC = 0.7532, Tratio = 7:1, 1st 
position weight = 5:1, 2nd position weight = 40:1, tree length = 1668).  Numbers above 
branches are bootstrap values. 
PMAN103 102
PP36
PI21
PPN17
GA18
AU01
SD01
BC01
TU01
BC04
MR01
CR03
FR02
EN01
GA30
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
CR01
PK07
89CI02
TH02
CH05
GB02
NB01
PO76
SC01
TL02
BT01
CT01
FA01
FA02
EG01
BE02
CT02
FR01
PMP93
SM04
BW02
100
88
65
89
63
63
64
62
100
65
80
94
88
PL103
PG36
PK02
NIV17
TCGA01
LCAL01
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
MM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
CCFL01
TRI07
PEN02*
ALL02**
ALL05
ALL02*
SCFL01
MCFL76
WCFL10
WCFL13
LCFL01
OCFL04
WCFL04
WCFL05
OCFL06
WCFL02
OCFL05
WCFL06
PMP93
PMS04
PMB02
64
100
99
114
 115
Fig. 21  Unweighted parsimony topology using the cytochrome-b region and D-loop with 
outgroup Peromyscus leucopus.  The topology represents a strict consensus tree of 26 
equally parsimonious topologies (CI = 0.7687, RI = 0.8273, RC = 0.6359, tree length = 
614).  Numbers above branches are bootstrap values. 
PMAN103 102
PP36
PI21
NB01
PO76
PPN17
SD01
BC01
TU01
BC04
MR01
CR03
FR02
EN01
GA30
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
CR01
PK07
89CI02
TH02
CH05
GB02
AU01
GA18
SC01
TL02
BT01
CT01
FA01
FA02
EG01
BE02
CT02
FR01
PMP93
SM04
BW02
89
80
85
91
92
99
65
64
63
64
53
100
82
61
100
PL103
PG36
PK02
SCFL02
MCFL76
NIV17
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL08
OCFL07
JCFL03
CFL01
TRI07
PEN02*
ALL02**
ALL05
ALL02*
LCAL01
TCGA01
WCFL10
WCFL13
LCFL01
OCFL04
WCFL04
WCFL05
OCFL06
WCFL02
OCFL05
WCFL06
PMP93
PMS04
PMB02
100
80
64
65
99
116
 117
Fig. 22  Weighted parsimony topology using the cytochrome-b region and D-loop with 
outgroup Peromyscus keeni.  The topology represents a strict consensus tree of 62 equally 
parsimonious topologies (CI = 0.8752, RI = 0.9092, RC = 0.7958, Tratio = 12:1, 1st 
position weight = 5:1, 2nd position weight = 37:1, tree length = 1002).  Numbers above 
branches are bootstrap values. 
PI21
PPN17
GA18
AU01
SD01
BC01
TU01
BC04
MR01
CR03
FR02
EN01
GA30
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
CR01
PK07
89CI02
TH02
CH05
GB02
NB01
PO76
SC01
TL02
BT01
CT01
FA01
FA02
EG01
BE02
CT02
FR01
PMP93
SM04
BW02
88
65
62
61
66
99
69
95
85
94
60
88
PK02
NIV17
TCGA01
LCAL01
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
CCFL01
TRI07
PNE02*
ALL02**
ALL05
ALL02*
SCFL01
MCFL76
WCFL10
WCFL13
LCFL01
OCFL04
WCFL04
WCFL05
OCFL07
WCFL02
OCFL05
WCFL06
P93
PMS04
PMB02
63
62
99
100
118
 119
Fig. 23  Unweighted maximum parsimony topology using the cytochrome-b region and 
D-loop with outgroup Peromyscus keeni.  The topology is represents a strict consensus 
tree of 440 equally parsimonious topologies (CI = 0.7853, RI = 0.8645, RC = 0.6789, tree 
length = 354).  Numbers above branches are bootstrap values. 
PI21
AU01
GA18
PO76
PPN17
NB01
SD01
BC01
TU01
BC04
MR01
CR03
FR02
EN01
GA30
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
CR01
PK07
89CI02
TH02
CH05
GB02
SC01
TL02
BT01
CT01
FA01
FA02
EG01
BE02
CT02
FR01
PMP93
SM04
BW02
60
100
100
82
63
100
88
64
91
65
63
99
66
80
92
88
PK02
LCAL01
TCGA01
MCFL76
NIV17
SCFL01
BCAL05
BCAL01
MCAL01
BCAL04
JCFL01
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
CCFL01
TRI07
PEN02*
ALL02**
ALL05
ALL02*
WCFL10
WCFL13
LCFL01
OCFL04
WCFL04
WCFL05
OCFL06
WCFL02
OCFL05
WCFL06
P93
PMS04
PMB02
99
120
 121
Fig. 24  Bayesian analysis using the cytochrome-b region with model GTR +? and 
outgroup Peromyscus leucopus.  The topology represents a 50% majority-rule consensus 
of 192 186 topologies.  Model selection was based on both log likelihood and AIC 
scores.  Numbers above branches are bootstrap values. 
PMAN103 102
PP36
PI21
NB01
PO76
PPN17
CR01
PK07
GA18
AU01
AL01
SR08
ST05
89CI02
TH02
CH05
GB02
BE01
FR05
MR03
BL01
EG02
BT01
FA02
FR01
FA01
CT02
CT01
BE02
EG01
SC01
TL02
SD01
BC04
CR03
BC01
FR02
TU01
MR01
EN01
GA30
PMP93
BW02
SM04
100
99
100
100
95
100
69
89
95
97
54
63
PL103
PG36
PK02
SCFL01
MCFL76
NIV17
CCFL01
TRI07
TCGA01
LCAL01
MM01
LEU08
PEN05
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
LCFL01
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
WCFL10
WCFL13
BCAL05
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
PMP93
PMB02
PMS04
88
96
67
60
100
122
 123
Fig. 25  Bayesian analysis using the cytochrome-b region with model HKY +? and 
outgroup Peromyscus keeni.  The topology represents a 50% majority-rule consensus tree 
of 191 840 topologies.  Model selection was based on log likelihood scores.  Numbers 
above branches are bootstrap values. 
PI21
BW02
SM04
PMP93
NB01
PO76
PPN17
CR01
PK07
GA18
AU01
AL01
SR08
ST05
89CI02
TH02
CH05
GB02
BE01
FR05
MR03
BL01
EG02
BT01
FA02
FR01
FA01
CT02
CT01
BE02
EG01
SC01
TL02
SD01
BC04
CR03
BC01
FR02
TU01
MR01
EN01
GA30
100
62
94
97
56
91
100
83
100
96
100
100
PK02
PMB02
PMS04
PMP93
JCFL01
MCFL76
NIV17
CCFL01
TRI07
TCGA01
LCAL01
AMM01
LEU08
PEN05
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
LCFL01
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
WCFL10
WCFL13
BCAL05
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
96
76
124
 125
Fig. 26  Bayesian analysis using the cytochrome-b region with model GTR +I and 
outgroup Peromyscus keeni.  The topology represents a 50% majority-rule of 193 162 
topologies.  Model selection was based on AIC score.  Numbers above branches are 
bootstrap values. 
PI21
BW02
SM04
PMP93
BT01
NB01
PO76
PPN17
GA18
AU01
CR01
PK07
AL01
SR08
ST05
89CI02
TH02
CH05
GB02
BE01
FR05
MR03
BL01
EG02
FA02
FR01
FA01
CT02
CT01
BE02
EG01
SC01
TL02
SD01
BC04
CR03
BC01
FR02
TU01
MR01
EN01
GA30
100
100
51
59
69
89
97
83
50
100
90
99
99
PK02
PMB02
PMS04
P93
LCFL01
SCFL02
MCFL76
MIV17
TCGA01
LCAL01
CCFL01
TRI07
AMM01
LEU08
PEN05
PEN02*
ALL02**
ALL05
ALL02*
WCFL01
WCFL09
JCFL03
OCFL01
OCFL07
WCFL05
WCFL06
WCFL04
OCFL05
OCFL04
WCFL02
OCFL06
WCFL10
WCFL13
BCAL05
BCAL04
OCFL03
BCAL01
WCFL07
MCAL01
JCFL01
RCAL01
TCGA03
96
126
 127
Fig. 27  Bayesian analysis using the D-loop with model HKY +I +? and outgroup 
Peromyscus leucopus.  The topology represents a 50% majority-rule of 191 313 
topologies.  Model selection was based on both log likelihood and AIC scores.  Numbers 
above branches are bootstrap values. 
PMAN103 102
PP36
PI21
BT01
NB01
CT01
EG01
FA01
FA02
BE02
CT02
FR01
AU01
GA18
CH05
GB02
89CI02
TH02
CR01
PK07
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
PO76
SC01
TL02
PPN17
PPN36
BC04
MR01
BC01
TU01
CR03
FR02
SD01
EN01
GA30
BW02
PMP93
SM04
90
100
54
71
100
51
90
75
100
100
54
65
72
95
79
94
99
93
98
92
100
96
PL103
PG36
PK02
LCFL01
SCFL01
OCFL04
OCFL06
WCFL04
WCFL05
WCFL02
OCFL05
WCFL06
LCAL01
TCGA01
ALL05
ALL02*
PEN02*
ALL02**
CCFL01
TRI07
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
MCFL76
WCFL10
WCFL13
NIV17
NIV36
BCAL04
JCFL01
BCAL01
MCAL01
OCFL03
WCFL07
BCAL05
RCAL01
TCGA03
PMB02
PMP93
PMS04
100
96
100
94
83
97
128
 129
Fig. 28  Bayesian analysis using the D-loop with model HKY +I +? and outgroup 
Peromyscus keeni.  The topology represents a 50% majority-rule of 191 737 topologies.  
Model selection was based on both log likelihood and AIC scores.  Numbers above 
branches are bootstrap values. 
PI21
BT01
NB01
CT01
EG01
FA01
FA02
BE02
CT02
FR01
AU01
GA18
CH05
GB02
89CI02
TH02
CR01
PK07
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
PO76
SC01
TL02
PPN17
PPN36
BC04
MR01
BC01
TU01
CR03
FR02
SD01
EN01
GA30
BW02
PMP93
SM04
100
100
100
62
56
75
65
100
75
91
88
100
83
99
91
83
59
64
99
95
100
73
PK02
LCFL01
SCFL01
OCFL04
OCFL07
WCFL04
WCFL05
WCFL02
OCFL05
WCFL06
LCAL01
TCGA01
ALL05
ALL02*
PEN02*
ALL02**
CCFL01
TRI07
PEN05
MM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
MCFL76
WCFL10
WCFL13
NIV17
NIV36
BCAL04
JCFL01
BCAL01
MCAL01
OCFL03
WCFL07
BCAL05
RCAL01
TCGA03
PMB02
P93
PMS04
97
100
93
96
130
 131
Fig. 29  Bayesian analysis using the cytochrome-b region and the D-loop with split 
models HKY +? (cty-b) and HKY +I +? (cyt-b and D-loop) with outgroup Peromyscus 
leucopus.  The topology represents a 50% majority-rule of 190 150 topologies.  Model 
selection is based on both log likelihood and AIC scores.  Numbers above branches are 
bootstrap values. 
PMAN103 102
PP36
PI21
SD01
BC04
MR01
BC01
TU01
CR03
FR02
EN01
GA30
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
CR01
PK07
89CI02
TH02
CH05
GB02
AU01
GA18
PPN17
PO76
SC01
TL02
NB01
BT01
CT01
FA01
FA02
EG01
BE02
CT02
FR01
BW02
PMP93
SM04
100
100
97
84
53
98
94
100
100
98
99
74
89
67
100
92
95
100
99
100
85
100
100
PL103
PG36
PK02
BCAL05
BCAL04
JCFL01
BCAL01
MCAL01
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
MM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
CCFL01
TRI07
PEN02*
ALL02**
ALL05
ALL02*
LCAL01
TCGA01
NIV17
MCFL76
WCFL10
WCFL13
SCFL01
LCFL01
OCFL04
WCFL04
WCFL05
OCFL06
WCFL02
OCFL05
WCFL06
PMB02
P93
PMS04
100
100
79
100
86
100
65
132
 133
Fig. 30  Bayesian analysis using the cytochrome-b region and the D-loop with split 
models HKY +? (cyt-b) and HKY +I +? (cyt-b and D-loop) with outgroup Peromyscus 
keeni.  The topology represents a 50% majority-rule of 194 997 topologies.  Model 
selection is based on both log likelihood and AIC scores.  Numbers above branches are 
bootstrap values. 
PI21
PPN17
SD01
BC04
MR01
BC01
TU01
CR03
FR02
EN01
GA30
ST05
AL01
SR08
BL01
BE01
FR05
EG02
MR03
CR01
PK07
89CI02
TH02
CH05
GB02
AU01
GA18
PO76
SC01
TL02
NB01
BT01
CT01
FA01
FA02
EG01
BE02
CT02
FR01
BW02
PMP93
SM04
100
99
100
82
100
100
96
98
63
57
100
100
93
90
86
100
96
99
96
79
80
96
100
PK02
NIV17
BCAL05
BCAL04
JCFL01
BCAL01
MCAL01
CCFL01
WCFL07
RCAL01
TCGA03
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL09
OCFL07
JCFL03
CCFL01
TRI07
PEN02*
ALL02**
ALL05
ALL02*
LCAL01
TCGA01
MCFL76
WCFL10
WCFL13
SCFL01
LCFL01
OCFL04
WCFL04
WCFL05
OCFL06
WCFL02
OCFL05
WCFL06
PMB02
PMP93
PMS04
97
100
87
100
63
134
 135
Fig. 31  Statistical parsimony network (Templeton et al. 1992) of Peromyscus polionotus 
haplotypes based on the number of substitutions.  Numbers represent haplotypes as 
indicated in Table 1.  Broken lines indicate alternative parsimony connections.  
Alternative loops are labeled 1A - 1C and 2.  Solid circles represent inferred haplotypes 
that were not present in the sample.  Lines between inferred and existing haplotypes 
represent a single mutational step. 
?
?
?
?
1A
49
17
?
?
?
?
?
? ??
? ? ? ? ? ? ? ? ?
?
?
?
?
?
?
?
?
?
??
?26
35
?
78
?
8
?
11
46
?
?
?
?
?
33
29
? 28 13
? ? ? ? ? ? ? ? ? 67
?
?
?
? ? 40
??31
?
?
?
?
?
?
?
?
?
?
?
?
?
22
?? 75
?
?
?? ? ? ? ?
?
73?
?
 
?
 
?
 
?
48
??
 ?
30
?
?
12
15
?? ? ?70
3
? ? ? ? ? ? ? 54
?
25
??
 ?
 ?
? ? ? ? ? ? ? ?
?
? 
? 
? 
? 
?
 ?
6
39
?
? ? ? ? ? ? ? ? ? ? ? ? ?
? ?
 ? 
? ?
1
?
? 
? 
? 
? 
? 
?
?
? 
?
64
63
59
?
?
 
?
 
?
 
?
76
65
2
1C
1B
136
 137
Fig. 32  Constant-rate time estimates for lineage divergence?s within Peromyscus 
polionotus where t1 = 306,000 ? 48,000, t2 = 243,000 ? 49,000, t3 = 195,000 ? 34,000, 
t4 = 185,000 ? 86,000, t5 = 159,000 ? 49,000, t6 = 151,000 ? 36,000, t7 = 63,000 ? 
32,000, t8 = 430,000 ? 145,000. 
 
t1
t2
t5
t4
t7
t3
t6
Time
t8
PK01
MCFL01
NIV17
BCAL05
BCAL04
JCFL01
BCAL01
MCAL01
OCFL03
WCFL07
RCAL01
TCGA03
PEN05
AMM01
LEU08
OCFL01
WCFL01
WCFL06
OCFL07
JCFL03
CCFL01
TRI07
PEN02*
ALL02**
ALL05
ALL02*
LCAL01
TCGA01
WCFL10
WCFL13
SCFL01
LCFL01
OCFL04
WCFL01
WCFL05
OCFL06
WCFL02
WCFL03
OCFL05
PMB02
PMP93
PMS04
138
 139
Fig. 33  Range map of Peromyscus polionotus allophrys (Bowen 1968) and location  of 
the debated haplotype (Crooked Island, FL).  Bowen (1968) concluded Crooked Island 
was a section of the range of Peromyscus polionotus peninsularis.  However, 
phylogenetic analyses placed haplotypes from Crooked Island with haplotypes from P.  p.  
allophrys. 
P
. 
p
. 
al
l
o
p
h
rys
range desi
gnat
e
d 
by
 B
o
wen (1
9
68).
C
ro
oked Is
l
a
nd, 
F
L
, a p
o
rt
i
o
n 
of 
P
. 
p
. 
peni
ns
ul
a
r
i
s
range 
per B
o
wen (
1
9
69).
140
Table 1  Key to specimens abbreviations relative to species and collection site (BS = Bon 
Secour Wildlife Refuge, WMRS = White Mountain Research Station).  The subspecific 
designation of some polionotus is uncertain as designated by ? 
 
Number Abbreviation Species Location 
  1 PEN02* P. p. peninsularis ? Crooked Island, FL 
  2 PEN03* P. p. peninsularis ? Crooked Island, FL 
  3 AMM01 P. p. ammobates BS Perdue unit, AL 
  4 AMM02 P. p. ammobates BS Fort Morgan Unit, AL 
  5 AMM03 P. p. ammobates BS Fort Morgan Unit, AL 
  6 LCAL01 P. p. polionotus Lee County, AL 
  7 LCAL02 P. p. polionotus Lee County, AL 
  8 BCAL01 P. p. polionotus Baldwin County, AL 
  9 BCAL02 P. p. polionotus Baldwin County, AL 
10 BCAL03 P. p. polionotus Baldwin County, AL 
11 BCAL04 P. p. polionotus Baldwin County, AL 
12 WCFL01 P. p. sumneri ? Walton County, FL 
13 WCFL02 P. p. sumneri ? Walton County, FL 
14 WCFL03 P. p. sumneri ? Walton County, FL 
15 OCFL01 P. p. polionotus Okaloosa County, FL 
16 OCFL02 P. p. polionotus Okaloosa County, FL 
17 LCFL01 P. p. subgriseus Liberty County, FL 
18 LCFL02 P. p. subgriseus Liberty County, FL 
19 PMB01 P. maniculatus bairdii Near Ann Arbor, MI 
20 PMB02 P. maniculatus bairdii Near Ann Arbor, MI 
21 PMB03 P. maniculatus bairdii Near Ann Arbor, MI 
22 ALL04 P. p. allophrys Shell Island, FL 
23 ALL05 P. p. allophrys Shell Island, FL 
24 ALL06 P. p. allophrys Shell Island, FL 
25 CCFL01 P. p. sumneri ? Calhoun County, FL 
26 OCFL03 P. p. albifrons Okaloosa County, FL 
27 OCFL04 P. p. albifrons Okaloosa County, FL 
28 OCFL05 P. p. alibfrons Okaloosa County, FL 
29 OCFL06 P. p. griseobracatus Okaloosa County, FL 
30 OCFL07 P. p. griseobracatus Okaloosa County, FL 
31 RCAL01 P. p. polionotus Russell County, AL 
32 WCFL04 P. p. polionotus Walton County, FL 
33 WCFL05 P. p. polionotus Walton County, FL 
34 WCFL06 P. p. griseobracatus Walton County, FL 
35 WCFL07 P. p. griseobracatus Walton County, FL 
36 WCFL08 P. p. griseobracatus Walton County, FL 
37 WCFL09 P. p. griseobracatus Walton County, FL 
 
141
 
142
38 TCGA06 P. p. polionotus Taylor County, GA 
39 TCGA01 P. p. polionotus Taylor County, GA 
40 TCGA03 P. p. polionotus Taylor County, GA 
41 TCGA04 P. p. polionotus Taylor County, GA 
42 TCGA05 P. p. polionotus Taylor County, GA 
43 ALL02* P. p. allophrys Grayton Beach, FL 
44 ALL03* P. p. allophrys Grayton Beach, FL 
45 ALL04* P. p. allophrys Grayton Beach, FL 
46 JCFL01 P. p. polionotus Jackson County, FL 
47 JCFL02 P. p. polionotus Jackson County, FL 
48 JCFL03 P. p. polionotus Jackson County, FL 
49 SCFL01 P. p. subgriseus Suwannee County, FL 
50 SCFL02 P. p. subgriseus Suwannee County, FL 
51 PK01 P. keeni Kittitas County, WA 
52 PK02 P. keeni Kittitas County, WA 
53 TRI01 P. p. trissyllepsis Johnson Beach, FL 
54 TRI07 P. p. trissyllepsis Florida Point, AL 
55 TRI08 P. p. trissyllepsis Florida Point, AL 
56 PL103 P. leucopus Unicoi County, TN 
57 PMP92 P. maniculatus pallescens Robertson County, TX 
58 PMP93 P. maniculatus pallescens Robertson County, TX 
59 MCFL76 P. p. subgriseus Marion County, FL 
60 MCFL77 P. p. subgriseus Marion County, FL 
61 PG06 P. gossypinus Baldwin County, AL 
62 PG36 P. gossypinus Baldwin County, AL 
63 NIV17 P. p. niveiventris Brevard County, FL 
64 NIV36 P. p. niveiventris Brevard County, FL 
65 WCFL10 P. p. roadsi Wakulla County, FL 
66 WCFL11 P. p. roadsi Wakulla County, FL 
67 BCAL05 P. p. polionotus Baldwin County, AL 
68 PMS04 P. maniculatus sonoriensis Near WMRS, CA 
69 PMS05 P. maniculatus sonoriensis Near WMRS, CA 
70 LEU08 P. p. leucocephalus Santa Rosa Island, FL 
71 LEU15 P. p. leucocephalus Santa Rosa Island, FL 
72 PEN04 P. p. peninsularis St. Joe State Park, FL 
73 PEN05 P. p. peninsularis St Joe State Park, FL 
74 PEN06 P. p. peninsularis St. Joe State Park, FL 
75 ALL02** P. p. allophrys Topsail Hill SP, FL 
76 WCFL12 P. p. roadsi Wakulla County, FL 
77 WCFL13 P. p. roadsi Wakulla County, FL 
78 MCAL01 P. p. polionotus Macon County, AL 
 
Table 2  Measures of diversity for two sets of sequence data that were used for 
phylogenetic analyses.  The first group contained 44 sequences with Peromyscus 
leucopus and Peromyscus gossypinus as outgroup haplotypes.  The second group  
  contained 42 sequences and Peromyscus keeni as the outgroup haplotype. 
 
Group N S Mutations Hd     Pi     k h 
P. leucopus        
Total 44 434 482 0.944 0.0217 52.97 39 
Cyt-b: 44 193 209 0.977 0.0202 22.99 33 
D-loop: 44 241 249 0.994 0.0294 28.37 36 
        
P. keeni 
Total: 42 267 277 0.993 0.0141 34.54 38 
Cyt-b: 42 121 125 0.974 0.0126 14.27 31 
D-loop: 42 132 138 0.988 0.0198 19.21 34 
 
 
143
Table 3
 Permutation Chi-square tests of geogr
aphical association of clades and th
e inferred biological
 explanation.   
Inferences were based on the 14 July 2004 key (
h
ttp
://d
arwin.u
v
i
g
o
.es/down
l
o
a
d
/
g
e
od
isKey_
14
Ju
l04
.
pd
f)
.  Abbreviated  
inferences R
G
F, IBD, PF, LDC, and LDD are restricted ge
ne flow, isolation by distance,
 past fragm
e
ntation, long  
distance colonization, and long dist
ance dispersal respectively.  M
odels were; all haplotypes, F
l
orida peninsula haplotypes  
rem
oved and peninsula rem
oved with LCAL01 a
nd TCGA01 
branching f
r
om
 a basal node of the beach clade. 
 
Scenario 
Clade 
level 
Chi-square 
P 
Chain of inference 
Inference 
     Model I 
2-21 
23.25 
0.0535 
1-19-202-11-17-4 
RGF with IBD 
 
3-32 
16.00 
0.0203 
1-2-11-17 
Inconclusive 
outcom
e
 
 
4-13 
21.00 
0.0321 
1-19-20-2-3-5-15 
PF and/or LDC 
 
4-44 
17.33 
0.2994 
1-19-20-2-11-17-
NO 
Inconclusive 
outcom
e
 
 
4-55 
11.00 
0.0782 
1-2-11-12 
Contiguous 
range 
expansion 
 
4-66 
11.00 
0.0433 
1-19-20-2-3-4-9 
Allopatric fragm
entation 
 
5-2 
20.00 
0.0030 
1-19-20-2-3-4 
RGF with IBD 
 
5-3 
22.00 
0.0026 
 
No significant clade distances 
 
Total 
Clade 
91.50 
<0.0000 
1-2-3-5-6-
7 
RGFwith Di
spersal with LDD 
 
 
 
 
 
 
     Model II 
2-21 
23.25 
0.0545 
1-19-20-2-11-17-4 
RGF with IBD 
 
3-22 
16.00 
0.0212 
1-2-11-17 
Inconclusive 
outcom
e
 
 
4-13 
21.00 
0.0334 
1-19-20-2-3-5-15 
PF and/or LDC 
 
4-34 
17.33 
0.2972 
1-19-20-2-11-17-
NO 
Inconclusive 
outcom
e
 
 
4-45 
11.00 
0.0755 
1-2-11-12 
Contiguous 
range 
expansion 
 
4-56 
11.00 
0.0457 
1-19-20-2-3-4-9 
Allopatric fragm
entation 
 
5-1 
40.54 
0.0082 
 
No significant clade distances 
 
5-2 
22.00 
0.0023 
 
No significant clade distances 
 
Total Clade 
37.59 
0.0015 
 
No significant clade distances 
144
 
 
 
 
 
 
     Model III 
2-21 
23.25 
0.0542 
1-19-20-2-11-17-4 
RGF with IBD 
 
3-22 
14.00 
0.0849 
1-2-11-17 
Inconclusive 
outcom
e
 
 
4-13 
21.00 
0.0292 
1-19-20-2-3-5-15 
PF and/or LDC 
 
4-24 
17.33 
0.2960 
1-19-20-2-11-17-
NO 
Inconclusive 
outcom
e
 
 
4-45 
11.00 
0.0794 
1-2-11-12 
Contiguous 
range 
expansion 
 
4-56 
11.00 
0.0441 
1-19-20-2-3-4-9 
Allopatric isolation 
 
5-2 
49.00 
0.0001 
1-2-3-4-9-N
O
 
No 
significant clade distances 
 
Total Clade
35.65 
0.0039 
 
No significant clade distances 
 
145
 
146
Table 4  Measures of diversity for clades estimated by maximum likelihood using the 
combined cytochrome-b and D-loop regions.  Measurements include S = the number of 
variable sites, Eta = the number of mutations, Hd = haplotype diversity, ??=?nucleotide?
diversity,?k?=?average number of nucleotide differences,?h?=?the number of haplotypes.  
Clade beach* was without haplotypes LCAL01 and TCGA01.  Clade complete* was 
without haplotypes from the Florida peninsula. 
 
Clade N S Eta Hd 
? 
k h Theta 
         
Beach         
Total: 23 57 59 0.963 ?0.02 0.0050 ?0.0004 12.28 15 0.0064 
Cyt-b: 23 19 19 0.950 ?0.02 0.0033 ?0.0004  3.70 13 0.0044 
D-loop: 23 34 36 0.953 ?0.02 0.0081 ?0.0006  8.05 13 0.0096 
         
Inland I         
Total: 12 34 34 0.955 ?0.05 0.0036 ?0.0008  8.73 9 0.0046 
Cyt-b: 12 16 16 0.939 ?0.05 0.0034 ?0.0010  3.84 8 0.0047 
D-loop: 12 14 14 0.924 ?0.06 0.0042 ?0.0009  4.09 8 0.0048 
         
Inland II         
Total: 11 23 23 0.855 ?0.09 0.0026 ?0.0008  6.47 6 0.0032 
Cyt-b: 11  5  5 0.600 ?0.15 0.0012 ?0.0004  1.31 4 0.0015 
D-loop: 11 17 17 0.818 ?0.08 0.0051 ?0.0017  4.20 5 0.0060 
         
Beach*         
Total: 22 42 44 0.957 ?0.02 0.0044 ?0.0003 10.74 13 0.0049 
Cyt-b: 22 12 12 0.939 ?0.02 0.0026 ?0.0002  2.98 11 0.0029 
D-loop: 22 26 28 0.944 ?0.02 0.0074 ?0.0005  7.16 11 0.0079 
         
Complete         
Total: 55 122 125 0.984 ?0.01 0.0074 ?0.0002 18.12 35 0.0112 
Cyt-b: 55 45 45 0.969 ?0.01 0.0052 ?0.0003  5.91 28 0.0087 
D-loop: 55 67 70 0.979 ?0.01 0.0120 ?0.0005 11.60 31 0.0158 
         
Complete*         
Total: 48 103 106 0.980 ?0.01 0.0070 ?0.0002 17.20 30 0.0098 
Cyt-b: 48 38 38 0.964 ?0.01 0.0051 ?0.0003  5.83 25 0.0075 
D-loop: 48 57 60 0.974 ?0.01 0.0112 ?0.0004 10.84 26 0.0139 
 
        
 
        
 
        
Table 5  Measures of diversity for 5
th
 level clades estimated by the model I nested clade 
analysis. Measurements include S = the number of variable sites, Eta = the number of 
mutations, Hd = haplotype diversity, ??=?nucleotide diversity, k = average number of 
nucleotide differences, h = the number of haplotypes. 
 
Clade N S Eta Hd 
? 
k h Theta 
         
Clade 5-1         
Total: 23 63 63 0.957 ?0.03 0.0059 ?0.0004 14.47 15 0.0070 
Cyt-b: 23 23 23 0.897 ?0.05 0.0044 ?0.0005 4.98 12 0.0055 
D-loop: 23 35 35 0.941 ?0.03 0.0093 ?0.0007 8.98 13 0.0098 
         
Clade 5-2         
Total: 10 46 46 0.933 ?0.06 0.0070 ?0.0006 17.04 7 0.0066 
Cyt-b: 10 15 15 0.867 ?0.07 0.0048 ?0.0006 5.42 5 0.0047 
D-loop: 10 29 29 0.933 ?0.06 0.0112 ?0.0011 10.91 7 0.0106 
         
Clade 5-3         
Total: 22 42 44 0.957 ?0.02 0.0044 ?0.0003 10.74 13 0.0049 
Cyt-b: 22 12 12 0.939 ?0.02 0.0026 ?0.0002 2.98 11 0.0029 
D-loop: 22 20 28 0.944 ?0.02 0.0074 ?0.0005 7.16 11 0.0079 
 
 
147
Table 6  Measures of diversity for 5
th
 level clades estimated by the model II nested clade 
analysis. Measurements include S = the number of variable sites, Eta = the number of 
mutations, Hd = haplotype diversity, ??=?nucleotide diversity, k = average number of 
nucleotide differences, h = the number of haplotypes. 
 
Clade N S Eta Hd 
? 
    k h Theta 
         
Clade 5-1         
Total: 26 75 75 0.963 ?0.02 0.0065 ?0.0004  15.79 17 0.0080 
Cyt-b: 26 28 28 0.917 ?0.04 0.0048 ?0.0005    5.50 14 0.0065 
D-loop: 26 42 42 0.951 ?0.02 0.0101 ?0.0007    9.84 15 0.0113 
         
Clade 5-2         
Total: 22 42 44 0.957 ?0.02 0.0044 ?0.0003  10.74 13 0.0049 
Cyt-b: 22 12 12 0.939 ?0.02 0.0026 ?0.0002    2.98 11 0.0029 
D-loop: 22 26 28 0.944 ?0.02 0.0074 ?0.0005    7.16 11 0.0079 
 
 
148
Table 7  Measures of diversity for 5
th
 level clades estimated by the model III nested clade 
analysis. Measurements include S = the number of variable sites, Eta = the number of 
mutations, Hd = haplotype diversity, ??=?nucleotide diversity, k = average number of 
nucleotide differences, h = the number of haplotypes. 
 
Clade N S Eta Hd 
? 
    k h Theta 
         
Clade 5-1         
Total: 23 63 63 0.957 ?0.03 0.0059 ?0.0004  14.47 15 0.0070 
Cyt-b: 23 23 23 0.897 ?0.05 0.0044 ?0.0005    4.98 12 0.0055 
D-loop: 23 36 36 0.941?0.03 0.0092 ?0.0007    9.15 13 0.0098 
         
Clade 5-2         
Total: 25 57 59 0.963 ?0.02 0.0050 ?0.0004  12.28 15 0.0064 
Cyt-b: 25 19 19 0.950 ?0.02 0.0033 ?0.0004    3.70 13 0.0044 
D-loop: 25 34 36 0.953?0.02 0.0081 ?0.0006    8.05 13 0.0096 
 
 
149
Table 8  Measures of diversity for clades estimated by maximum likelihood using the 
combined cytochrome-b and D-loop regions.  Measurements include S = the number of 
variable sites, Eta = the number of mutations, Hd = haplotype diversity, ??=?nucleotide?
diversity,?k?=?average number of nucleotide differences,?h?=?the number of haplotypes.  
Clade beach* was without haplotypes LCAL01 and TCGA01.  Clade complete* was 
without haplotypes from the Florida peninsula.  For Fu and Li?s D and F tests all P 
values were > 0.10. 
 
Clade Fu and Li's D Fu and Li's F McDonald-Kreitman 
 
Complete    
 
Total: -1.8272 -1.9972 
 
Cyt-b: -1.4629 -1.8123 P-value: 0.0494 
 
D-loop: -1.3800 -1.4748  
 
   
 
Beach  
 
Total: -1.4779 -1.6929 
 
Cyt-b: -1.3924 -1.6934 P-value: 0.0585 
 
D-loop: -1.1325 -1.2717  
 
   
 
Inland I    
 
Total: -0.6456 -0.9265 
 
Cyt-b: -0.5636 -0.9037 P-value: 0.1956 
 
D-loop: -0.5040 -0.6088  
 
   
 
Inland II    
 
Total: -0.2131 -0.4770 
 
Cyt-b: -0.5795 -0.9078 P-value: 0.3909 
 
D-loop: 0.1495 -0.0561  
 
   
 
Beach*  
 
Total: -1.0777 -1.1911 
 
Cyt-b: -1.0223 -1.1669 P-value:0.0771 
 
D-loop: -0.6596 -0.7183  
 
   
 
Complete*  
 
Total: -1.6520 -1.7600 
 
Cyt-b: -1.5003 -1.7170 P-value: 0.0460 
 
D-loop: -1.2056 -1.2479  
 
   
 
 
 
150
Table 9  Fu and Li?s D and F tests for Peromyscus polionotus  
clades or groups with other Peromyscus polionotus as outgroup  
haplotypes.  Clade beach* was without haplotypes LCAL01 
and TCGA01.  Significance is indicated by *. 
 
Clade      Fu and Li's D Fu and Li's F
Beach   
(inland I)   
Total:      -1.7239              -1.8771
Cyt-b:      -2.0283              -2.2241
D-loop:      -1.1532              -1.2666
  
Beach*  
(inland I)  
Total:      -1.3985              -1.4496
Cyt-b:      -1.4650              -1.5427
D-loop:      -0.9416              -0.9486
  
Beach  
(inland II)  
Total:      -1.4800           -1.6720 
Cyt-b:        -2.3462* -2.4894*
D-loop:      -0.5711            -0.7845 
  
Beach*  
(inland II)  
Total:      -0.9366              -1.0633
Cyt-b:      -1.4650              -1.5427
D-loop:      -0.2361              -0.3557
   
Inland I   
(inland II)   
Total:      -1.1633              -1.3696
Cyt-b:      -1.2261              -1.4749
D-loop:      -0.5040              -0.6088
 
 
151
Table 10  Fu and Li?s D* and F* and Tajima?s D tests for each clade or group.   
Within each clade or group the complete sequence and each major gene were  
tested.  Clade beach* was without haplotypes LCAL01 and TCGA01.  Clade  
complete* is without Florida peninsula haplotypes.  All P values were > 0.05. 
 
Clade Singletons Fu and Li's D* Fu and Li's F* Tajima's D 
Beach     
Total: 32 -1.3361 -1.4797 -1.1319 
Cyt-b: 13 -1.8340 -1.9784 -1.4039 
D-loop: 16 -0.8193 -0.9389 -0.7910 
     
Beach* 
Total: 22 -0.9734 -1.0399 -0.7199 
Cyt-b: 7 -1.1252 -102023 -0.8395 
D-loop: 12 -0.6400 -0.6649 -0.4155 
     
Inland I     
Total: 18 -0.7489 -0.9336 -1.0174 
Cyt-b: 9 -0.8461 -1.0655 -1.1938 
D-loop: 6 -0.2790 -0.3830 -0.5016 
     
Inland II     
Total: 13 -0.7920 -0.9010 -0.8031 
Cyt-b: 3 -0.7638 -0.8976 -0.8933 
D-loop: 9 -0.6271 -0.7125 -0.6345 
     
Complete 
Total: 55 -1.9113 -2.0062 -1.2807 
Cyt-b: 22 -2.0651 -2.2103 -1.4799 
D-loop: 88 -1.7169 -2.0280 -0.8759 
     
Complete*     
Total: 46 -1.7262 -1.7634 -1.0372 
Cyt-b: 19 -1.9855 -2.0343 -1.1998 
D-loop: 21 -1.0223 -1.0629 -0.6566 
 
 
152
Table 11
  Mitochondrial DNA poly
m
or
phism
 within different 
Peromyscus
 
polionotus
 clades.  Coalescen
t based m
a
xim
u
m  
likelihood estim
ates of N
e
, theta (
?
), and exponential growth rate 
(95% C
I) under m
odels of constant size and growth as  
im
ple
m
ented by Kuhner 
et al
. (1998).  Clade beach* was wi
thout haplotypes LCAL01 and TCGA01.  Clade com
plete*  
was without Florida Pe
ninsula haplotypes. 
  
            Beach
           Beach*
         Inland I
        Inland II
       Com
p
lete
     Com
p
lete*
 
 
Sa
m
ple Size 
17
15
12
11
44
40
 
 
Base pairs 
2449
2449
2449
2449
2449
2449
 
 
Haplotypes 
15
13
9
6
34
30
 
 
Polym
orphic sites 
57
42
34
23
122
103
 
 
?
M
L (constant size) 
0.0138
0.0099
0.0059
0.0030
0.0292
0.0226
 
 
Ne (constant size) 
1256906
899631
533476
269146
2657062
2051542
 
 
?
M
L (growth) 
0.0481 
?0.0115
0.0311 
?0.0081
0.0091 
?0.0022
0.0029 
?0.0007
0.0563 
?0.0069
0.0433 
?0.0058
 
 
Ne (growth) 
4 376 909
2 830 090
826 454
259 272
5 120 272
3 939 818
 
 
Growth rate 
1066.39 ?172.95
1212.47 ?230.51
469.13
?173.58
-30.24
?206.18
524.24 ?68.65
578.52 ?83.65
 
 
Substitu
tion rate (?) 
1.1 X 10-8
1.1 X 10-
8
1.1 X 10-8
1.1 X 10-8
1.1 X 10-8
1.1 X 10-8
 
 
Models agree 
     No 
     No 
     Yes 
     Yes 
     No 
     No 
 
 
?
M
L (constant) 
         0.008 -           0.0251 
       0.0055 -         0.0188 
       0.0029 -         0.0126 
       0.0014 -         0.0069 
       0.0209 -         0.0419 
       0.0157 -         0.0331 
 
 
 
153
  
 154  
Chapter Two 
 
Spatiotemporal Dynamics of Gene Diversity in the  
Alabama Beach Mouse (Peromyscus polionotus ammobates) 
 
Introduction 
The presence of genetic variation is an essential prerequisite for the evolutionary 
process.  In order for any species to evolve, it must both accrue and maintain variation.   
In natural systems, this requirement represents an obstacle to evolution.  Two key 
mechanisms that drive the evolutionary process, natural selection and genetic drift, both 
generally result in the removal of allelic variants.  On an evolutionary timescale, these 
losses can be partially balanced by mutation, the ultimate source of all new variation.  
But, at best, the accumulation of variation via mutation is a slow process that contributes 
little to the genomic variance present at any given moment.  In fact, as demonstrated by 
Fisher (1930), the ultimate fate of 99% of all new mutations is eventual loss from the 
gene pool.  Clearly under these constraints, the evolution of mechanisms for genetic 
management is, for most species, as critical to long-term evolutionary success as are the 
classical studied markers of reproductive success.  What avenues might be open to a 
species for gene pool ?management?, given that the goal is to slow the rate of decay of 
  
 155
variation?  If we view genetic variation as a resource that a species is attempting to 
protect, then biology provides us with one immediate answer.  The most common 
approach used by any biological entity to conserve a resource, be it a community / a guild 
/ a population, or a single cell, is to partition that resource.  Isolation of the resource into 
separate subunits for management, while not necessarily the best approach in some cases, 
is clearly a natural response to many biological problems.  The management of gene pool 
diversity appears to adhere to this rule.  For example, the partitioning of the genetic 
blueprint into separate chromosomes by the earliest cellular forms eventually led to 
meiosis, recombination, and sexual reproduction; mechanisms which collectively work as 
one of the most effective variation generators known.  At every level, this pattern 
continues.  Many organisms partition their gene pools into male and female units, where 
environmental and physiological effects can be managed separately.  These separations 
also provide the basis for higher-level units that are formed by organisms such as kin 
groups, breeding groups, pods, harems, coteries, etc, each representing a viable, 
additional subdivision of the gene pool (Dobson 1998).  Following from these 
observations, organisms should be expected to produce and maintain systems that 
partition genotypes.  As do many others, I view these systems as ?traits?, subject to 
modification and refinement by the evolutionary process. 
Contemporary concepts of genetic structuring through partitioning of genotypes 
are direct descendents of the theoretical framework established by Sewall Wright (1931, 
1951, 1969, 1978).  Wright recognized and promoted the importance of population 
subdivision.  Theoretically, in large panmictic populations, the primary determinant of 
genetic variation is selection.  However, as populations become smaller, such as through 
  
 156
partitioning or otherwise depart from panmixis, other factors can have a significant effect 
on genetic variation (Wright 1978).  Primary of these is genetic drift, which, along with 
the related mechanisms of inbreeding, demographic bottlenecks, and founder events, 
tends to reduce genetic variation.  As a legacy from Wright, population models have 
proven to be crucial in the study of evolution and population genetics.  A principle 
question arising among these models is whether a population functions as a single entity 
moving along one evolutionary path, or as an amalgamation comprised of multiple 
entities exploring different evolutionary pathways.  Understanding how populations are 
genetically structured is a vital step towards understanding and predicting the forces that 
shape populations of organisms (Wright 1931). 
As the theoretical base for population models grew, the importance of spatial 
variance in migration among subpopulations and its genetic consequences were soon 
realized (Giles & Goudet 1997).  This knowledge led to the development of a set of 
dispersal-based models (Hamrick & Nason 1996).  One, the continent-island model, 
represents what is essentially one-way migration.  This model is structured as a single 
large population (continent) and a small population (island) where migration occurs from 
the continent to the island.  Dependent upon the amount of migration over-time, the 
island population is predicted to assume the allele frequency of the continent population.  
Because of their value to population genetics, other models followed, these include 
Wright?s (1931) island model, the stepping-stone model (Kimura 1955; Kimura & Weiss 
1964), and the isolation-by-distance model (Wright 1943).  These models and their 
descendants vary in complexity, represent different types of population structure and 
begin to account for the possibility of nonrandom mating (Gile & Goudet 1997).  
  
 157
Wright?s island model assumes that all populations are equal in size and contribute 
equally to the migrant pool.  Spatially, each population is effectively equally close to all 
other populations; thus, each population receives a proportionate number of migrants.  
This model moves the complex of populations towards an equilibrium frequency.   
Kimura (1955, 1964) proposed the stepping-stone model to incorporate the effects 
of spatial variance in migration rate.  It assumes that migration is restricted to adjacent 
populations.  Thus, a new allele or mutation could move to an adjacent population in one 
generation and move to a second population in the next generation.  This model creates 
the greatest amount of genetic differentiation among the subpopulations.  While the 
previous two models are composed of subpopulations that are panmictic, Wright 
developed the isolation-by-distance model (IBD) to account for population assemblages 
in which mating is nonrandom (Hamrick & Nason 1996).  In the IBD model, migration 
occurs among local communities in a continuously distributed population.  The IBD 
model is intermediate to the stepping stone model and the island model in producing 
levels of within population variation (Hamrick & Nason 1996).  The three models 
discussed above do share important assumptions.  Random mating is assumed to occur in 
each subpopulation or community and effective population size is expected to remain 
constant. Typically the models also assume that no selection or mutation occurs.  Lastly, 
the basic models describe diploid, sexually reproducing organisms with nonoverlapping 
generations.  
Dispersal models have continued to evolve and to incorporate more realistic 
assumptions.  Wright (1940) suggested that extinctions of subpopulations might increase 
between population variation.  Following this logic, Levins (1969), developed the 
  
 158
concept of a metapopulation.  Within the metapopulation model, space is discrete and 
composed of suitable and unsuitable habitat.  Extinction and recolonization on patches of 
suitable habitat is an essential component of metapopulation theory.  Recolonization of 
patches is important in the overall genetic structure of the populations.  The effects of 
recolonization on genetic structure depend on whether all populations or a single 
population contributed migrants.  Slatkin (1977) modified Levins basic model to account 
for variation in patch size, variance of migrants between occupied patches, and to 
acknowledge that vacant patches may or may not be immediately recolonized (Giles & 
Goudet 1997).  While metapopulation theory has evolved dramatically since its inception, 
its primary focus continues to be between subpopulation dynamics as driven by 
migration.  Regardless of the assumptions of each model, they are all simplifications of 
spatial genetic structure. 
Traditionally, empirical evaluations of spatial partitioning have focused on 
situations where clear patterns of partitioning existed (e. g. environmental barriers and 
mating systems).  More recently, however, techniques such as spatial autocorrelation 
(Sokal & Oden 1978a, b) and Mantel?s Test (Smouse et al. 1986) along with population 
assignment methods (Paetkau et al. 1995; Waser & Strobeck 1998; Banks & Eichert 
2000) are increasingly being applied to hypervariable microsatellite data sets.  Thus, 
detection of microgeographic genetic structuring along with assignment of individuals to 
their original population is now quite possible.  Using autocorrelation analysis, 
populations characterized by limited dispersal and the lack of selection will demonstrate 
positive genetic association within smaller distance classes before becoming negative as 
distance classes increase (Sokal & Wartenberg 1983; Smouse & Peakall 1999).  
  
 159
Populations that are subdivided into microgeographic groups might be characterized by 
increased inbreeding and homozygosity and distinguished by microgeographic 
assemblages of like genotypes (Turner et al. 1982).  However, little empirical work has 
focused on potential breeding units within a population that is not distinctly subdivided 
by a habitat matrix.  Arguably, within contiguous populations, semi-independent 
breeding units may occur that are critical to the gene dynamics of populations.  
Nonrandom mating and individual dispersal events would be the major factors affecting 
the genetic structuring within such a population. 
Theoretically, subdivision of a population can impact the maintenance of genetic 
diversity over time (Chesser et al. 1980; Karlin & Campbell 1980).  Genetic partitioning 
is also hypothesized to increase the ability of a species to adapt to environmental changes 
more quickly (Wright 1969, 1978).  Thus, species should be expected to evolve 
mechanisms to partition genotypes to counter factors that can act to reduce overall 
variation.  According to Waser (1993), populations are genetically structured by the 
coevolution of mating systems, dispersal behavior, mate choice, and the fitness effects of 
mating.  Studies have demonstrated that both spatial segregation and breeding structure 
can be important causal agents of genotype partitioning (Chesser 1991a; Chesser 1998; 
Gerlach & Musolf 2000; Tiedemann et al. 2000).  However, many populations exist in 
more contiguous habitats than are normally considered in genetical structure studies.   
The premise underlying my research is that within contiguous habitats, 
microgeographic structure can mimic higher-order breeding systems (coteries, harems, 
etc.,) such that alleles are partitioned and genetic diversity is maintained at higher levels 
than would otherwise be expected.  Most mammal species exhibit some type of 
  
 160
population structuring (Greenwood 1980).  This has been demonstrated for house mice, 
Mus musculus, (Selander 1970a) deer mice, Peromyscus maniculatus, (Wright 1978) 
pocket gophers, Thomomys bottae, (Patton & Yang 1977) prairie dogs, Cynomys 
ludovicianus, (Chesser 1983) and the Australian bush rat, Rattus fuscipes, (Peakall et al. 
2003).  The reoccurrence of partitioning among species suggests that there are likely to 
be multiple mechanisms (dispersal patterns, single sex clusters, etc.) and, thus, multiple 
scenarios that favor the formation of such elements.  One model (Isolation by Distance) 
invoked to explain such formations is characterized by short dispersal differences and 
nonrandom mating inferring some level of inbreeding may result (Wright 1943).  
Mammals, in particular, because of low reproductive rates, may benefit from maintaining 
high degrees of relatedness within social groups (Chesser 1998).  Inbreeding may serve to 
increase the proportion of an individuals genome that is passed on to its offspring, and, if 
dispersal costs are high, it is potentially beneficial for related individuals to mate 
(Bengtsson 1978).  Thus, low levels of inbreeding may be less costly than dispersal.  
Also, gene combinations that are selectively advantageous may increase in frequency 
more quickly in small inbreeding groups rather than in larger randomly mating 
populations (Slatkin 1976).  
Historically, allozymes have been used to investigate genetic structuring within a 
population.  However, allozymes are probably best suited for detection of structure on a 
larger scale or where some discernable form of isolation among populations is present.  
As might be expected, some of these studies have detected the presence of genetic 
structure on a local scale while others have not.  Waser & Elliott (1991) failed to detect 
local structure within a population of bannertailed kangaroo rats (Dipodomys spectabilis).  
  
 161
However, van Staaden et al. (1996) demonstrated substantial structure within a 
population of Richardson?s ground squirrels (Spermophilus richardsonii).  A behavioral 
characteristic of Richardson?s ground squirrels is philopatry by females.  The clustering 
of related females strongly suggests that genetic structure will be present.  While there 
are multiple means by which species might develop structure (behavioral, social, 
dispersal) rarely have populations been temporally examined to determine if and how 
structure is influenced temporally.  However, Scribner & Chesser (1993) investigated 
genetic structuring in the eastern cottontail (Sylvilagus floridanus)and demonstrated 
significant nonrandom genetic structuring across multiple time periods.  They also 
observed a cyclic temporal pattern where during some time periods autocorrelation was 
less evident.  Importantly, spatial autocorrelation of genotypes was less evident during 
the winter season.  This result was attributed to the winter season being the 
prereproductive season.  While genetic structure has been demonstrated on a local scale, 
microgeographic structuring was difficult to assess.  Recent application of large numbers 
of hypervariable microsatellite loci as markers in ecological studies have substantially 
increased the precision by which questions concerning spatial structure can be addressed.  
Microsatellite markers, found abundantly within the genome and possessing a high 
mutation rate, are now being applied to the investigation of microgeographic genetic 
structuring.  Microsatellites have been used to investigate spatial structure at a fine-scale 
with some success in mammals, birds, snakes and fire ants, Solenopsis invicta, (Peakall et 
al. 2003; Albrecht et al. 2001; Double et al. 2000; Gibbs et al. 1997; Ross et al. 1997).  
Also, Double et al. (2005) used microsatellites to show sex-biased dispersal and positive 
spatial structure among males within a population of the superb fairy-wren (Malurus 
  
 162
cyaneus).  Increased detection of genetic structuring within populations is beginning to 
provide key insights into such evolutionary factors as dispersal patterns.  In particular, 
this new wave of data is fostering empirical examination of long-standing theories 
regarding the partitioning of genotypes as a means of preserving or slowing the rate of 
loss of global genetic variance (Chesser et al. 1980; Karlin & Campbell 1980).  Rohlf & 
Schnell (1971) used a single-locus two-allele model and showed local genetic 
differentiation can occur in a contiguous population.  Models of noncontiguous 
populations have also yielded evidence for this critical element of spatial structuring 
(Maruyama 1970; Christensen 1974; Slatkin 1985). 
My study animal, beach mice (Peromyscus polionotus ssp.) has not been reported 
to be territorial, form single sex clusters, or exhibit other characteristics that would 
normally drive behavioral partitioning of genotypes.  All evidence to date indicates that 
beach mice are monogamous (Foltz 1981; Swilling & Wooten 2002).  Monogamy may 
produce varying effects on population structure because it tends to limit reproductive 
variation among individuals.  Contributing to their uniqueness are the habitat associations 
of beach mice.  Along the Gulf Coast of Alabama, beachfront habitat is long, linear, and 
seemingly contained few areas that beach mice did not inhabit.  Until recent decades, 
beach mouse populations appear to have existed in this semicontiguous habitat where 
spatial separation (i.e. a course habitat matrix) was not the defining characteristic.  Thus, 
geographic boundaries would not seem to be the foremost agent in fine-scale genetic 
structuring within local beach mouse populations.  Temporal variance is however a 
significant element among beach mouse populations as they experience pronounced 
demographic annual fluctuations that are compounded by cyclic patterns of tropical 
  
 163
storms and hurricanes that can reduce population size for several years.  Such fluctuations 
in population size are known to have significant effects on genetic variation (Zhang et al. 
2004).  The recovery period following a hurricane Opal for the segment of the beach 
mouse population representing my study site was approximately four years.  Theory 
predicts that the amount of genetic variation remaining in the population following a 
bottleneck is a function of the severity of the bottleneck and the rate of demographic 
change (Nei et al. 1975).  Compounding the annual cycle, is the fact that a major 
hurricane affects the area, on average, every 10 years.  Thus, bottlenecks arise through an 
annual cycle as well as through periodic hurricanes.  This combination of conditions (i. e. 
cyclic bottleneck, behavioral tendencies, semicontiguous habitat) where each tends to 
elevate the rate of genetic decay gave rise to my central question, namely, what options 
are available to populations of beach mice to aid in mitigating the loss of genetic 
variance?  In fact, it is due to such conditions that I believe beach mice (Peromyscus 
polionotus ssp.) represent an appropriate model organism for use in examining questions 
concerning genetic partitioning and the rate of loss of genetic variance.  If partitioning 
occurs in a ?system? such as beach mice it indicates that a level of partitioning other than 
that defined by spatial boundaries and mating systems (i.e. single sex clusters) can 
develop. 
It was my supposition that the dynamics of gene diversity within beach mice 
populations are not panmictic or metapopulation but do share characteristics with each 
model.  Within these insular populations experiencing regular and repeated factors that 
tend to reduce genetic variation, temporal partitioning of genotypes may occur if groups 
are formed by low dispersal rates, distances and some toleration of inbreeding.  Under 
  
 164
these conditions, one might predict that, due to chance alone, individual familial lineages 
would be established across the habitat.  Within any local region of a beach mouse 
population, structural population units are likely to be comprised of closely related 
individuals.  Genetic structuring would be maintained through social dynamics, dispersal 
patterns, and, to a lesser degree, broken habitat.  Within this model, local breeding 
clusters are familial and exist as smaller, interconnected units within the population.  
Genetic partitioning exists due to the temporal formation of familial breeding units on a 
microgeographic scale.  An excess of homozygous individuals would be expected to 
characterize these familial lineages.  Periodically, catastrophic events (e. g. hurricanes, 
tropical storms) increase mortality thus breaking down familial lineages and forcing 
individuals to seek new mating opportunities.  If genotype partitioning exists, shortly 
following such events as hurricanes individual heterozygosity should increase, however 
over time familial lineages will reform with the genetic structuring again being 
characterized by an excess of homozygotes.  The increase in heterozygosity likely 
represents ?isolate breaking?.  Spatial autocorrelation should be significantly positive 
prior to spikes in heterozygosity but significantly positive during spikes in 
heterozygosity.  Wooten and Holler (1998) observed such an increase in heterozygosity 
following hurricane Opal within this population of Alabama beach mice.  Their 
observation was based on three loci and while the pattern appeared to be real, adding loci 
would provide a more definitive test. 
Here, I investigated not only genotype partitioning, but also the potential for this 
hypothesized system of genotype partitioning to be a significant component of the 
evolutionary process.  Specifically, I examined how gene pool management occurs in a 
  
 165
natural population that is subject to multiple, severe events known to increase genetic 
decay.  An important twist to this research is that my model species, Peromyscus 
polionotus ammobates, operates in a wildly unstable environment while maintaining a 
monogamous mating system, and no traditional breeding structures, within a locally 
contiguous habitat.  First, I investigated predicted microgeographic spatiotemporal 
partitioning of genotypes based primarily on social dynamics (monogamy, increased 
inbreeding levels) and individual dispersal patterns exhibited by P. polionotus.  Second, I 
investigated the effect of spatiotemporal gene partitioning on the rate of loss of genetic 
variation.  My overall goal was to test the hypothesis that microgeographic spatial 
structuring alone could reduce the observed rate of genetic decay relative to a predicted 
rate.  More globally, it is my contention that such structure plays a critical role as the core 
modulator of gene pool composition. 
Methods 
Tissue Collection 
 Tissue samples were obtained from mice captured on two permanent grids. 
Established in fall of 1994, the grids consisted of 25 stations set at 20 m intervals east to 
west and 12-13 stations at 20 m intervals extending south to north (Fig. 1).  The two grids 
(Gazebo and Vet Village) began in the foreslope of the primary dunes and extended north 
into scrub habitat and were separated east and west by approximately 620 m (Swilling et 
al. 1998).  Two Sherman live traps baited with rolled oats were placed at each station.  
Captured mice were weighed, sexed, aged (adult or subadult), and reproductive condition 
determined (scrotal or nonscrotal for males; not pregnant, pregnant, lactating, or pregnant 
and lactating for females).  Each mouse was assigned a unique number for individual 
  
 166
identification.  Mice were released at the capture site immediately after data collection.  
Traps were opened in the late afternoon and checked for captures prior to dawn unless 
temperatures were expected to drop below 13
o 
C.  All traps were left closed after predawn 
checking.  Cotton was used inside traps to provide insulation during cool nights.  
Trapping periods were designed to run for five consecutive nights.  However, trapping 
was stopped for nights of rain and/or temperatures deemed to be too cold.  Interrupted 
trapping periods were resumed after passing of inclement weather.   
For years 1995, and 1996 and half the year of 1997, the grids were trapped 
approximately bimonthly.  For 1995 and 1996, I used tissue collected during February, 
April and June on Gazebo grid and March, April and June on Vet Village grid.  Only a 
single trapping period for both grids from 1997 (March) was examined.  In fall of 1997, 
both grids were reduced in size.  This modification consisted of removing seven 
north/south lines from the western section of Gazebo grid and seven from the eastern 
section of Vet Village grid.  The result was additional spacing of approximately 280 m 
between the two grids, extending the total separation to approximately 920 m (Fig. 2).  
Thus, all trapping periods from late 1997-2002 were conducted on smaller grids spaced 
farther apart.  Only spring trapping periods from 1998 and 1999 were examined.  For 
2000 and 2001, both a spring and fall trapping period were available, and, from 2002, a 
single spring trapping period was available for analysis.  The final data set represented 14 
trapping periods spanning eight years (1995-2002). 
DNA preparation 
 DNA was extracted from tissue samples using Qiagen DNeasy? tissue kits.  The 
Polymerase Chain Reaction (PCR) method was used to amplify specific regions of DNA 
  
 167
to determine the nucleotide sequence and/or fragment length.  For PCR reactions, Qiagen 
Taq PCR Core Kit and Promega PCR Nucleotide Mix were used.  Reactions were run 
with a Perkin Elmer GeneAmp PCR System 2400, a Hybaid Omn-E HBTRE02, or a MJ 
Research PTC-200.  Nucleotide sequence determination and fragment analysis was 
completed using an ABI Prism? 3100 Genetic Analyzer.  GENESCAN? analysis software 
along with GENOTYPER? software was used for fragment scoring.  Several of the 
microsatellites were sequenced to be sure of amplification of the proper alleles.  BIOEDIT 
v5.0.9 (Hall 1999) software was used primarily to manipulate nucleotide sequence data. 
Polymerase chain reactions (PCR) were conducted with 11 primer sets known to 
amplify dinucleotide repeat regions.  These 11 loci were selected after testing 24 primer 
sets.  Some of the initial primer sets did not amplify or generated inconsistent peaks 
where scoring was not reliable.  The 11 loci that amplified and scored consistently were 
Ppa-01, Ppa-12 and Ppa-46 (Wooten et al. 1999), Pml-03, Pml-04, Pml-06 and Pml-11 
(Chirhart et al. 2000) and PO3-85 and PO3-68 (Prince et al. 2002).  Also, two primer 
sets, PO-25 (AF380236) and PO-71 (AF380240), were obtained directly from NCBI 
(Direct submission; Prince et al. 2001).  Forward primers contained fluorescent dye-
labeled either 5? 6-FAM?, 5? HEX?, and 5? NED?.  Each primer set consisted of 25 
?L total volume and contained 12.6-54.7 ng DNA, 0.25 uM primer, 0.5 mM MgCl
2
, 100-
uM dNTPs, 1X buffer, 1.0 U of Taq.  PCR reactions were conducted using Taq PCR 
Core Kit (Qiagen
?
, Valencia, California) and PCR Nucleotide Mix (Promega
?
, Madison, 
Wisconsin).  Characterization of the 11 microsatellite loci is reported in Table 1.  
Conformation of PCR product was determined by running electrophoresis on a 2% 
agarose gel at 75 milli amps and viewing under UV light.  Fragment analysis was 
  
 168
conducted using an ABI 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA) 
located at the Auburn University Genomics and Sequencing Lab (Auburn University).  A 
total of 1085 individuals were examined at 11 loci resulting in 11 935 individual 
genotypes. 
Statistical Analyses 
Data from each trapping period for Gazebo and Vet Village grids were analyzed 
independently.  Also, trapping periods from each grid covering approximately the same 
time period were combined and analyzed as a single population.  Exact tests for 
deviations from Hardy-Weinberg equilibrium were performed using GENEPOP on the 
Web (Raymond & Rousset 1995).  The Markov chain method (Guo & Thompson 1992) 
with the parameters dememorization, iterations set at 1000 and batches set at 100 were 
used to estimate P-values.  GENEPOP was also used to test for heterozygote deficiency 
with dememorization and iterations set at 10 000 and batches set at 1000.  Population 
differentiation as estimated by Analysis of molecular variance (AMOVA) (Excoffier et al. 
1992), F
ST
 and R
ST
 values were calculated using the program ARLEQUIN version 2.0 
(Schneider et al. 2000).  Significance tests were based on 1000 permutations.  ARLEQUIN 
performs an AMOVA based on the method of Excoffier et al. (1992).  Differentiation was 
tested between grids, as well as among select trapping periods on individual grids.  These 
models allow for the grids to be split into different subpopulations for analysis of 
differentiation within grids.  Individual grid analyses were conducted by dividing each 
grid into four quadrants reflecting the four directional corners, north/south, and east/west 
(Fig. 1).  Thus, one analysis consisted of four subpopulations (four corners) while the 
other two consisted of two populations (north/south, east/west).  Also, for comparison, 
  
 169
twice I randomly assigned individuals into the same quadrants (four corners, north/south, 
east/west).  Standard population genetic parameters (number of alleles, allele frequency, 
observed and expected heterozygosity, and fixation index) were calculated using the 
software program GENALEX (Peakall & Smouse 2005).  GENALEX was also used to 
investigate spatial genetic structuring and conduct Mantel tests.  For spatial genetic 
structure analysis, the method of Smouse & Peakall (1999) was used.  This approach 
employs a multivariate method that combines alleles and loci rather than analyzing a 
single locus at a time.  Spatial autocorrelation was tested at multiple distance classes 
within and among sites.  To determine significance, the approach of Peakall et al. (2003) 
was used.  Mantel tests are designed to test the correlation between two matrices.  The 
two matrices used in GENALEX are genetic and geographic distance.  For conducting 
Mantel tests GENALEX follows the methods of Smouse & Long (1992) and Smouse et al. 
(1986).  Significance of the Mantel statistics were determined using 999 permutations.  
Population assignment tests were conducted using WHICHRUN version 4.1 (Banks & 
Eichert 2000).  Population assignment was based on the jackknife method.  The jackknife 
method samples an individual while recalculating allele frequencies without the 
previously sampled genotypes and then estimates the most likely source of that 
individual. The computer program POPGENE (version 1.32; University of Alberta, 
Edmonton, Canada) was used to calculate F
IS
 (Wright 1978).  Program SAS (SAS Institute 
2000) was used to compute means, standard errors, and random assignment of 
individuals.  The program MICRO-CHECKER (van Oosterhout et al. 2004) was used to test 
for null alleles (nonamplifying alleles).  MICRO-CHECKER uses the methods of Brookfield 
(1996) and others to determine if null alleles may be present.  Two equations were used 
  
 170
to predict heterozygosity as a function of time.  First, average heterozygosity was 
predicted for selected trapping periods (Gazebo, Feb-96, Feb-97, Apr-98, Feb-99, Mar-
00, Mar-01, Mar-02; Vet Village, Mar-96, Mar-97, Feb-98, Mar-00, Mar-01, Feb-02) 
using equation 13.12 from Nei (1987).  Effective population size was estimated using 
equation 7.22 in Hartl & Clark (1997).  Population estimates were done using the 
jackknife method within program CAPTURE (Otis et al. 1978).  Secondly, I used equation 
7.14 in Hartl & Clark (1997) to estimate the rate of change in heterozygosity for all 
trapping periods.  The upper limit of the 95
th
 CI was used to estimate effective population 
size for equation 7.22 and the rate of change of heterozygosity for equation 7.14. 
Results 
Summary 
Data were obtained during eight years from 14 trapping periods (28 sessions) on 
both Gazebo and Vet Village grids.  A total of 14 112 trap nights were represented with 
1869 captures of 1163 animals.  Tissue was available from 1007 of these individuals 
(Table 2).  Trapping on Gazebo grid resulted in a total of 890 captures of 539 animals 
with tissue being obtained from 477 individuals (Table 3).  Captures per trapping period 
on Gazebo grid ranged from 9 to 145.  Trapping on Vet Village grid produced 979 
captures of 625 animals with tissue being obtained from 530 individuals (Table 4).  
Captures on Vet Village grid ranged from 7 to152 animals over 14 sessions.  
Amplification was attempted for all 1007 individuals at 11 loci.  The number of alleles 
per locus on Gazebo grid ranged from 3 to10 and 3 to 8 on Vet Village grid (Table 5).  
To be conservative, I used predicted heterozygosity from equation 13.12 (Nei 1987) 
because it produced higher estimates than equation 7.14 (Hartl & Clark 1997). 
  
 171
Gazebo Grid 
The March 1999 trapping period on Gazebo grid was not used due unsuccessful 
amplification of primer sets for those individuals.  Tables 6-18 summarize genetic 
diversity information by trapping period for Gazebo grid for each of the 11 loci 
examined, including locus sample size, number of alleles, effective number of alleles per 
locus, average observed heterozygosity, expected average heterozygosity, and the 
fixation index (F
IS
).  The total number of observed alleles on Gazebo grid across all 
trapping periods was 70 (Table 5).  The total number of alleles detected per locus across 
trapping periods ranged from 3 to 10.  The number of alleles observed by trapping period 
on Gazebo grid ranged from 38 to 57.  The highest number of alleles observed (57) was 
during the June 1996 trapping period while the lowest number (38) occurred three times 
(Feb-97, Nov-00, Nov-01).  Typically, for a trapping period when sample size was large 
(number of captures) the observed number of alleles was also high relative to trapping 
periods with low captures.  The average number of alleles per locus by trapping period 
ranged from 3.45 to 5.18.  Allele frequencies by trapping period are presented in 
Appendix 1.  Common alleles tended to persist across trapping periods.  Nine alleles were 
present only in the June 1996 trapping period.  Population pairwise F
ST
 values for 
trapping periods February 1995 and February 1996, April 1995 and April 1996, June 
1995 and June 1996 were not significant, however, F
ST
 for the February 1995 and March 
2002 trapping periods was significant (F
ST
 = 0.0254; P = 0.0000). 
The average observed heterozygosity per trapping period ranged from 0.5187 to 
0.6446 (Table 6-18).  It was highest during the November 2001 trapping period (0.6446) 
and lowest during the March 2000 trapping period (0.5187).  Average observed 
  
 172
heterozygosity oscillated throughout the study but temporally increased following 
hurricane Opal predominately in the June 1996 trapping period (Table 11).  Predicted 
heterozygosity (equation 13.12 Nei 1987) did not increase following the hurricane.  The 
predicted heterozygosity from equation 13.12 (Nei 1987) ranged from 0.5143 during 
March of 2002 to 0.5910 during February 1996 (Fig. 3).  Using equation 7.14 (Hartl & 
Clark 1997) the predicted heterozygosity ranged from 0.4895 in March 2002 to 0.6069 
during February 1995.  The observed heterozygosity plotted as a linear best-fit line (H
O
 
Net) indicates a lower rate of decay than predicted (H
T
) (Fig. 3).  The average expected 
heterozygosity ranged from 0.5441 to 0.6212 where the low occurred in March 2000 and 
the high in February 1996.  Across all trapping periods the mean observed heterozygosity 
was 0.5909 and the mean expected heterozygosity was 0.5896.  The fixation index, F
IS
, 
occasionally indicated that inbreeding and outbreeding were occurring within the grid.  
However, estimates of F
IS
 did not typically coincide with predictions of when inbreeding 
or outbreeding may occur.  Mantel tests were significant for four of the 14 trapping 
periods on Gazebo grid (Table 19) indicating that for these trapping periods (Feb-95, 
Mar-01, Nov-01, Mar-02) there was a positive relationship between genetic 
differentiation and geographic distance.  Exact tests of Hardy-Weinberg expectations 
(HWE) across all trapping periods and loci revealed few significant deviations (0-2 loci 
per trapping period; Appendix 2).  On Gazebo grid, three of the eleven loci were found to 
significantly deviate (5% level) from HWE across 14 trapping periods.  However, these 
loci did not consistently deviate across trapping periods.   
Investigation of possible structuring within the grid produced some interesting 
results (Table 20).  For the February 1995 and March 2002 trapping periods, the four 
  
 173
corner quadrant estimate produced the highest values of F
ST
 (0.0143; 0.0159).  Estimates 
from February 1996 and June 1996 were higher in the east/west (0.0125) and north/south 
(0.0220) analysis respectively.  Only for the June 1995 trapping period was F
ST
 highest 
(0.0219) based on a random assignment of individuals to grids.  While F
ST
 values were 
high considering the proximity of groupings, a particular group (e.g. four corner, etc.) 
was not consistently higher than other groups.  Analysis of molecular variance (AMOVA) 
framework was used to estimate hierarchical analysis of genetic differentiation within 
and among quadrants of the grid for trapping periods February 1995, June 1995, February 
1996, June 1996 and March 2002 and are reflected in the F
ST
 values (Table 20).  The 
north/south sectioning of the grid produced the highest estimate of genetic variance 
(2.20%) within Gazebo grid.  However, the east/west sectioning produced the highest 
estimate for the June 1995 trapping period (0.69%) while the four corner sectioning 
produced the highest estimate for February 1995 (1.43%) and March 2002 (1.59%).  
Significant estimates of F
ST
 from within grid, nonrandom associations were revealed for 
the February 1995, June 1996 and March 2002 trapping periods.  The estimates of genetic 
variance between grids were generally higher than within grids.  
Analysis of spatial genetic autocorrelation was conducted for each trapping period 
using multiple distance class sizes (20 m, 50 m, 100 m, 250 m).  Significant spatial 
autocorrelation was found for eight of fourteen trapping periods and multiple distance 
classes on Gazebo grid (Appendix 3).  However, for distance classes that were significant 
during the February 1996, February 1997, March 2001 (except the 100 m distance class) 
and November 2001 sample size was ? 24.  For six trapping periods significant 
autocorrelation was not found (Jun-95, Apr-96, Jun-96, Apr-98, Feb-99, Nov-00).  
  
 174
Autocorrelation patterns across trapping periods and distance classes were generally not 
consistent.  However, this result was expected for trapping periods following hurricane 
Opal.  Spatial autocorrelation data sets were comprised of males and females.  This 
indicates that proximal individuals were more genetically similar regardless of sex.  
Individual trapping periods, typically at larger distance classes (250 m), were marked by 
an increasing autocorrelation coefficient (r).  However, these particular analyses were 
limited due to small sample sizes. 
Veterans Village Grid 
Tables 6-18 summarize genetic diversity information by trapping period for Vet 
Village grid for each of the 11 loci examined, including locus sample size, number of 
alleles, effective number of alleles per locus, average observed heterozygosity, expected 
average heterozygosity, and the fixation index (F
IS
).  Through all trapping periods the 
total number of alleles on Vet Village grid was 59 (Table 5).  A larger number of samples 
from Vet Village grid (530) produced 11 fewer alleles than 477 samples from Gazebo 
grid.  The total number of alleles detected per locus across trapping periods was 3 to 8.  
The number of alleles found on Vet Village grid per trapping period ranged from 37 to 
53.  The highest number of alleles during a single trapping period on Vet Village grid 
(53) was from April 1995 while the lowest (37) was from December 2000.  This is 
different from Gazebo grid when the most alleles during a trapping period was observed 
in June 1996.  However, a low number of alleles observed during a single trapping period 
occurred at approximately the same time.  Typically, when sample size was low so was 
the number of observed alleles.  However, a larger overall sample size on Vet Village 
grid did not produce a larger number of alleles.  The average number of alleles per 
  
 175
trapping period per locus ranged from 3.36 to 4.82.  Allele frequencies by trapping period 
are presented in Appendix 1.  As expected, common alleles tended to persist across 
trapping periods while rare alleles were not observed when sample size was small.  
Population pairwise F
ST
 values for the trapping periods of March 1995/March 1996 and 
June 1995/June 1996 were not significant, however, F
ST
 for the April 1995/April 1996 
and March 1995/February 2002 trapping periods was significant (FST = 0.0050, P = 
0.0137; F
ST
 = 0.0254, P = 0.0000 respectively). 
The average observed heterozygosity per trapping period on Vet Village grid 
ranged from 0.4496 to 0.6669.  The highest estimate of average observed heterozygosity 
was during the April 1996 trapping period while the lowest occurred in December of 
2000.  Highest and lowest estimates of mean heterozygosity occurred at approximately 
the same time period on both sample grids.  However, even more dramatically than on 
Gazebo grid, observed heterozygosity temporarily increased in the first trapping period 
following hurricane Opal on Vet Village grid (Table 9).  Estimates of predicted 
heterozygosity over time were slightly higher on Vet Village grid than on Gazebo grid.  
A linear trend line of observed heterozygosity (H
O
 Net) indicates a much reduced slope 
versus predicted loss of heterozygosity (H
T
) (Fig. 4).  Using equation 13.12 (Nei 1987) 
predicted heterozygosity ranged from 0.5500 during February of 2002 to 0.6009 during 
March 1996 (Fig. 4).  Equation 7.14 (Hartl & Clark 1997) predicted heterozygosity was 
lowest February 2002 (0.5339) and highest during March 1995 (0.6109).  Observed 
heterozygosity on Vet Village grid was higher than predicted following hurricane Opal 
and was higher than predicted for all but a single trapping period (Dec-00).  The mean 
expected heterozygosity ranged from 0.6420 to 0.5196 and occurred during March 1996 
  
 176
and December 2000 respectively.  Average observed heterozygosity across all trapping 
periods was 0.6084 while average expected heterozygosity was lower at 0.6000.  These 
estimates are slightly higher than estimates for Gazebo grid.  The fixation index F
IS
, 
typically near zero, did not indicate a pattern of higher rates of inbreeding or outbreeding.  
Mantel tests were significant for four (Mar-96, Mar-01, Oct-01, Feb-02) of the 13 
trapping periods on Vet Village grid (Table 19) indicating a periodic relationship between 
genetic distance and geographic distance.  Mantel tests were significant for three of the 
same trapping periods on Vet Village and Gazebo grids indicating that a relationship 
between genetic and geographic distances occurred on both grids simultaneously.  Exact 
tests of Hardy-Weinberg expectations per trapping period for each locus on Vet Village 
grid found significance deviations, similar to Gazebo, ranging from 0 to 2 loci per 
trapping period.  Six of eleven loci deviated significantly from HWE across 13 trapping 
periods (Appendix 2), however, there was no obvious pattern associated with these 
deviations. 
For select trapping periods (Mar-95, Jun-95, Mar-96, Jun-96, Feb-02) the fixation 
index (F
ST
) was estimated for Vet Village grid by sectioning the grid into the four 
directional corners, north/south, and east/west quadrants and treating individuals in each 
quadrant as a subpopulation (Table 20).  Also, for comparison, F
ST
 was estimated by 
twice randomly assigning individuals to these quadrants (Table 20).  For the March 1995 
and March 1996 trapping periods, the four corner estimate produced the highest F
ST
 value 
(0.0247, 0.0369 respectively).  The F
ST
 estimate from the March 2002 trapping period 
was highest in the east/west analysis (0.0231).  The June 1995 and June 1996 estimates of 
F
ST
 were highest based on random assignment of individuals into quadrants (0.0078, 
  
 177
0.0337 respectively).  Similar to Gazebo grid, F
ST
 values were relatively high, but a 
particular quadrant (i. e. four corner, etc.) was not consistently higher than other groups.  
Similar to Gazebo grid, significant estimates of F
ST
 from within grid, nonrandom 
associations were revealed for the March 1995, March 1996 and March 2002 trapping 
periods.  Analysis of molecular variance (AMOVA) framework was used to estimate 
hierarchical analysis of genetic differentiation within and among quadrants of the grid 
and are reflected in the F
ST
 values (Table 20).  The sectioning of the grid into four corners 
produced the highest estimate of genetic variance (3.69%) within Vet Village grid.  
However, the east/west sectioning produced the highest estimates for the June 1996 
(0.70%) and March 2002 (2.31%) trapping periods.  As on Gazebo grid, some of the F
ST
 
estimates were significant.  
Analysis of spatial genetic autocorrelation was conducted for each trapping period 
using multiple distance class sizes (20 m, 50 m, 100 m, 250 m; Appendix 3).  Significant 
patterns were identified.  Significant spatial autocorrelation found during the March 1997 
trapping period for the 20 and 50 m distance classes was most likely the result of small 
sample sizes (?  28).  Spatial autocorrelation was not detected during all trapping periods.  
This outcome was expected based on predictions and spatial autocorrelation results were 
similar to Vet Village grid.  For trapping periods where spatial autocorrelation was 
detected, not all distance classes were significant.  Significant autocorrelation was found 
during seven of the thirteen trapping periods.  Corresponding trapping periods where 
positive significant autocorrelation was found on both grids include six trapping periods 
covering winter and spring 1995, winter 1997, winter and fall 2001 and winter 2002.  
Significant positive autocorrelation was generally found at the same distance classes 
  
 178
between grids for corresponding trapping periods.  Significant positive autocorrelation 
was not detected for all trapping periods and distance classes.  Males and females were 
combined in all autocorrelation analysis indicating that regardless of sex, proximal 
individuals were more genetically similar.  The correlograms, primarily for smaller 
distance classes (20 m to 50 m), show the autocorrelation coefficient (r) oscillating in a 
high to low pattern (Fig. 12-18).  The pattern of larger distance classes (250 m) marked 
by an increasing autocorrelation coefficient (r) was not found on Vet Village grid when 
sample size was sufficient.  Overall, the patterns of autocorrelation found on Vet Village 
grid were strikingly similar to the patterns observed on Gazebo grid. 
Gazebo and Veterans Village combined 
 Trapping periods on Gazebo and Vet Village grids that occurred at 
approximately the same time were combined and analyzed as single trapping periods (e. 
g. Gazebo Feb-95 and Vet Village Mar-95, Gazebo Apr-95 and Vet Village Apr-95, etc.).  
Analyses of these data included exact tests of Hardy-Weinberg, spatial autocorrelation, 
F
ST
, R
ST
, number of migrants (N
M
), AMOVA, Mantel tests, and Population assignment.  
Combining trapping periods from both Gazebo and Vet Village grids into single trapping 
periods resulted in captures ranging from 16 to 329 (Table 2).  Tables 21-27 summarize 
information by trapping period for the combined trapping periods for each of the 11 loci 
examined, locus sample size, number of alleles per locus, effective number of alleles per 
locus, average observed heterozygosity, average expected heterozygosity, and the 
fixation index (F
IS
).  The total number of alleles observed for the global population was 
73 (Table 5).  The genetic variability for the combined data sets was moderate.  The 
number of alleles per data set after combining ranged from 41 to 63.  The high occurred 
  
 179
during the 3
rd
 trapping period of 1996 (June for both grids), with the low in November 
2000 on Gazebo grid and December 2000 on Vet Village grid.  The average number of 
alleles per locus across trapping periods ranged from 3.73 to 5.73.  The number of alleles 
per locus across trapping periods ranged from 3 to 10.  Allele frequencies by trapping 
period are presented in Appendix 1.  Twenty-seven of 73 alleles were observed during 
each trapping period.  Population pairwise F
ST
 values for the combined trapping periods 
of the 1
st
 trapping period of 1995/1
st
 trapping period of 1996 and the 2nd trapping period 
of 1995/2
nd
 trapping period of 1996 were not significant.  F
ST
 values for the 3
rd
 trapping 
period of 1995/3
rd
 trapping period of 1996 and the 1
st
 trapping period of 1995/1
st
 trapping 
period of 2002 were significant (FST = 0.0046, P = 0.0215; F
ST
 = 0.0327, P = 0.0000 
respectively). 
The average observed heterozygosity per trapping period across both grids ranged 
from 0.5014 to 0.6455.  It was highest during the 2
nd
 trapping period of 1996 (April on 
both grids) and lowest during the 2
nd
 2000 trapping period (Gazebo November, Vet 
Village December).  The mean expected heterozygosity ranged from 0.5770 to 0.6388.  It 
was lowest during the 2
nd
 2000 trapping period (Gazebo November, Vet Village 
December) and highest during the 1
st
 1996 trapping period (Gazebo February, Vet 
Village March).  For all trapping periods the average observed heterozygosity was 0.6015 
and the average expected heterozygosity was 0.6085.  F
IS
 values for the combined data 
sets ranged from -0.0268 to 0.1464.  The lowest estimate occurred during the 2
nd
 1996 
trapping period (April for both grids) while the highest estimate occurred during the 2
nd
 
2000 trapping period (Gazebo November, Vet Village December).  Mantel tests were 
significant for 11 of 13 trapping periods examined (Table 28).  Thus, for the majority of 
  
 180
the trapping periods there was a relationship between genetic differentiation and 
geographic distance.  Locus by locus exact tests for Hardy-Weinberg expectations for 
each combined trapping periods found significant deviations (5% level) ranging from 0 to 
3 loci per trapping period (Appendix 2).  For the combined data eight of the eleven loci 
were found to significantly deviate from HWE.  However, there was not a consistent 
pattern of loci or a locus deviating from HWE.  
The fixation index F
ST
 and an analog, R
ST,
 along with N
M
 were calculated for all 
combined trapping periods (Table 29).  F
ST
 values for 10 of 13 trapping periods, while 
small, were significantly different from zero.  R
ST
 values were similar to F
ST
 estimates.  
The number of migrants was also estimated for each combined trapping period and 
ranged from 9.4 to inf.  Both F
ST 
and N
M
 suggest that gene flow among grids varied 
widely.  Also, AMOVA was used to estimate hierarchical analysis of genetic 
differentiation for each combined trapping period (Table 29).  The combined trapping 
periods from March 2000 produced the highest estimate of genetic variance between 
grids (5.05%) while the 1
st
 1998 trapping period was the lowest (0.00).  Within grid 
AMOVA values were approximately equal to combined trapping periods.  The percentage 
of individuals assigned to the population from which they were trapped for Gazebo grid 
ranged from 63.6% to 81.7% and Vet Village grid ranged from 59.1% to 81.2% (Table 
30).  The highest proportion of individuals correctly assigned on Gazebo grid was from 
the March 2002 trapping period while the lowest proportion were from the March 2001 
trapping period.  On Vet Village grid, the highest proportion of individuals correctly 
assigned was from the February 2002 trapping period and the lowest number correctly 
assigned were from the March 1996 trapping period. 
  
 181
Analysis of spatial genetic autocorrelation was conducted for each combined 
trapping period using unique distance classes and distance classes equal to those used for 
individual grids.  Significance was demonstrated for at least one distance class over 12 of 
the 13 trapping periods (Appendix 3).  Only the 2
nd
 2000 trapping period did not show 
significance for at least one distance class.  For the 1998 trapping period only the 200 m 
distance class was significant and only at a marginal level.  As with the previous 
analyses, males and females were combined again indicating that proximal individuals 
are more genetically similar regardless of sex. 
Discussion 
Differentiation among grids 
Analyses were conducted to determine if genetical spatial structure could be 
detected within a population of the Alabama beach mouse (Peromyscus polionotus 
ammobates) and, if structure was present, to determine if it had an effect on the rate of 
genetic decay within the population.  As previously discussed, the Alabama beach mouse 
displays none of the traditional mechanisms by which genetical structure might develop.  
Regardless, however, genetical spatial structure was demonstrated within this population.  
Structure was present on two levels.  First, the total data set comprised of combined 
trapping periods on both Gazebo and Vet Village grids showed a consistent pattern of 
microgeographic genetic structure (Appendix 3).  Second, genetical spatial structure was 
detected within each grid, however, less consistently than with the overall data set.  It is 
also shown that the rate of genetic decay is reduced compared to a predicted loss of 
heterozygosity based on a simple drift model (Fig. 3, 4).  Thus, genetical spatial structure 
was shown to be temporally present within this population of Alabama beach mice on a 
  
 182
microgeographic scale (<500 m), and importantly, the rate of genetic decay was 
effectively reduced. 
Wooten & Holler (1999) noted an increase in observed heterozygosity within a 
population of Alabama beach mice following hurricane Opal in October 1995.  Here, 
using a substantially larger data set of both individuals and loci, their observation is 
substantiated as these data also show a temporary increase in observed heterozygosity on 
both Gazebo and Vet Village grids during the year (1996) following hurricane Opal 
(Table 9 - 11).  A possible explanation for this temporary increase in observed 
heterozygosity would be ?isolate breaking? which occurs when subpopulations are forced 
to interbreed.  This type of phenomenon is generally referred to as a Wahlund effect 
(Wahlund 1928) and can indicate the presence of subdivision within the population.  The 
observed ?isolate breaking? is notable in the context of genetical spatial structure because 
beach mice exist in a narrow, contiguous habitat where large-scale habitat heterogeneity 
is not typically evident.  Clearly, narrow, linear beach habitat is not a system where 
genetical spatial structure would likely develop through landscape heterogeneity.  
However, within this system, consistent microgeographic spatial structure has been 
shown with positive autocorrelation detected across multiple time periods and distance 
classes.  The spatial structure of genotypes is believed to be the reason for the reduced 
rate of genetic decay that was measured.  Also, the recovery of spatial structure following 
recovery from a catastrophic event was observed.  Based on a suite of analyses, I believe 
that the data are sufficient to warrant rejection of the null hypothesis that genotypes are 
randomly distributed within the Alabama beach mouse population and conclude that 
proximal beach mice share more similar genotypes than do more removed beach mice. 
  
 183
The fixation index (F
IS
) values for the combined trapping periods were positive 
from February 1995 through February 1996 (1
st
 trapping period post hurricane) then 
became negative for the remainder of the 1996 trapping periods as negative F
IS
 values 
suggest outbreeding.  I interpret this as an indication that structure within the population 
had deteriorated.  However, F
IS
 values, measured for each trapping period on both grids 
separately, were less clear concerning spatial structuring within grids.  Measured over the 
total population F
IS
 values did indicate an excess of homozygotes prior to the hurricane 
and an excess of heterozygotes consistently following the hurricane.  This finding is 
consistent with the hypothesized presence of genetical structure within the population.  
The recovery period, 1997 through 2001, was marked by both positive and negative F
IS
 
values.  The inconsistency in these values during the recovery period is likely, to some 
extent, a reflection of small sample size.  Small sample size during the recovery period 
was primarily on Gazebo grid.  F
IS
 values for separate grids across trapping periods 
demonstrated the difficulty of detecting a clear signal at such a microgeographic level.  
However, concerning F
IS
 values, it is also probable that the grid does not represent a 
single subpopulation but a composite of subpopulations.  The grids were placed with only 
a basic knowledge of how breeding units within the population might be positioned both 
spatially and temporally.  Tests for heterozygote deficiency were significant at some loci 
for trapping periods even when the average heterozygosity for those periods was high.  
Null alleles were unlikely to account for the observed patterns because significant 
heterozygote deficiency was inconsistent across trapping periods and loci.  High levels of 
inbreeding were also not likely during this time, again due to the inconsistencies of 
  
 184
heterozygote deficiency across loci.  Higher levels of inbreeding usually affect all loci 
(Hartl & Clark 1989). 
Given the microgeographic scale of my analysis, it was of course difficult to 
distinguish true signal from noise.  However, F
ST
 values, while not globally high, were 
significant between grids and exhibited a substantial range (0.006-0.051; Table 29).  
Calculation of R
ST
 produced results essentially the same as F
ST
.  Estimates of F
ST
 among 
populations (grids) were significant for all but the June 1996, winter/spring 1998 and fall 
2000 trapping periods.  Collectively, my results suggest that individuals inhabiting each 
grid, approximately 630-910 m apart, were arranged across space in a distinctive allelic 
pattern.  Analyses of these data also suggest that the partitioning of alleles across space 
was temporal.  On a scale of < 1 km, F
ST
 values reported by Peakall et al. (2003) for 
Australian bush rats ranged from 0.04 to 0.08.  Also, on a scale of < 1 km, Root et al. 
(2003) calculated F
ST
 values for P. maniculatus at 0.031 and 0.043.  Thus, F
ST
 values 
reported here were consistent with other microsatellite analysis.  However, F
ST
 values for 
beach mice may be unparalleled in the context of the contiguous habitat pattern and small 
distance between grids.  Mantel tests among grids were significant for all of the 
combined trapping periods except June 1996 and fall 2000 suggesting a pattern of 
restricted gene flow between grids.  Typically, the association between genetic and 
geographic distance is not detected by a Mantel test unless the association is very strong 
(Peakall et al. 2003).  These analyses (F
IS
, F
ST
, Mantel test) all suggest the presence of 
genetical spatial structure on a microgeographic scale within a contiguous population of 
Alabama beach mice.  The AMOVA results, and microgeographic scale, are similar to 
those reported by Peakall et al. (2003).  AMOVA results for the total population ranged 
  
 185
from 2% to 5% of genetic variance among groups, while within grids AMOVA values, 
under some scenarios, were as high as 3.69%.  Thus, within grid differentiation (< 500 m 
through 1997;  < 360 m through 2002) was similar to between grid differentiation (< 
1600 m).   
Spatial autocorrelation analyses demonstrated microgeographic structure both 
among and within grids.  Within grids, the pattern of autocorrelation was not as 
consistent across seasons but was evident.  The model predicted that autocorrelation 
would be present prior to the hurricane and deteriorate following the hurricane.  When 
densities returned to prehurricane levels (2002), autocorrelation should have been 
strongly detected.  In fact, this is essentially what occurred (Fig. 25).  Prior to the 
hurricane spatial autocorrelation was detected at all distance classes except for the 20 m 
distance class during the combined trapping period of June 1995.  Also during the June 
1995 trapping period distance classes of 50 and 100 m were approaching nonsignificance.  
This may, in part, be due to the large annual population decline that beach mice 
experience each summer (Oli et al. 2001).  A large population decline might produce the 
same effect as the hurricane but on a smaller scale.  Significant spatial autocorrelation 
was detected for each trapping period up to approximately one year following the 
hurricane suggesting that the genetic spatial autocorrelation was left intact by the initial 
storm event.  However, by the June 1996 trapping period structure was not detected.  
During the summer period, structure was expected to be weak at smaller distance classes.  
However, a further reduction in structure during June 1996 may also indicate that 
factor(s) resulting from the hurricane was/were responsible for driving the deterioration 
of genetical spatial structure.   
  
 186
There are several possible factors that could explain the presence of genetic 
spatial structure within the Alabama beach mouse population including selection, surges 
in reproduction, dispersal patterns and social structure (Peakall et al. 2003).  In my case, 
selection is unlikely to explain the spatial autocorrelation because microsatellites are 
neutral loci and typically are spread throughout the genome (Jarne & Lagoda 1996).  
According to Epperson (1990), selection should generate spatial genetic structure through 
particular coding genes or loci linked to particular coding genes.  It is unlikely that the 11 
loci examined here are linked to coding genes that were each under similar selection.  
According to Scribner and Chesser (1993), the reproductive period of a population can 
lead to a clustering of similar genotypes.  However, data sets here were comprised 
predominately of adults so it is unlikely that clustering of juveniles created positive 
spatial structure that deteriorated as juveniles dispersed.  Rather, limited dispersal of 
juveniles promoted spatial structure.  Swilling & Wooten (2002) showed that beach mice, 
on average, tend to disperse about 160 m or within a distance of one homerange of their 
natal site.  The reproductive period of beach mice begins in November/December and 
juveniles begin entering the population en mass during January.   The reproductive 
season typically continues into April; afterward the population begins a dramatic decline.  
Thus, one would expect spatial structuring to be strongest during the peak of the 
reproductive season because juveniles tend to remain associated with their natal site.  
This is in fact what was observed.  Spatial structure was typically not detected during 
June trapping periods.  The apparent break down of spatial structure follows a 
demographic decline that is likely to increase mate searching among the adult age class. 
  
 187
For the three trapping periods of 1995 and the three of 1996 spatial 
autocorrelation was detected at the largest distance class of 800 m.  However, this pattern 
was not found during the recovery period (1997-2000).  In fact, spatial structure was 
rarely detected during the recovery period.  Lower survivorship and/or more frequent 
movement by individuals might account for this observation.  However, during 2001 and 
into 2002 spatial structure was clearly evident on both grids.  This may represent a point 
at which detrimental effects relating to the hurricane had diminished allowing population 
recovery.  Most important to my model, this pattern does indicate that genetical spatial 
structure will reorganize following disruption by a catastrophic event.  This is a critical 
point because it demonstrates that factors inherent in the population result in recurring 
spatial structure.  The precise location of clusters may change over time but, importantly, 
groups do appear to have reformed in this population. 
Within this population of Alabama beach mice, the most likely explanation for 
positive spatial autocorrelation is the social organization, nonrandom mating and the 
dispersal pattern of beach mice.  A limited pattern of dispersal, by both sexes, along with 
a monogamous mating system would necessarily create familial clusters.  Once 
established, kinship groups themselves can act to limit gene flow (Chesser 1991b; 
Scribner & Chesser 1993).  Also, according to van Staaden et al. (1996), positive spatial 
genetic structure may not occur over short distances unless dispersal is limited.  Here, 
when spatial structure was detected for the combined trapping periods, 58% of the time it 
was detected in the 20 m distance class.  Independently, on Gazebo and Vet Village grids 
spatial structure was detected at the 20 m distance class in 75% and 57% of the trapping 
periods, respectively.  Several authors have used computer simulations to demonstrate 
  
 188
that a vacillating pattern of positive and negative genetic structure can be brought about 
by restricted gene flow (Turner 1982; Sokal & Wartenburg 1983; Sokal et al. 1989; 
Epperson 1990; Epperson & Li 1997).  Such vacillations, while not significant, were 
present in some analyses within smaller distance classes (20 m, 50 m).  Thus, the pattern 
of genetic structure observed in beach mice is consistent with that predicted from a 
combination of social group formation, enforced by limited dispersal resulting in a 
clumped pattern of similar genotypes.  The genetic data produced in my study were 
consistent with field observations concerning dispersal patterns made by Swilling & 
Wooten (2000).  They showed that dispersal by juveniles was typically limited to within 
one homerange of their natal site or about 160 m. 
Differentiation within grids 
Gazebo 
The increase of observed heterozygosity on Gazebo grid following hurricane Opal 
was not as pronounced as the Vet Village increase.  However, patterns for both grids 
were consistent with the concept of isolate braking.  F
IS
 values from each trapping period 
on Gazebo grid were not always consistent with expectations based on spatial structure.  
Average F
IS
 values for trapping periods were both positive and negative but not generally 
in a predictable pattern.  However, it is unlikely that individuals captured on the grid 
represent a single, structurally distinct segment of the population.  F
ST
 values derived by 
dividing the grid into various quadrants indicated that under some scenarios subdivision 
of the grid could produce relatively high values.  On Gazebo grid, based on the three 
models of subdividing the population used here, the most likely scenario of spatial 
structure within the grid was the four corner quadrants and the east/west quadrants (Fig. 
  
 189
1).  However, the north/south quadrants were consistently the highest F
ST
 estimates.  The 
north/south quadrants reflect the primary and secondary dunes (south) and the scrub 
habitat (north).  Swilling & Wooten (2000) showed that dispersal between the two habitat 
types was limited.  My results support the idea that the fore and back dune habitats 
effectively function as separate units in spite of their close spatial relationship. 
Mantel tests were significant for the February 1995, March 2001, November 2001 
and March 2002 trapping periods.  Thus, within grid Gazebo (Feb-95 < 500m; Mar-01, 
Nov-01, Mar-02 < 360 m), an association between genetic and microgeographic distances 
was demonstrated.  The pattern of association detected by Mantel tests on Gazebo grid 
indicated that structure existed prior to the hurricane and reorganized following the 
recovery period (1997-2000).  While the Mantel test for April 1995 was not significant, it 
did approach significance (P = 0.057).  Mantel tests on Gazebo grid also indicated a 
pattern of association that is highest during late winter and declines throughout the 
remainder of the year.  Analyses of spatial autocorrelation on Gazebo grid produced eight 
trapping periods with positive autocorrelation for at least one distance class.  However, 
sample size was a limiting factor for four of these trapping periods.  Regardless, similar 
to the Mantel tests, the spatial autocorrelation analyses indicated that spatial structure 
existed prior to the hurricane and reorganized following the recovery period.  While the 
June 1995 trapping period occurred prior to the hurricane, not finding positive 
autocorrelation was consistent with other analyses that suggested a reduction in spatial 
genetic association during summer/fall.  Again, this observation is likely due to factors 
related to the sharp annual demographic decline experienced each summer.  Overall, the 
  
 190
seasonal pattern appears to be that structure within the grid is most defined during the late 
winter/spring then declines throughout the year as population recruitment wanes. 
Vet Village 
Analyses of Vet Village grid produced very similar results to Gazebo grid, 
however, some differences were observed.  The increase of observed heterozygosity on 
Vet Village grid following hurricane Opal was more pronounced and immediate than the 
increase on Gazebo grid.  The increase on Vet Village grid is, again, consistent with the 
concept of isolate braking.  F
IS
 values from each trapping period on Vet Village grid were 
generally more consistent with expectations based on spatial structure than F
IS
 values 
from Gazebo grid.  Typically, for trapping periods when F
IS
 was positive an excess of 
homozygotes was expected.  Similar to Gazebo grid, estimates of F
ST
 derived by dividing 
Vet Village grid into quadrants showed that defining subpopulations within the grid could 
produce relatively high values.  On Vet Village grid, the most likely scenario for 
demonstrating spatial structure was the four corner quadrants (Fig. 1).  The four corner 
quadrant estimates of F
ST
 on Vet Village grid indicated that structure was most likely to 
occur under this scenario during March.  This is consistent with a model that predicts 
structure is more defined during March but diminishes over summer.  Unlike Gazebo 
grid, on Vet Village grid the east/west quadrants consistently produced the highest F
ST
 
values. 
Mantel tests for Vet Village grid data were significant for the March 1996, March 
2001, November 2001 and March 2002 trapping periods.  Thus, within Vet Village grid 
(Mar-96 < 500m; Mar-01, Nov-01, Mar-02 < 360 m), an association between genetic and 
geographic distances was demonstrated.  On Vet Village grid, Mantel tests for five other 
  
 191
trapping periods approached significance suggesting an eight-year pattern indicating 
structure was more evident during winter/spring rather than summer/fall.  The pattern of 
genetic and geographic association on Vet Village grid, like Gazebo grid, indicates that 
structure existed prior to the hurricane and reorganized following the recovery period 
(1997-2000).  Analyses of spatial autocorrelation on Vet Village grid produced seven 
trapping periods with positive autocorrelation for at least one distance class.  Sample size 
was limiting for only one of these trapping periods.  Interestingly, spatial autocorrelation, 
unlike the Mantel test, was not significant during the March 1996 trapping period.  
Regardless, similar to the Mantel tests, the spatial autocorrelation analyses indicated that 
spatial structure existed on Vet Village grid prior to the hurricane and reorganized 
following the recovery period (1997-2000).  As on Gazebo grid, the June 1995 trapping 
period on Vet Village grid did not produce significant positive autocorrelation supporting 
the idea that positive spatial structure abated throughout the summer and fall.  This 
pattern has been observed on both Gazebo and Vet Village grids.  This phenomenon is 
likely due to a reorganization of the population following a sharp demographic decline.  
The conclusions reached based on separate data and analyses from each grid are in 
agreement concerning genetical spatial structure.  However, based on these analyses, 
spatial structure may have been more conspicuous on Vet Village grid than Gazebo grid.  
Following hurricane Opal, sections of Vet Village grid were fertilized and vegetation 
recovered more quickly than on Gazebo grid.  Vet Village grid produced more captures 
than Gazebo grid thus effective population size (N
e
) was higher on Vet Village grid.  This 
may also be partially explainable by the fact that Vet Village grid is more difficult to 
  
 192
emigrate to.  Water borders the north boundary of Vet Village grid effectively isolating 
the grid from the deep scrub habitat (Fig. 1). 
Conclusions 
Based on microsatellite analyses, beach mice exhibit moderate levels of genetic 
variation (Peakall et al. 2003; Williams et al. 2002; Kretzmann et al. 2001).  This level of 
variation persists even under the pressure of intensive annual demographic bottlenecks 
compounded by hurricanes that can further reduce population size across multiple years.  
In many cases, severe, persistent bottlenecks reduce genetic variation (Nei et al. 1975) 
and coupled with a contiguous habitat that readily allows dispersal (up to several 
kilometers) genetic variance was expected to be low.  These factors likely place 
immediate ecological stress on individuals as well as having evolutionary implications 
for the population.  Given the annual reduction in population size, one might expect 
frequent, long distance dispersal by beach mice.  However, in agreement with field 
observations, I found little evidence of a high rate of gene flow between grids (< 930 m).  
The observed low dispersal rates and distances were the predicted pattern for beach mice 
based on observation (Swilling & Wooten 1998) and genetics (Wooten & Holler 1998). 
With such a restricted pattern of gene flow microgeographic structure should 
develop even within this contiguous population.  Spatial autocorrelation analyses 
conducted for each grid, as well as combined data sets, indicated that beach mice 
populations were indeed comprised of clusters of similar genotypes as was predicted.  
These clusters, however, were apparently more temporal than anticipated as they were 
undetectable during the early summer and fall, even during periods of relatively high 
density.  Two possible scenarios by which structure could erode are populations where 
  
 193
significant demographic declines tend to leave neighboring individuals that are on 
average less genetically similar and if individuals are more likely to disperse following 
population declines.  However, field observations along with genetic data suggest 
dispersal is limited.  Supporting limited dispersal, Mantel tests and spatial autocorrelation 
analyses indicated that between grid differentiation was generally significant.  
Regeneration of spatial structure would occur through a reproductive cycle with limited 
dispersal by juveniles.  However, further compounding a reforming of spatial structure 
are periodic hurricanes that place severe stresses on the population.  Low population 
density during hurricane recovery periods might be characterized by higher dispersal 
rates than is typical.  Recovery period trapping results were marked by small sample sizes 
that were generally difficult to interpret.  Following the recovery period, however, 
structure reorganized both within and among grids.  This indicates properties of the beach 
mouse population that lead to the formation and reformation of genetical spatial structure. 
These properties that lead to genetical spatial structure appear to be an inherent 
aspect of the beach mouse population and are likely to have some evolutionary 
justification.  My research suggests that spatial structure within this population of 
Alabama beach mice reduces the rate of genetic decay.  Three of the main factors that can 
affect genetic diversity within populations are demographic bottlenecks, genetic drift and 
inbreeding.  Beach mice populations experience regular and persistent demographic 
bottlenecks.  Thus, genetic drift and inbreeding are likely, on some level, persistent forces 
within beach mice populations.  Regardless, observed heterozygosity declined at a slower 
rate than the predicted loss of heterozygosity.  Genetic diversity within beach mice 
populations appears to reside at a microgeographic level.  By partitioning genetic 
  
 194
variation in this manner, beach mice are able to slow the rate of loss of genetic variation.  
Restricted dispersal and social structure within beach mice populations are evolutionary 
adaptations that aid in the preservation of genetic variation.  I predict that genetic 
diversity resides at a microgeographic level for many species and represents an 
evolutionary adaptation by which genetic variation can be maintained at a higher level 
than would otherwise be expected.  In doing so, microspatial structure sets the boundaries 
within which all evolution, including natural selection must operate. 
Clearly, biological aspects of beach mice serve as a mechanism that functions to 
preserve genetic variation.  These biological ?traits? negate the effects of forces such as 
genetic drift that typically act to reduce genetic variation.  The findings of 
microgeographic genetic structure and its relationship in preserving genetic variation 
establish a connection between microevolution and the ecology of beach mice.  The 
dispersal ?ability? and social interactions of beach mice has a clear association with the 
magnitude and spatial scale of genetic differentiation within the population of Alabama 
beach mice.  The maintenance of genetic variation is essential for the evolutionary 
process, and genetic differentiation within the population provides the potential for local 
adaptation or even speciation. 
Combining demographic and genetic data in a spatiotemporal manner has 
provided insights into population processes that might have been undetected by a single 
method.  My results demonstrate that behavioral partitioning of genotypes can occur on a 
microgeographic scale in the absence of traditional breeding structures (e. g. single sex 
clusters, etc.).  Importantly, following a catastrophic event genotype partitioning was 
shown to reorganize.  Also, and unexpectedly, genotype partitioning deteriorated and 
  
 195
reorganized on an annual basis.  This indicates that partitioning of genotypes is an 
intrinsic aspect of the population.  Further, this partitioning of genotypes acts as a 
mechanism that slows the rate at which genetic variation is lost.  Given the process of 
genetic partitioning, future studies might investigate both the spatial and temporal 
formation and decay of actual breeding groups.  The demographic and genetic dynamics 
of both within and between breeding groups offers intriguing insights into such 
fundamental questions as the amount of gene flow during different periods, mate choice, 
and paternity analysis.  Large numbers of microsatellites markers have already been 
developed (Mullen et al. 2005).  Thus, laboratory developmental work that is both 
expensive and time consuming would be minimal.  Further and more extensive analyses 
of genetical spatial structure with the population of Alabama beach mice are warranted. 
 196
Fig. 1  Vet Village grids established during the fall of 1994.  Each grid is approximately 
25 X 13 m.  The grids were divided into four corner, north/south and east/west quadrants 
to model subdivision within each grid.  Water borders the north boundary of Vet Village 
grid. 
Gaz
e
bo grid
Vet Village grid
water
~ 620 m
North
197
 198
Fig. 2  The reduction of both grids increased the distance between them to approximately 
920 m. 
V
e
t Village 
grid
Gazebo gr
id
~ 920 m
North
199
 200
Fig. 3  Gazebo grid observed heterozygosity (H
o
) for each trapping period along with the 
predicted heterozygosity (H
t
) for each trapping period.  Predicted heterozygosity is based 
on equation 7.14 in Hartl & Clark (1997).  Observed heterozygosity increases following 
hurricane Opal in October 1995 while predicted heterozygosity declines. 
0.
50
0.
52
0.
54
0.
56
0.
58
0.
60
0.
62
02_95
02_96
02_97
04_98
02_99
03_00
03_01
03_02
H
O
Net
H
T
H
O
Heterozygosity
Trapping pe
riod (ti
m
e)
201
 202
Fig. 4  Vet Village observed heterozygosity (H
o
) for each trapping period along with the 
predicted heterozygosity (H
t
) for each trapping period.  Predicted heterozygosity is based 
on equation 7.14 in Hartl & Clark (1997).  Observed heterozygosity increases following 
hurricane Opal in October 1995 while predicted heterozygosity declines. 
 
0.
54
0.
56
0.
58
0.
60
0.
62
0.
64
0.
66
0.
68
03_95
03_96
03_97
02_98
03_00
03_01
02_02
H
O
Net
H
T
H
O
Heterozygos ity
Trapping pe
riod (ti
m
e)
203
 204
Fig. 5  Correlograms of Gazebo grid for the trapping period of February 1995 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null  
hypothesis of randomly distributed genotypes.  The 95% CI intervals about r are  
estimated by bootstrapping.  Autocorrelation analysis are presented for distance  
class sizes of 20 m (A), 100 m (B) and 250 m (C). 
 
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
100 200 300 400 500
r
-0.03
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
0.02
250 500
r
A
B
C
Distance (m)
205
 206
Fig. 6  Correlograms of Gazebo grid for the trapping period of April 1995 shows genetic 
correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
 
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
100 200 300 400 500
r
-0.03
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
0.02
250 500
r
A
B
C
Distance (m)
207
 208
Fig. 7  Correlograms of Gazebo grid for the trapping period of June 1995 shows genetic 
correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
100 200 300 400 500
r
-0.03
-0.03
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
0.02
250 500
r
Distance (m)
A
B
C
209
 210
Fig. 8  Correlograms of Gazebo grid for the trapping period of February 1996 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.40
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
0.40
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.12
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
100 200 300 400 500
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
250 500
r
Distance (m)
A
B
C
211
 212
Fig. 9  Correlograms of Gazebo grid for the trapping period of April 1996 shows genetic 
correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
100 200 300 400 500
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
250 500
r
A
B
C
Distance (m)
213
 214
Fig. 10  Correlograms of Gazebo grid for the trapping period of June 1996 shows genetic 
correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
 
-0.50
-0.40
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
20 60 100 140 180 220 260 300 340 380 420 460
r
-0.15
-0.10
-0.05
0.00
0.05
0.10
100 200 300 400 500
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
250 500
r
A
B
C
Distance (m)
215
 216
Fig. 11  Correlograms of Gazebo grid for the trapping period of March 2002 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
20 60 100 140 180 220 260 300 340 380 420
r
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
100 200 300 400 500
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
250 500
r
A
B
C
Distance (m)
217
 218
Fig. 12  Correlograms of Vet Village grid for the trapping period of March 1995 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
100 200 300 400 500
r
-0.02
-0.01
-0.01
0.00
0.01
0.01
250 500
r
A
B
C
Distance (m)
219
 220
Fig. 13  Correlograms of Vet Village grid for the trapping period of April 1995 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
 
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
100 200 300 400 500
r
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
250 500
r
A
B
C
Distance (m)
221
 222
Fig. 14  Correlograms of Vet Village grid for the trapping period of June 1995 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
100 200 300 400 500
r
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
250 500
r
A
B
C
Distance (m)
223
 224
Fig. 15  Correlograms of Vet Village grid for the trapping period of March 1996 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
100 200 300 400 500
r
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
250 500
r
A
B
C
Distance (m)
225
 226
Fig. 16  Correlograms of Vet Village grid for the trapping period of April 1996 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.15
-0.10
-0.05
0.00
0.05
0.10
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
100 200 300 400 500
r
-0.03
-0.03
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
0.02
250 500
r
A
B
C
Distance (m)
227
 228
Fig. 17  Correlograms of Vet Village grid for the trapping period of June 1996 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
20 60 100 140 180 220 260 300 340 380 420 460 500
r
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
100 200 300 400 500
r
-0.03
-0.03
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
0.02
0.03
250 500
r
A
B
C
Distance (m)
229
 230
Fig. 18  Correlograms of Vet Village grid for the trapping period of February 2002 shows 
genetic correlation (r) as a function of distance and the 95% CI about a null hypothesis of 
randomly distributed genotypes.  The 95% CI intervals about r are estimated by 
bootstrapping.  Autocorrelation analysis are presented for distance class sizes of 20 m 
(A), 100 m (B) and 250 m (C). 
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
20 60 100 140 180 220 260 300 340 380
r
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
100 200 300 400
r
-0.03
-0.03
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
250 500
r
A
B
C
Distance (m)
231
 232
Fig. 19  Correlograms from combined trapping periods of February 1995 on Gazebo grid 
and March 1995 on Vet Village grid demonstrating genetic correlation (r) as a function of 
distance and the 95% CI about a null hypothesis of randomly distributed genotypes.  The 
95% CI intervals about r are estimated by bootstrapping.  Autocorrelation analysis are 
presented for distance class sizes of 50 m (A), 200 m (B) and 800 m (C). 
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
50 150 250 350 450 550 650 750 850 950 1050 1150 1250 1350 1450 1550
r
-0.06
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
200 400 600 800 1000 1200 1400 1600
r
-0.03
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
0.02
0.03
800 1600
r
A
B
C
Distance (m)
233
 234
Fig. 20  Correlograms from combined trapping periods of April 1995 on both Gazebo 
and Vet Village grids demonstrating genetic correlation (r) as a function of distance and 
the 95% CI about a null hypothesis of randomly distributed genotypes.  The 95% CI 
intervals about r are estimated by bootstrapping.  Autocorrelation analysis are presented 
for distance class sizes of 50 m (A), 200 m (B) and 800 m (C). 
 
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
50 150 250 350 450 550 700 800 900 1000 1100 1200 1300 1400 1500 1600
r
-0.06
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
200 400 600 800 1000 1200 1400 1600
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
800 1600
r
A
B
C
Distance (m)
235
 236
Fig. 21  Correlograms from combined trapping periods of June 1995 on both Gazebo and 
Vet Village grids demonstrating genetic correlation (r) as a function of distance and the 
95% CI about a null hypothesis of randomly distributed genotypes.  The 95% CI intervals 
about r are estimated by bootstrapping.  Autocorrelation analysis are presented for 
distance class sizes of 50 m (A), 200 m (B) and 800 m (C). 
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
50 150 250 350 450 550 700 800 900 1000 1100 1200 1300 1400 1500 1600
r
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
200 400 600 800 1000 1200 1400 1600
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
800 1600
r
A
B
C
Distance (m)
237
 238
Fig. 22  Correlograms from combined trapping periods of February 1996 on Gazebo grid 
and March 1996 on Vet Village grid demonstrating genetic correlation (r) as a function of 
distance and the 95% CI about a null hypothesis of randomly distributed genotypes.  The 
95% CI intervals about r are estimated by bootstrapping.  Autocorrelation analysis are 
presented for distance class sizes of 50 m (A), 200 m (B) and 800 m (C). 
 
-0.40
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
0.60
50 150 250 350 450 550 700 800 900 1000 1100 1200 1300 1400 1500 1600
r
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
200 400 600 800 1000 1200 1400 1600
r
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
800 1600
r
A
B
C
Distance (m)
239
 240
Fig. 23  Correlograms from combined trapping periods of April 1996 on both Gazebo 
and Vet Village grids demonstrating genetic correlation (r) as a function of distance and 
the 95% CI about a null hypothesis of randomly distributed genotypes.  The 95% CI 
intervals about r are estimated by bootstrapping.  Autocorrelation analysis are presented 
for distance class sizes of 50 m (A), 200 m (B) and 800 m (C). 
 
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
50 150 250 350 450 550 700 800 900 1000 1100 1200 1300 1400 1500 1600
r
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
200 400 600 800 1000 1200 1400 1600
r
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
800 1600
r
A
B
C
Distance (m)
241
 242
Fig. 24  Correlograms from combined trapping periods of June 1996 on both Gazebo and 
Vet Village grids demonstrating genetic correlation (r) as a function of distance and the 
95% CI about a null hypothesis of randomly distributed genotypes.  The 95% CI intervals 
about r are estimated by bootstrapping.  Autocorrelation analysis are presented for 
distance class sizes of 50 m (A), 200 m (B) and 800 m (C). 
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
0.40
50 150 250 350 450 550 750 850 950 1050 1150 1250 1350 1450 1550
r
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
200 400 600 800 1000 1200 1400 1600
r
-0.03
-0.02
-0.02
-0.01
-0.01
0.00
0.01
0.01
0.02
0.02
0.03
800 1600
r
A
B
C
Distance (m)
243
 244
Fig. 25  Correlograms from combined trapping periods of March 2002 on Gazebo grid 
and February 2002 on Vet Village grid demonstrating genetic correlation (r) as a function 
of distance and the 95% CI about a null hypothesis of randomly distributed genotypes.  
The 95% CI intervals about r are estimated by bootstrapping.  Autocorrelation analysis 
are presented for distance class sizes of 50 m (A), 200 m (B) and 800 m (C). 
-0.25
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
50 150 250 350 450 950 1050 1150 1250 1350 1450 1550
r
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
200 400 600 1000 1200 1400 1600
r
-0.05
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
800 1600
r
A
B
C
Distance (m)
245
Table 1
  Characterization of 11 
m
i
crosatellite us
ed for 
Peromyscus polionotus ammobates
 including prim
er sequences, 
Accession n
u
m
b
er, repeat m
o
tif, annealing tem
p
eratu
r
e (T
a
) and num
ber of observed alleles. 
 Locus 
Accession no. 
Repeat m
o
tif 
Prim
er sequence (5'
 - 3'
) 
T
a
(
o
C) 
Ppa01 
AF016855 
(AC)
24
F:ATGTGCAGCTTTGTTTGAGTTA 
58 
Allele
s 
     3 
 
 
 
R:ATTTCCATCTTGTGTCTTTCCTA 
 
 
Ppa12 
AF016858 
(AG)
37
F:TGCCAAACTACA
CAAAGAGACCCT 
58 
     9 
 
 
 
R:GGACACAGAAGGTCTCAAAGCTGA
 
 
Ppa46 
AF016860 
(GT)
6
TT(GT)
18
F:AGTC
C
ACAGAGGGAGCT
TCT
AAG 
58 
     6 
 
 
 
R:TTTTACAGAAG
AAGGCAACATTTC 
 
 
PO-25 
AF380236 
(AG)
23
F:GGAC
AGCCAGGAC
TGTTACAC 
65 
     8 
 
 
 
R:CCCACT
C
ATCTCAATGCC 
 
 
PO-71 
AF380240 
(AC)
10
(AG)
4
AA(AG)
28
F:CAGC
CAGAACAAAATAG
CACT 
64 
   10 
 
 
 
R:AGCT
T
C
A
TGC
C
TC
CTAT
ATT
C
 
 
 
PO3-68 
AF380248 
(TG)
22+
F:GTAGT
C
T
GAGAAGCGAAAGG 
58 
     8 
 
 
 
R:TTTATTTGGGTCAGCTCGAC 
 
 
PO3-85 
AF380250 
(AC)
21
F:TGGC
AAGGCT
GAT
AAC 
58 
     6 
 
 
 
R:CTCCCT
CTTGGTCATCAGTA 
 
 
Pm
l03 
AF251777 
(CA)
22
F:GCCATT
AGTC
T
AT
GTGACAG 
54 
     8 
 
 
 
R:GCGATGTACC
CAGAAAT 
 
 
Pm
l04 
AF251778 
(CA)
27
F:CATAAGGTGGCTC
GGAATC
A
 
57 
     5 
 
 
 
R:CAGGAAGGGGAAATGACC
AT 
 
 
Pm
l06 
AF251780 
(CA)
24
F:CAGGGC
TGTAGAGGGAGAAC
 
54 
     3 
 
 
 
R:ACTGGAGCAGAGGCATTT
G 
 
 
Pm
l11 
AF251785 
(CA)
23
F:ACCC
CC
GAGT
GCT
GAGATT 
57 
     7 
 
 
 
RGC
C
AAC
C
ATTT
CTT
C
AAGTG 
 
 
 
246
 247
Table 2  Total captures, new captures per trapping period and the  
number of individuals from which tissue was obtained. 
  
Trap period Captures Tissue New captures Tissue 
1st-95 322 248 322 248 
2nd-95 329 235 98 89 
3rd-95 189 138 15 14 
1st-96 134 92 78 73 
2nd-96 178 124 78 65 
3rd-96 115 78 31 30 
1st-97 45 37 39 34 
1st-98 33 31 31 30 
1st-99 16 15 16 15 
1st-00 64 58 62 58 
2nd-00 26 21 22 18 
1st-01 83 65 71 56 
2nd-01 61 48 48 43 
1st-02 274 237 253 234 
     
Totals 1869 1427 1163 1007 
 
Table 3  Total captures, new captures and the number of individuals from  
which tissue was obtained on Gazebo grid. 
 
Gazebo 
 
Trap period Captures Tissue  New captures            Tissue 
Feb-95 170 145 170 145 
Apr-95 166 122 41 38 
Jun-95 91 68 6 6 
Feb-96 77 48 41 37 
Apr-96 75 52 25 24 
Jun-96 53 34 12 12 
Feb-97 11 9 9 8 
Apr-98 10 9 9 9 
Feb-99 9 9 9 9 
Mar-00 33 32 33 32 
Nov-00 16 13 14 11 
Mar-01 29 22 24 18 
Nov-01 12 12 11 11 
Mar-02 138 120 135 117 
    
Totals 890 695 539 477 
 
 248
Table 4  Total captures, new captures and the number of individuals  
from which tissue was obtained on Vet Village grid. 
 
Vet Village     
Trap period Captures Tissue New captures Tissue 
Mar-95 152 103 152 103 
Apr-95 163 113 57 51 
Jun-95 98 70 9 8 
Mar-96 57 44 37 36 
Apr-96 103 72 53 41 
Jun-96 62 44 19 18 
Mar-97 34 28 30 26 
Feb-98 23 22 22 21 
Feb-99 7 6 7 6 
Mar-00 31 26 29 26 
Dec-00 10 8 8 7 
Mar-01 54 43 47 38 
Oct-01 49 36 37 32 
Feb-02 136 117 118 117 
     
Totals 979 732 625 530 
 
 249
Table 5  The total number of alleles observed for each 
trapping period on each grid. 
 
Locus Gazebo Vet Village Total 
Ppa-01 3 3 3 
Ppa-12 8 6 9 
Ppa-46 6 3 6 
PO3-85 6 6 6 
PO3-68 8 7 8 
PO-25 8 6 8 
Pml-03 7 6 8 
Pml-11 7 7 7 
Pml-04 4 4 5 
PO-71 10 8 10 
Pml-06 3 3 3 
    
Total 70 59 73 
 
 250
Table 6  Summary statistics from Gazebo trap period February  
1995 and Vet Village trap period March, 1995 included are  
sample size (N), number of alleles, effective number of alleles  
(N
e
), observed heterozygosity (H
OBS
), expected heterozygosity  
(H
EXP
), and the fixation index (F). 
 
Gazebo (Feb-95) 
Locus N Alleles N
e
H
OBS
H
EXP
F     
Ppa-01 142 3.0 1.6632 0.3485 0.3987  0.1260    
Ppa-12 142 4.0 2.4367 0.6761 0.5896 -0.1466     
Ppa-46   97 3.0 2.2040 0.5979 0.5463 -0.0946     
PO3-85 143 5.0 2.2975 0.6014 0.5647 -0.0649     
PO3-68 141 7.0 4.1798 0.7305 0.7608  0.0398     
PO-25 145 5.0 2.6969 0.6069 0.6292  0.0355     
Pml-03 142 5.0 2.4608 0.5435 0.5936  0.0845     
Pml-11 142 6.0 4.0243 0.7042 0.7515  0.0629     
Pml-04 118 3.0 1.9340 0.5508 0.4829 -0.1406     
PO-71 142 8.0 3.2747 0.6831 0.6946  0.0166     
Pml-06 143 3.0 2.6326 0.6643 0.6202 -0.0712     
       
Mean 136 4.73 2.7096 0.6095 0.6031 -0.0140     
St. Dev  1.74 0.8046 0.1062 0.1084  0.0936     
       
Vet Village (Mar-95)    
Locus N Alleles N
e
H
OBS
H
EXP
F     
Ppa-01 100 2.0 1.8247 0.4100 0.4520  0.0928     
Ppa-12   98 5.0 2.9451 0.6531 0.6605  0.0112     
Ppa-46   94 3.0 2.0406 0.4894 0.5100  0.0404     
PO3-85 102 5.0 3.4061 0.7353 0.7064 -0.0409     
PO3-68 101 6.0 4.4682 0.8119 0.7762 -0.0460     
PO-25 100 4.0 2.5777 0.7000 0.6120 -0.1437     
Pml-03 103 4.0 2.5521 0.5534 0.6082  0.0900     
Pml-11 103 6.0 4.1727 0.7282 0.7603  0.0423     
Pml-04 103 3.0 1.6303 0.4175 0.3866 -0.0798     
PO-71   86 8.0 3.3489 0.6395 0.7014  0.0882     
Pml-06 101 3.0 2.7656 0.6040 0.6384  0.0540     
       
Mean 99 4.45 2.8848 0.6128 0.6191  0.0097     
St. Dev  1.75 0.9102 0.1328 0.1243  0.0780     
 
 
251
Table 7  Summary statistics from Gazebo trap period April, 1995 and  
Vet Village trap period April, 1995 included are sample size (N),  
number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the fixation  
index (F). 
 
Gazebo (Apr-95) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 112 3.0 1.7174 0.4286 0.4177 -0.0260 
Ppa-12 119 4.0 2.5075 0.6218 0.6012 -0.0344 
Ppa-46   95 3.0 1.8742 0.5263 0.4664 -0.1284 
PO3-85 121 5.0 2.2361 0.5785 0.5528 -0.0465 
PO3-68 119 7.0 4.2886 0.7227 0.7668  0.0576 
PO-25 121 5.0 2.6654 0.5950 0.6248  0.0477 
Pml-03   91 5.0 2.5247 0.4835 0.6039  0.1994 
Pml-11 118 6.0 4.0553 0.7288 0.7534  0.0326 
Pml-04 106 3.0 1.8863 0.5094 0.4699 -0.0842 
PO-71 122 7.0 3.5400 0.7131 0.7175  0.0061 
Pml-06 121 3.0 2.5957 0.6612 0.6147 -0.0755 
       
Mean 113 4.64 2.7173 0.5973 0.5991 -0.0045 
St. Dev.  1.57 0.8758 0.1028 0.1167  0.0890 
      
Vet Village (Apr-95) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 111 2.0 1.8804 0.4955 0.4682 -0.0583 
Ppa-12 108 5.0 2.9710 0.6667 0.6634 -0.0049 
Ppa-46 111 3.0 2.0790 0.5135 0.5190  0.0106 
PO3-85 113 6.0 3.2783 0.6726 0.6950  0.0322 
PO3-68 112 7.0 4.8545 0.8214 0.7940 -0.0345 
PO-25 113 5.0 2.6153 0.6903 0.6176 -0.1176 
Pml-03 113 4.0 2.4888 0.5664 0.5982  0.0532 
Pml-11 113 7.0 3.8271 0.7434 0.7387 -0.0063 
Pml-04 113 3.0 1.7451 0.4425 0.4270 -0.0363 
PO-71   98 8.0 3.0916 0.6837 0.6765 -0.0105 
Pml-06 113 3.0 2.7445 0.5664 0.6356  0.1090 
       
Mean 111 4.82 2.8704 0.6237 0.6213 -0.0058 
St. Dev.  1.99 0.9042 0.1157 0.1124  0.0595 
 
 
252
Table 8  Summary statistics from Gazebo trap period June, 1995  
and Vet Village trap period June, 1995 included are sample size (N),  
number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo (Jun-95) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 60 3.0 1.7557 0.4167 0.4304  0.0319 
Ppa-12 66 3.0 2.5407 0.6970 0.6064 -0.1493 
Ppa-46 53 3.0 2.0700 0.5283 0.5169 -0.0220 
PO3-85 67 5.0 2.3441 0.6567 0.5734 -0.1453 
PO3-68 65 6.0 4.3467 0.7385 0.7699  0.0409 
PO-25 66 5.0 2.3357 0.5606 0.5719  0.0197 
Pml-03 47 5.0 2.5060 0.4255 0.6010  0.2919 
Pml-11 64 6.0 4.2314 0.6875 0.7637  0.0997 
Pml-04 61 2.0 1.7698 0.5410 0.4350 -0.2437 
PO-71 66 7.0 3.1114 0.7121 0.6786 -0.0494 
Pml-06 67 3.0 2.6578 0.6418 0.6237 -0.0289 
       
Mean 62 4.36 2.6972 0.6006 0.5974 -0.0139 
St. Dev.  1.63 0.8777 0.1130 0.1127  0.1424 
      
Vet Village (Jun-95) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 68 2.0 1.9490 0.4265 0.4869  0.1241 
Ppa-12 67 5.0 2.8358 0.7015 0.6474 -0.0836 
Ppa-46 68 3.0 2.0805 0.5441 0.5194 -0.0477 
PO3-85 70 5.0 3.1705 0.6571 0.6846  0.0401 
PO3-68 70 6.0 4.5837 0.8286 0.7818 -0.0598 
PO-25 70 5.0 2.5708 0.6857 0.6110 -0.1222 
Pml-03 70 4.0 2.7169 0.5143 0.6319  0.1862 
Pml-11 70 7.0 3.8750 0.7286 0.7419  0.0180 
Pml-04 70 3.0 1.6228 0.4000 0.3838 -0.0423 
PO-71 60 8.0 3.1607 0.7000 0.6836 -0.0240 
Pml-06 70 3.0 2.7754 0.5571 0.6397  0.1290 
       
Mean 69 4.64 2.8493 0.6130 0.6194  0.0106 
St. Dev.  1.86 0.8524 0.1344 0.1159  0.0987 
 
 253
Table 9  Summary statistics from Gazebo trap period February, 1996  
and Vet Village trap period March, 1996 included are sample size (N),  
number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo (Feb-96) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 47 3.0 1.7858 0.3830 0.4400  0.1296 
Ppa-12 47 3.0 2.7121 0.5745 0.6313  0.0900 
Ppa-46 48 3.0 2.3963 0.5208 0.5827  0.1061 
PO3-85 48 4.0 2.5815 0.6458 0.6126 -0.0542 
PO3-68 46 6.0 4.5850 0.8261 0.7819 -0.0565 
PO-25 47 5.0 2.6598 0.6383 0.6240 -0.0229 
Pml-03 10 4.0 2.6667 0.5000 0.6250  0.2000 
Pml-11 48 6.0 3.6923 0.7083 0.7292  0.0286 
Pml-04 48 3.0 2.1294 0.6250 0.5304 -0.1784 
PO-71 47 5.0 2.8284 0.7021 0.6464 -0.0861 
Pml-06 48 3.0 2.6995 0.5833 0.6296  0.0734 
       
Mean 44 4.09 2.7943 0.6096 0.6212  0.0209 
St. Dev.  1.22 0.7568 0.1186 0.0900  0.1111 
      
Vet Village (Mar-96) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 44 2.0 1.8615 0.5455 0.4628 -0.1786 
Ppa-12 43 4.0 3.1286 0.6512 0.6804  0.0429 
Ppa-46 44 3.0 2.2776 0.5455 0.5610  0.0276 
PO3-85 44 5.0 3.3876 0.7955 0.7048 -0.1286 
PO3-68 44 6.0 4.5877 0.8409 0.7820 -0.0753 
PO-25 43 5.0 2.3569 0.5349 0.5757  0.0709 
Pml-03 43 4.0 2.9141 0.5349 0.6568  0.1857 
Pml-11 42 6.0 4.3557 0.8140 0.7704 -0.0565 
Pml-04 44 3.0 1.8447 0.4773 0.4579 -0.0423 
PO-71 41 7.0 4.2883 0.8049 0.7668 -0.0497 
Pml-06 44 3.0 2.8038 0.7045 0.6433 -0.0951 
       
Mean 43 4.36 3.0735 0.6589 0.6420 -0.2720 
St. Dev.  1.57 0.9867 0.1375 0.1158  0.1025 
 
 
254
Table 10  Summary statistics from Gazebo trap period April, 1996  
and Vet Village trap period April, 1996 included are sample size (N),  
number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo (Apr-96) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 52 3.0 1.9616 0.5192 0.4902 -0.0592 
Ppa-12 52 3.0 2.7217 0.6154 0.6326  0.0272 
Ppa-46 52 3.0 2.3482 0.6538 0.5741 -0.1388 
PO3-85 52 5.0 2.1367 0.4615 0.5320  0.1324 
PO3-68 52 6.0 4.8372 0.8269 0.7933 -0.0424 
PO-25 51 5.0 2.8964 0.6667 0.6547 -0.0182 
Pml-03 29 5.0 1.7376 0.4483 0.4245 -0.0560 
Pml-11 51 6.0 3.9290 0.7059 0.7455  0.0531 
Pml-04 49 3.0 1.9433 0.5306 0.4854 -0.0931 
PO-71 51 5.0 2.7685 0.7451 0.6388 -0.1664 
Pml-06 49 3.0 2.3714 0.5306 0.5783  0.0825 
       
Mean 49 4.27 2.6956 0.6095 0.5953 -0.0254 
St. Dev.  1.27 0.9334 0.1219 0.1121  0.0921 
      
Vet Village (Apr-96) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 72 3.0 1.7089 0.4167 0.4148 -0.0044 
Ppa-12 71 4.0 3.0276 0.6620 0.6697  0.0116 
Ppa-46 72 3.0 2.3409 0.6806 0.5728 -0.1881 
PO3-85 72 5.0 3.0903 0.7361 0.6764 -0.0883 
PO3-68 72 6.0 5.2390 0.8333 0.8091 -0.0299 
PO-25 69 5.0 2.4803 0.6522 0.5968 -0.0927 
Pml-03 71 5.0 3.1774 0.6761 0.6853  0.0135 
Pml-11 70 6.0 4.1315 0.7286 0.7580  0.0388 
Pml-04 72 4.0 1.9304 0.5139 0.4820 -0.0662 
PO-71 69 8.0 3.6595 0.7826 0.7267 -0.0769 
Pml-06 72 3.0 2.7234 0.6528 0.6328 -0.0316 
       
Mean 71 4.73 3.0462 0.6669 0.6386 -0.0467 
St. Dev.  1.56 1.0168 0.1167 0.1167  0.0646 
 
 
255
Table 11  Summary statistics from Gazebo trap period June, 1996 and  
Vet Village trap period June, 1996 included are sample size (N),  
number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo   (Jun-96) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 34 3.0 1.9744 0.5294 0.4935 -0.0727 
Ppa-12 34 5.0 2.8508 0.6765 0.6492 -0.0420 
Ppa-46 34 4.0 2.3616 0.6765 0.5766 -0.1733 
PO3-85 34 5.0 2.4083 0.4706 0.5848  0.1953 
PO3-68 34 7.0 5.2785 0.8824 0.8106 -0.0886 
PO-25 34 8.0 3.0542 0.7059 0.6726 -0.0495 
Pml-03 25 6.0 2.1259 0.5200 0.5296  0.0181 
Pml-11 33 7.0 3.8077 0.7273 0.7374  0.0137 
Pml-04 34 4.0 2.2468 0.7353 0.5549 -0.3250 
PO-71 29 5.0 2.7894 0.7241 0.6415 -0.1288 
Pml-06 31 3.0 2.2045 0.4375 0.5464  0.1993 
       
Mean 32 5.18 2.8275 0.6440 0.6180 -0.0414 
St. Dev.  1.66 0.9676 0.1360 0.0954  0.1513 
      
Vet Village (Jun-96) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 43 2.0 1.7018 0.5798 0.4124 -0.1843 
Ppa-12 42 4.0 3.0309 0.5952 0.6701  0.1117 
Ppa-46 43 3.0 2.3420 0.5349 0.5730  0.0665 
PO3-85 42 5.0 3.2015 0.7857 0.6876 -0.1426 
PO3-68 43 6.0 5.0727 0.7674 0.8029  0.0441 
PO-25 41 5.0 3.0018 0.7073 0.6669 -0.0607 
Pml-03 43 6.0 2.8100 0.6047 0.6441  0.0613 
Pml-11 42 6.0 3.8983 0.7143 0.7435  0.0393 
Pml-04 44 4.0 1.7209 0.5227 0.4189 -0.2478 
PO-71 42 8.0 4.2000 0.7857 0.7619 -0.0312 
Pml-06 42 3.0 2.5658 0.4524 0.6103  0.2587 
       
Mean 42 4.73 3.0496 0.6409 0.6355 -0.0077 
St. Dev.  1.74 1.0265 0.1165 0.1273  0.1453 
 
 
256
Table 12 Summary statistics from Gazebo trap period February,  
1997 and Vet Village trap period March, 1997 included are sample  
size (N), number of alleles, effective number of alleles (N
e
),  
observed heterozygosity (H
OBS
), expected heterozygosity (H
EXP
),  
and the fixation index (F). 
 
Gazebo (Feb-97) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 9 2.0 1.9059 0.5556 0.4753 -0.1688 
Ppa-12 8 4.0 3.2821 0.3750 0.6953  0.4607 
Ppa-46 9 2.0 1.9059 0.5556 0.4753 -0.1688 
PO3-85 9 3.0 2.4179 0.6667 0.5864 -0.1368 
PO3-68 8 6.0 4.4138 0.7500 0.7734  0.0303 
PO-25 9 4.0 3.1765 0.6667 0.6852  0.0270 
Pml-03 8 4.0 2.9767 0.5000 0.6641  0.2471 
Pml-11 9 4.0 2.9455 1.0000 0.6605 -0.5140 
Pml-04 9 3.0 2.1600 0.3333 0.5370  0.3793 
PO-71 6 3.0 1.4118 0.3333 0.2917 -0.1429 
Pml-06 9 3.0 2.0506 0.4444 0.5123  0.1325 
       
Mean 8 3.45 2.6704 0.5619 0.5776  0.0132 
St. Dev.  1.13 0.8515 0.2017 0.1365  0.2820 
      
Vet Village (Mar-97) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 28 2.0 1.9382 0.5357 0.4841 -0.1067 
Ppa-12 28 3.0 2.2958 0.6429 0.5644 -0.1390 
Ppa-46 28 3.0 1.6915 0.3929 0.4088  0.0390 
PO3-85 28 4.0 3.2464 0.6429 0.6920   0.0710 
PO3-68 27 6.0 4.4316 0.8519 0.7743 -0.1001 
PO-25 28 4.0 2.5496 0.5357 0.6078  0.1186 
Pml-03 28 4.0 3.1173 0.7500 0.6792 -0.1042 
Pml-11 28 5.0 3.2131 0.6786 0.6888  0.0148 
Pml-04 28 4.0 1.8044 0.5000 0.4458 -0.1216 
PO-71 28 6.0 3.4161 0.8214 0.7073 -0.1614 
Pml-06 28 3.0 2.3473 0.5714 0.5740  0.0044 
       
Mean 28 4.00 2.7318 0.6295 0.6023 -0.0441 
St. Dev.  1.26 0.8336 0.1404 0.1183  0.0959 
 
 
257
Table 13  Summary statistics from Gazebo trap period April, 1998  
and Vet Village trap period February, 1998 included are sample size (N),  
number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the 
 fixation index (F). 
 
Gazebo (Apr-98) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 8 2.0 1.8824 0.5000 0.4688 -0.0667 
Ppa-12 7 4.0 3.1613 0.4286 0.6837  0.3731 
Ppa-46 9 4.0 2.2817 0.4444 0.5617  0.2088 
PO3-85 8 3.0 1.8551 0.2500 0.4609  0.4576 
PO3-68 8 5.0 3.7647 1.0000 0.7344 -0.3617 
PO-25 8 4.0 3.1220 1.0000 0.6797 -0.4713 
Pml-03 9 4.0 3.2400 0.3333 0.6914  0.5179 
Pml-11 8 4.0 2.8444 0.3750 0.6484  0.4217 
Pml-04 9 3.0 1.7419 0.3333 0.4259  0.2174 
PO-71 8 5.0 2.6667 0.6250 0.6250  0.0000 
Pml-06 8 3.0 2.4151 0.7500 0.5859 -0.2800 
       
Mean 8 3.73 2.6341 0.5490 0.5969  0.0925 
St. Dev  0.90 0.6593 0.2634 0.1053  0.3504 
      
Vet Village (Feb-98) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 22 3.0 2.0907 0.5909 0.5217 -0.1327 
Ppa-12 22 3.0 2.1753 0.4545 0.5403  0.1587 
Ppa-46 22 3.0 2.1802 0.4545 0.5413  0.1603 
PO3-85 22 4.0 3.8413 0.7273 0.7397  0.0168 
PO3-68 22 5.0 3.7231 0.6818 0.7314  0.0678 
PO-25 22 3.0 2.4631 0.6364 0.5940 -0.0713 
Pml-03 22 4.0 3.6255 0.8182 0.7242 -0.1298 
Pml-11 22 4.0 3.3495 0.7727 0.7014 -0.1016 
Pml-04 22 3.0 1.7505 0.5909 0.4287 -0.3783 
PO-71 20 5.0 2.6936 0.7500 0.6287 -0.1928 
Pml-06 22 3.0 2.2830 0.6818 0.5620 -0.2132 
       
Mean 22 3.64 2.7431 0.6509 0.6103 -0.0742 
St. Dev  0.81 0.7520 0.1202 0.1029  0.1646 
 
 
258
Table 14  Summary statistics from Gazebo trap period March, 2000  
and Vet Village trap period March, 2000 included are sample size (N),  
number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo (Mar-00) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 32 2.0 1.8824 0.3750 0.4688  0.2000 
Ppa-12 31 3.0 2.0253 0.3871 0.5062  0.2354 
Ppa-46 32 3.0 2.2165 0.5938 0.5488 -0.0819 
PO3-85 31 5.0 2.6732 0.6452 0.6259 -0.0308 
PO3-68 31 5.0 3.7539 0.6452 0.7336  0.1206 
PO-25 31 4.0 2.0869 0.5161 0.5208  0.0090 
Pml-03 32 4.0 1.3412 0.2500 0.2544  0.0173 
Pml-11 32 5.0 3.3964 0.7500 0.7056 -0.0630 
Pml-04 32 2.0 1.9692 0.4375 0.4922  0.1111 
PO-71 27 6.0 2.1378 0.4815 0.5322  0.0954 
Pml-06 32 3.0 2.5473 0.6250 0.6074 -0.0289 
       
Mean 31 3.82 2.3662 0.5187 0.5451  0.0530 
St. Dev.  1.33 0.6947 0.1485 0.1295  0.1064 
      
Vet Village  (Mar-00) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 26 2.0 1.9737 0.6538 0.4933 -0.3253 
Ppa-12 25 3.0 2.5667 0.6400 0.6104 -0.0485 
Ppa-46 26 3.0 1.9123 0.6538 0.4771 -0.3705 
PO3-85 26 4.0 2.5851 0.6538 0.6132 -0.0663 
PO3-68 26 4.0 2.3032 0.6538 0.5658 -0.1556 
PO-25 24 4.0 3.1911 0.7917 0.6866 -0.1530 
Pml-03 26 4.0 2.5037 0.6538 0.6006 -0.0887 
Pml-11 26 6.0 3.1369 0.8462 0.6812 -0.2421 
Pml-04 25 3.0 1.6513 0.2000 0.3944  0.4929 
PO-71 24 6.0 3.7770 0.7083 0.7352  0.0366 
Pml-06 25 3.0 2.0227 0.6000 0.5056 -0.1867 
       
Mean 26 3.82 2.5113 0.6415 0.5785 -0.1006 
St. Dev.  1.25 0.6430 0.1632 0.1032  0.2307 
 
 
259
Table 15  Summary statistics from Gazebo trap period November,  
2000 and Vet Village trap period December, 2000 included are sample  
size (N), number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo (Nov-00) 
Locus N Alleles N
e
H
OBS
H
EXP
F   
Ppa-01 8 2.0 1.8824 0.5000 0.4688 -0.0667   
Ppa-12 11 3.0 2.3725 0.3636 0.5785  0.3714   
Ppa-46 11 2.0 1.5414 0.0909 0.3512  0.7412   
PO3-85 10 3.0 1.9417 0.5000 0.4850 -0.0309    
PO3-68 11 5.0 3.7231 0.9091 0.7314 -0.2429    
PO-25 10 4.0 3.7736 0.6000 0.7350  0.1837   
Pml-03 11 4.0 2.2000 0.4545 0.5455  0.1667   
Pml-11 11 4.0 3.5072 0.8182 0.7149 -0.1445   
Pml-04 11 2.0 1.8615 0.5455 0.4628 -0.1786   
PO-71 11 6.0 4.0333 0.6364 0.7521  0.1538   
Pml-06 9 3.0 2.3143 0.5556 0.5679  0.0217   
       
Mean 10 3.45 2.6501 0.5431 0.5812  0.0885   
St. Dev.  1.29 0.9163 0.2167 0.1354  0.2835    
      
Vet Village (Dec-00)   
Locus N Alleles N
e
H
OBS
H
EXP
F   
Ppa-01 8 2.0 1.6000 0.2500 0.3750  0.3333   
Ppa-12 8 3.0 2.8444 0.7500 0.6484 -0.1566   
Ppa-46 8 3.0 2.1333 0.5000 0.5313  0.0588   
PO3-85 8 3.0 2.6122 0.8750 0.6172 -0.4177   
PO3-68 8 6.0 2.3704 0.5000 0.5781  0.1351   
PO-25 7 4.0 2.9697 0.5714 0.6633  0.1385   
Pml-03 8 1.0 1.0000 0.0000 0.0000 #N/A   
Pml-11 8 4.0 3.1220 0.6250 0.6797  0.0805   
Pml-04 8 2.0 1.9692 0.1250 0.4922  0.7460   
PO-71 8 6.0 3.7647 0.5000 0.7344  0.3191   
Pml-06 8 3.0 1.6623 0.2500 0.3984  0.3725   
       
Mean 8 3.36 2.3679 0.4496 0.5196  0.1608   
St. Dev.  1.57 0.7964 0.2667 0.2069  0.3156    
 
 
260
Table 16  Summary statistics from Gazebo trap period March, 2001  
and Vet Village trap period March, 2001 included are sample size (N),  
number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo (Mar-01) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 20 2.0 1.8824 0.4500 0.4688  0.0400 
Ppa-12 22 3.0 2.2776 0.5000 0.5610  0.1087 
Ppa-46 22 3.0 2.0125 0.5000 0.5031  0.0062 
PO3-85 22 4.0 2.9877 0.6364 0.6653  0.0435 
PO3-68 22 6.0 4.3214 0.7273 0.7686  0.0538 
PO-25 21 4.0 2.4845 0.7619 0.5975 -0.2751 
Pml-03 22 4.0 1.7441 0.4091 0.4267  0.0412 
Pml-11 22 5.0 3.8876 0.7273 0.7428  0.0209 
Pml-04 22 2.0 1.8161 0.5000 0.4494 -0.1126 
PO-71 21 7.0 2.7477 0.6667 0.6361 -0.0481 
Pml-06 22 3.0 2.7115 0.8182 0.6312 -0.2962 
       
Mean 21 3.91 2.6248 0.6087 0.5865 -0.0380 
St. Dev.  1.58 0.8441 0.1414 0.1159  0.1352 
      
Vet Village (Mar-01) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 43 2.0 1.4947 0.3721 0.3310 -0.1242 
Ppa-12 39 3.0 2.8298 0.6154 0.6466  0.0483 
Ppa-46 43 3.0 2.1525 0.4419 0.5354  0.1747 
PO3-85 42 4.0 2.3758 0.5476 0.5791  0.0543 
PO3-68 43 6.0 2.7618 0.6279 0.6379  0.0157 
PO-25 32 4.0 2.8967 0.6875 0.6548 -0.0500 
Pml-03 43 4.0 1.9443 0.4884 0.4857 -0.0056 
Pml-11 40 5.0 3.3862 0.7250 0.7047 -0.0288 
Pml-04 43 3.0 2.0330 0.5581 0.5081 -0.0985 
PO-71 43 6.0 3.6796 0.7907 0.7282 -0.0858 
Pml-06 43 3.0 1.9210 0.5349 0.4794 -0.1156 
       
Mean 42 3.91 2.4979 0.5809 0.5719 -0.0196 
St. Dev.  1.30 0.6724 0.1245 0.1177  0.0900 
 
 
261
Table 17  Summary statistics from Gazebo trap period November,  
2001 and Vet Village trap period October, 2001 included are sample  
size (N), number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo (Nov-01) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 12 2.0 1.9459 0.6667 0.4861 -0.3714 
Ppa-12 12 3.0 1.6457 0.5000 0.3924 -0.2743 
Ppa-46 12 3.0 2.0719 0.6667 0.5174 -0.2886 
PO3-85 11 3.0 1.7536 0.3636 0.4298  0.1538 
PO3-68 12 6.0 3.8919 0.8333 0.7431 -0.1215 
PO-25 12 5.0 2.4828 0.6667 0.5972 -0.1163 
Pml-03 12 3.0 2.1333 0.5833 0.5313 -0.0980 
Pml-11 12 4.0 3.2360 0.8333 0.6910 -0.2060 
Pml-04 12 3.0 1.9862 0.5833 0.4965 -0.1748 
PO-71 11 3.0 2.1416 0.7273 0.5331 -0.3643 
Pml-06 12 3.0 2.6422 0.6667 0.6215 -0.0726 
       
Mean 12 3.45 2.3575 0.6446 0.5490 -0.1757 
St. Dev.  1.13 0.6767 0.1368 0.1061  0.1513 
      
Vet Village (Oct-01) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 35 2.0 1.4706 0.2286 0.3200  0.2857 
Ppa-12 34 3.0 2.6544 0.5294 0.6233  0.1506 
Ppa-46 35 3.0 1.9460 0.4000 0.4861  0.1772 
PO3-85 36 4.0 3.1116 0.5278 0.6786  0.2223 
PO3-68 35 5.0 3.3243 0.7143 0.6992 -0.0216 
PO-25 32 5.0 3.8861 0.7188 0.7427  0.0322 
Pml-03 34 4.0 1.7088 0.4412 0.4148 -0.0636 
Pml-11 33 5.0 3.6667 0.7273 0.7273  0.0000 
Pml-04 36 2.0 1.7693 0.5833 0.4348 -0.3416 
PO-71 36 5.0 2.9827 0.7500 0.6647 -0.1283 
Pml-06 36 3.0 1.9445 0.5278 0.4857 -0.0866 
       
Mean 35 3.73 2.5877 0.5589 0.5707  0.0205 
St. Dev.  1.19 0.8770 0.1630 0.1462  0.1809 
 
 
262
Table 18  Summary statistics from Gazebo trap period March, 2002  
and Vet Village trap period February, 2002 included are sample size  
(N), number of alleles, effective number of alleles (N
e
), observed  
heterozygosity (H
OBS
), expected heterozygosity (H
EXP
), and the  
fixation index (F). 
 
Gazebo (Mar-02) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 119 2.0 1.8491 0.4622 0.4592 -0.0065 
Ppa-12 110 5.0 2.0748 0.4364 0.5180  0.1576 
Ppa-46 120 3.0 1.9907 0.5167 0.4977 -0.0382 
PO3-85 120 4.0 2.5473 0.6083 0.6074 -0.0015 
PO3-68 117 6.0 4.5630 0.7607 0.7808  0.0258 
PO-25 118 5.0 3.3986 0.6610 0.7058  0.0634 
Pml-03 88 4.0 2.1863 0.5455 0.5426 -0.0052 
Pml-11 117 6.0 4.4539 0.8462 0.7755 -0.0911 
Pml-04 119 3.0 1.8318 0.3866 0.4541  0.1487 
PO-71 113 8.0 2.3280 0.5575 0.5704  0.0227 
Pml-06 117 3.0 2.8569 0.6496 0.6500  0.0006 
       
Mean 114 4.45 2.7346 0.5846 0.5965  0.0253 
St. Dev.  1.76 0.9922 0.1390 0.1181  0.0742 
      
Vet Village (Feb-02) 
Locus N Alleles N
e
H
OBS
H
EXP
F 
Ppa-01 117 2.0 1.4950 0.3162 0.3311  0.0449 
Ppa-12 113 4.0 2.8442 0.5752 0.6484  0.1129 
Ppa-46 117 3.0 1.7828 0.4103 0.4391  0.0656 
PO3-85 116 4.0 2.9129 0.5603 0.6567  0.1467 
PO3-68 117 5.0 3.1329 0.7350 0.6808 -0.0797 
PO-25 103 5.0 3.2830 0.6602 0.6954  0.0506 
Pml-03 117 3.0 1.8017 0.4914 0.4450 -0.1043 
Pml-11 112 6.0 3.5216 0.7857 0.7160 -0.0973 
Pml-04 115 3.0 1.9856 0.5826 0.4964 -0.1737 
PO-71 115 7.0 3.4257 0.7391 0.7081 -0.0438 
Pml-06 117 3.0 1.8590 0.5385 0.4621 -0.1653 
       
Mean 114 4.09 2.5496 0.5812 0.5707 -0.0220 
St. Dev.  1.51 0.7658 0.1438 0.1375  0.1114 
 
 
263
Table 19  Mantel tests for Gazebo and Vet Village grids along with the number  
of samples for each trapping period.  Individuals from the March 1999 trapping  
period did not amplify well during PCR reactions and were not used. 
 
Gazebo           Vet Village 
Period N     Rxy     P  Period N     Rxy      P 
Feb-95 145 0.076 0.003  Mar-95 103 0.049 0.081 
Apr-95 122 0.045 0.057  Apr-95 113 -0.010 0.651 
Jun-95   68 0.015 0.314  Jun-95   70 -0.089 0.994 
Feb-96   48 0.041 0.218  Mar-96   44 0.158 0.004 
Apr-96   52 0.028 0.281  Apr-96   72 0.059 0.058 
Jun-96   34 -0.095 0.872  Jun-96   44 -0.071 0.836 
Feb-97    9 -0.054 0.587  Mar-97   28 0.090 0.077 
Apr-98    9 -0.017 0.488  Feb-98   22 0.133 0.057 
Feb-99    9 -0.100 0.700  Mar-99     8   
Mar-00   32 0.011 0.099  Mar-00   26 0.160 0.053 
Nov-00   11 0.063 0.329  Dec-00    8 0.053 0.397 
Mar-01   22 0.365 0.001  Mar-01   43 0.240 0.001 
Nov-01   12 0.266 0.029  Oct-01   36 0.127 0.037 
Mar-02 120 0.052 0.023  Feb-02 117 0.106 0.001 
 
 
264
Table 20  F
ST
 values based on various models of subdivision within  
Gazebo and Vet Village grids.  The four corner, north/south and  
east/west models are based on sectioning the grid (Fig. 1).  The four  
other models are based on randomly assigning individuals into the  
four corner, north/south and east/west quadrants, random one two  
subpopulations (R12P), random one four subpopulations (R14P),  
random two two subpopulations (R22P) and random two four  
subpopulations. 
 
Gazebo      
Feb-95 Jun-95 Feb-96 Jun-96 Mar-02 
 F
ST
F
ST
F
ST
F
ST
F
ST
Four corner 0.0143 0.0005 0.0050 0.0187 0.0159 
North/South 0.0112 0.0048 0.0104 0.0220 0.0098 
East/West 0.0046 0.0069 0.0125 -0.0096 0.0151 
R12P -0.0042 -0.0111 -0.0037 0.0203 -0.0016 
R14P -0.0002 -0.0189 -0.0173 0.0006 -0.0009 
R22P -0.0020 0.0219 0.0098 -0.0057 -0.0012 
R24P 0.0002 0.0102 0.0080 -0.0107 -0.0005 
      
Vet Village      
 Mar-95 Jun-95 Mar-96 Jun-96 Mar-02 
F
ST
F
ST
F
ST
F
ST
F
ST
Four corner 0.0247 0.0009 0.0369 0.0082 0.0210 
North/South 0.0165 -0.0028 0.0244 -0.0026 -0.0029 
East/West 0.0125 0.0008 0.0092 0.0070 0.0231 
R12P 0.0025 0.0050 0.0034 0.0002 0.0005 
R14P 0.0008 0.0005 -0.0000 -0.0054 -0.0004 
R22P 0.0007 0.0015 0.0064 0.0337 0.0025 
R24P -0.0030 0.0078 0.0068 0.0158 0.0010 
 
 
265
 
266
Table 21  Summary statistics for the combined trapping periods of February  
1995 on Gazebo grid and March 1995 on Vet Village grid (1
st
-95) and April 
1995 on both Gazebo and Vet Village girds (2
nd
_95).  Included are sample 
size (N), number of alleles (Na), effective number of alleles (N
e
), observed 
heterozygosity (H
o
), expected heterozygosity (H
e
) and the fixation index (F
IS
). 
 
1st-95      
Locus      N Na N
e
H
o
H
e
F 
Ppa-01 232 3.00 1.737 0.375 0.424 0.116 
Ppa-12 240 5.00 2.719 0.667 0.632 -0.055 
Ppa-46 191 3.00 2.138 0.545 0.532 -0.023 
PO3-85 245 5.00 2.731 0.657 0.634 -0.037 
PO3-68 242 7.00 4.503 0.764 0.778 0.017 
PO-25 245 5.00 2.650 0.645 0.623 -0.036 
Pml-03 195 5.00 2.538 0.549 0.606 0.095 
Pml-11 245 6.00 4.282 0.714 0.766 0.068 
Pml-04 221 3.00 1.821 0.489 0.451 -0.084 
PO-71 228 9.00 3.337 0.667 0.700 0.048 
Pml-06 244 3.00 2.690 0.639 0.628 -0.018 
   
Mean 230 4.909 2.832 0.610 0.616 0.008 
St. Dev. 19 1.921 0.895 0.111 0.113 0.652 
       
2nd-95 
Locus     N Na N
e
H
o
H
e
F 
Ppa-01 223 3.00 1.811 0.462 0.448 -0.032 
Ppa-12 227 5.00 2.785 0.643 0.641 -0.003 
Ppa-46 206 3.00 2.016 0.519 0.504 -0.031 
PO3-85 234 6.00 2.704 0.624 0.630 0.010 
PO3-68 231 7.00 4.725 0.771 0.788 0.023 
PO-25 234 6.00 2.644 0.641 0.622 -0.031 
Pml-03 204 5.00 2.519 0.529 0.603 0.122 
Pml-11 231 7.00 4.250 0.736 0.765 0.038 
Pml-04 219 3.00 1.832 0.475 0.454 -0.045 
PO-71 220 8.00 3.384 0.700 0.705 0.006 
Pml-06 234 3.00 2.681 0.615 0.627 0.018 
       
Mean 224 5.090 2.850 0.611 0.617 0.007 
St. Dev. 10 1.868 0.937 0.104 0.113 0.467 
 
 
267
Table 22  Summary statistics for the combined trapping periods of June 1995  
on Gazebo grid and Vet Village grid (3
rd
-95) along with February 1996 on  
Gazebo grid and March 1996 on Vet Village grid (1
st
-96).  Included are sample  
size (N), number of alleles (Na), effective number of alleles (N
e
), observed 
heterozygosity (H
o
), expected heterozygosity (H
e
) and the fixation index (F
IS
). 
 
3rd-95      
Locus      N Na N
e
H
o
H
e
F 
Ppa-01 128 3.00 1.891 0.422 0.471 0.105 
Ppa-12 133 5.00 2.738 0.699 0.635 -0.102 
Ppa-46 121 3.00 2.086 0.537 0.521 -0.032 
PO3-85 137 5.00 2.741 0.657 0.635 -0.034 
PO3-68 135 6.00 4.600 0.785 0.783 -0.003 
PO-25 136 5.00 2.483 0.625 0.597 -0.046 
Pml-03 117 5.00 2.645 0.479 0.622 0.230 
Pml-11 134 7.00 4.184 0.709 0.761 0.068 
Pml-04 131 3.00 1.714 0.466 0.417 -0.118 
PO-71 126 8.00 3.218 0.706 0.689 -0.025 
Pml-06 137 3.00 2.727 0.599 0.633 0.055 
   
Mean 130 4.818 2.821 0.608 0.615 0.009 
St. Dev. 7 1.722 0.893 0.118 0.112 0.100 
       
1st-96 
Locus       N Na N
e
H
o
H
e
F 
Ppa-01 91 3.00 1.837 0.462 0.456 -0.013 
Ppa-12 90 4.00 2.955 0.611 0.662 0.076 
Ppa-46 92 3.00 2.355 0.533 0.575 0.074 
PO3-85 92 5.00 2.966 0.717 0.663 -0.082 
PO3-68 90 6.00 4.929 0.833 0.797 -0.045 
PO-25 90 5.00 2.554 0.589 0.608 0.032 
Pml-03 53 4.00 2.883 0.528 0.653 0.191 
Pml-11 91 6.00 4.138 0.758 0.758 0.000 
Pml-04 92 3.00 2.008 0.554 0.502 -0.104 
PO-71 88 7.00 3.514 0.750 0.715 -0.048 
Pml-06 92 3.00 2.762 0.641 0.638 -0.005 
   
Mean 87 4.455 2.991 0.634 0.639 0.007 
St. Dev. 11 1.440 0.911 0.116 0.102 0.084 
 
 
268
Table 23  Summary statistics for the combined trapping periods of April 1996 
on both Gazebo and Vet Village grids (2
nd
-96) along with June 1996 on both  
Gazebo and Vet Village grids (3
rd
-96).  Included are sample size (N), number  
of alleles (Na), effective number of alleles (N
e
), observed heterozygosity (H
o
),  
expected heterozygosity (H
e
) and the fixation index (F
IS
). 
 
2nd-96       
Locus     N Na N
e
H
o
H
e
F 
Ppa-01 124 3.00 1.812 0.460 0.448 -0.026 
Ppa-12 123 4.00 2.937 0.642 0.660 0.026 
Ppa-46 124 3.00 2.345 0.669 0.574 -0.167 
PO3-85 124 5.00 2.661 0.621 0.624 0.005 
PO3-68 124 6.00 5.212 0.831 0.808 -0.028 
PO-25 120 5.00 2.667 0.658 0.625 -0.053 
Pml-03 100 5.00 2.792 0.610 0.642 0.050 
Pml-11 121 6.00 4.213 0.719 0.763 0.057 
Pml-04 121 4.00 1.942 0.521 0.485 -0.073 
PO-71 120 8.00 3.270 0.767 0.694 -0.104 
Pml-06 121 3.00 2.592 0.603 0.614 0.018 
   
Mean 120 4.727 2.949 0.646 0.631 -0.027 
St. Dev. 7 1.555 0.990 0.104 0.106 0.069 
       
3rd-96 
Locus N Na N
e
H
o
H
e
F 
Ppa-01 77 3.00 1.820 0.506 0.450 -0.125 
Ppa-12 76 6.00 2.997 0.632 0.666 0.052 
Ppa-46 77 4.00 2.351 0.597 0.575 -0.039 
PO3-85 76 5.00 2.850 0.645 0.649 0.007 
PO3-68 77 7.00 5.449 0.818 0.816 -0.002 
PO-25 75 8.00 3.049 0.707 0.672 -0.052 
Pml-03 68 7.00 2.618 0.574 0.618 0.072 
Pml-11 75 7.00 3.893 0.720 0.743 0.031 
Pml-04 78 5.00 1.959 0.615 0.490 -0.257 
PO-71 71 8.00 3.652 0.761 0.726 -0.047 
Pml-06 74 3.00 2.417 0.446 0.586 0.239 
   
Mean 75 5.727 3.005 0.638 0.636 -0.011 
St. Dev. 3 1.849 1.030 0.109 0.108 0.124 
 
 
269
Table 24  Summary statistics for the combined trapping periods of February  
1997 on Gazebo grid and March 1997 on Vet Village grid (1
st
-97) along with  
April 1998 on Gazebo grid and February 1998 on Vet Village grid (1
st
-98).   
Included are sample size (N), number of alleles (Na), effective number of  
alleles (N
e
), observed heterozygosity (H
o
), expected heterozygosity (H
e
) and  
the fixation index (F
IS
). 
 
1st-97       
Locus       N Na N
e
H
o
H
e
F 
Ppa-01 37 2.00 1.931 0.541 0.482 -0.121 
Ppa-12 36 4.00 2.546 0.583 0.607 0.039 
Ppa-46 37 3.00 1.899 0.432 0.473 0.086 
PO3-85 37 4.00 3.195 0.649 0.687 0.056 
PO3-68 35 6.00 4.605 0.829 0.783 -0.058 
PO-25 37 4.00 2.733 0.568 0.634 0.105 
Pml-03 36 4.00 3.149 0.694 0.682 -0.018 
Pml-11 37 5.00 3.154 0.757 0.683 -0.108 
Pml-04 37 4.00 1.892 0.459 0.472 0.026 
PO-71 34 6.00 2.934 0.735 0.659 -0.115 
Pml-06 37 3.00 2.651 0.541 0.623 0.132 
   
Mean 36 4.090 2.790 0.617 0.617 0.002 
St. Dev. 1 1.221 0.785 0.126 0.102 0.092 
       
1st-98 
Locus       N Na N
e
H
o
H
e
F 
Ppa-01 30 3.00 2.062 0.567 0.515 -0.100 
Ppa-12 29 4.00 2.470 0.448 0.595 0.247 
Ppa-46 31 4.00 2.232 0.452 0.552 0.182 
PO3-85 30 4.00 3.377 0.600 0.704 0.148 
PO3-68 30 5.00 3.774 0.767 0.735 -0.043 
PO-25 30 4.00 2.635 0.733 0.621 -0.182 
Pml-03 31 4.00 3.579 0.677 0.721 0.060 
Pml-11 30 5.00 3.279 0.667 0.695 0.041 
Pml-04 31 3.00 1.750 0.516 0.429 -0.204 
PO-71 28 6.00 2.751 0.714 0.636 -0.122 
Pml-06 30 3.00 2.469 0.700 0.595 -0.176 
   
Mean 30 4.090 2.762 0.622 0.618 -0.014 
St. Dev. 1 0.944 0.657 0.113 0.095 0.159 
 
 
270
Table 25  Summary statistics for the combined trapping periods of March 
2000 on Gazebo and Vet Village grids (1
st
-00) along with November 2000  
on Gazebo grid and December 2000 on Vet Village grid (2
nd
-00).  Included are  
sample size (N), number of alleles (Na), effective number of alleles (N
e
),  
observed heterozygosity (H
o
), expected heterozygosity (H
e
) and the fixation  
index (F
IS
). 
 
1st-00       
Locus       N Na N
e
H
o
H
e
F 
Ppa-01 58 2.00 1.931 0.500 0.482 -0.037 
Ppa-12 56 3.00 2.305 0.500 0.566 0.117 
Ppa-46 58 3.00 2.086 0.621 0.521 -0.192 
PO3-85 57 5.00 2.841 0.649 0.648 -0.002 
PO3-68 57 5.00 3.272 0.649 0.694 0.065 
PO-25 55 4.00 2.606 0.636 0.616 -0.033 
Pml-03 58 4.00 1.822 0.431 0.451 0.044 
Pml-11 58 6.00 3.528 0.793 0.717 -0.107 
Pml-04 57 3.00 1.875 0.333 0.467 0.286 
PO-71 51 6.00 2.998 0.588 0.666 0.117 
Pml-06 57 3.00 2.346 0.614 0.574 -0.070 
   
Mean 56 4.00 2.510 0.574 0.582 0.017 
St. Dev. 2 1.342 0.586 0.125 0.094 0.130 
       
2nd-00 
Locus       N Na N
e
H
o
H
e
F 
Ppa-01 16 2.00 1.753 0.375 0.430 0.127 
Ppa-12 19 3.00 2.625 0.526 0.619 0.150 
Ppa-46 19 3.00 1.774 0.263 0.436 0.397 
PO3-85 18 4.00 2.445 0.667 0.591 -0.128 
PO3-68 19 6.00 3.297 0.737 0.697 -0.058 
PO-25 17 4.00 3.879 0.588 0.742 0.207 
Pml-03 19 4.00 1.563 0.263 0.360 0.269 
Pml-11 19 4.00 3.557 0.737 0.719 -0.025 
Pml-04 19 2.00 1.978 0.368 0.494 0.255 
PO-71 19 6.00 4.102 0.579 0.756 0.234 
Pml-06 17 3.00 2.014 0.412 0.503 0.182 
   
Mean 18 3.727 2.635 0.501 0.577 0.146 
St. Dev. 1 1.348 0.923 0.176 0.140 0.158 
 
 
271
Table 26  Summary statistics for the combined trapping periods of March 
2001 on both Gazebo and Vet Village grids (1
st
-01) along with November 2001  
on Gazebo grid and October 2001 on Vet Village grid (2
nd
-01).  Included are  
sample size (N), number of alleles (Na), effective number of alleles (N
e
),  
observed heterozygosity (H
o
), expected heterozygosity (H
e
) and the fixation  
index (F
IS
). 
 
1st-01       
Locus       N Na N
e
H
o
H
e
F 
Ppa-01 63 2.00 1.630 0.397 0.387 -0.026 
Ppa-12 61 3.00 2.657 0.574 0.624 0.080 
Ppa-46 65 3.00 2.109 0.462 0.526 0.122 
PO3-85 64 4.00 2.654 0.578 0.623 0.072 
PO3-68 65 6.00 3.303 0.662 0.697 0.051 
PO-25 53 4.00 2.853 0.717 0.650 -0.104 
Pml-03 65 4.00 1.875 0.462 0.467 0.011 
Pml-11 62 6.00 3.710 0.726 0.730 0.006 
Pml-04 65 3.00 1.978 0.538 0.495 -0.089 
PO-71 64 7.00 3.542 0.750 0.718 -0.045 
Pml-06 65 3.00 2.305 0.631 0.566 -0.114 
   
Mean 63 4.090 2.602 0.591 0.589 -0.003 
St. Dev. 4 1.578 0.698 0.118 0.111 0.080 
       
2nd-01 
Locus       N Na N
e
H
o
H
e
F 
Ppa-01 48 2.00 1.627 0.354 0.385 0.081 
Ppa-12 48 3.00 2.395 0.542 0.582 0.070 
Ppa-46 48 3.00 1.957 0.458 0.489 0.063 
PO3-85 48 4.00 2.748 0.479 0.636 0.247 
PO3-68 48 6.00 3.583 0.750 0.721 -0.040 
PO-25 48 6.00 3.689 0.708 0.729 0.028 
Pml-03 48 5.00 1.863 0.479 0.463 -0.034 
Pml-11 48 5.00 3.704 0.771 0.730 -0.056 
Pml-04 48 3.00 1.822 0.583 0.451 -0.293 
PO-71 48 6.00 2.810 0.750 0.644 -0.164 
Pml-06 48 3.00 2.182 0.563 0.542 -0.038 
   
Mean 48 4.182 2.580 0.585 0.579 -0.012 
St. Dev. 0 1.471 0.785 0.141 0.122 0.140 
 
 
272
Table 27  Summary statistics for the combined trapping periods of March 
2002 on Gazebo grid and February 2002 on Vet Village grid (1
st
-02).   
Included are sample size (N), number of alleles (Na), effective number of  
alleles (N
e
), observed heterozygosity (H
o
), expected heterozygosity (H
e
)  
and the fixation index (F
IS
). 
 
1st-02       
Locus      N Na N
e
H
o
H
e
F 
Ppa-01 236 2.00 1.685 0.390 0.407 0.041 
Ppa-12 223 5.00 2.568 0.507 0.611 0.170 
Ppa-46 237 3.00 1.887 0.464 0.470 0.013 
PO3-85 236 4.00 2.737 0.585 0.635 0.079 
PO3-68 234 6.00 3.917 0.748 0.745 -0.004 
PO-25 221 6.00 3.412 0.661 0.707 0.065 
Pml-03 204 4.00 1.961 0.515 0.490 -0.050 
Pml-11 229 6.00 4.138 0.817 0.758 -0.077 
Pml-04 234 4.00 1.928 0.483 0.481 -0.003 
PO-71 228 9.00 2.957 0.649 0.662 0.019 
Pml-06 234 3.00 2.426 0.594 0.588 -0.011 
   
Mean 229 4.727 2.692 0.583 0.596 0.022 
St. Dev. 10 1.954 0.840 0.128 0.120 0.067 
 
 
273
Table 28  Mantel tests for the combined trapping  
periods along with the number of samples for each  
trapping period. 
 
Period           N                 Rxy                  P 
1st-95 248 0.079 0.001
2nd-95 235 0.113 0.001
3rd-95 138 0.099 0.001
1st-96 92 0.268 0.001
2nd-96 124 0.108 0.001
3rd-96 78 0.034 0.096
1st-97 37 0.326 0.001
1st-98 31 0.254 0.003
1st-00 58 0.128 0.001
2nd-00 19 0.070 0.153
1st-01 65 0.127 0.001
2nd-01 48 0.444 0.001
1st-02 237 0.140 0.001
 
 
274
Table 29  Between grid measures of differentiation that include F
ST
 and R
ST
 and  
the number of migrants per generation based on F
ST
 ranges from 9.90 to infinity.   
Also, between grid genetic differentiation as estimated by AMOVA (% variation). 
 
Period N F
ST
P-value SE R
ST
N
M
 (F
ST
) % Var. 
1st-95 248 0.019 0.000 0.000 0.019 25.66 1.91 
2nd-95 235 0.019 0.000 0.000 0.019 25.97 1.89 
3rd-95 138 0.009 0.002 0.001 0.010 52.46 0.94 
1st-96 92 0.012 0.009 0.003 0.012 41.22 1.20 
2nd-96 124 0.006 0.013 0.003 0.006 80.59 0.62 
3rd-96 78 0.004 0.137 0.013 0.004 127.34 0.39 
1st-97 37 0.046 0.009 0.003 0.048 10.44 4.57 
1st-98 31 -0.004 0.568 0.016 0.000 Inf. 0.00 
1st-00 58 0.051 0.000 0.000 0.053 9.40 5.05 
2nd-00 19 0.038 0.128 0.012 0.039 12.68 3.79 
1st-01 65 0.032 0.001 0.001 0.033 15.15 3.19 
2nd-01 48 0.033 0.011 0.003 0.034 14.64 3.30 
1st-02 237 0.038 0.000 0.000 0.040 12.66 3.80 
 
 
275
Table 30  Population assignment using the jackknife  
method within program WHICHRUN  
(Banks & Eichert 2000) for each grid by trapping  
period. 
 
Trap period Grid             n    % correct
1st-95 Gazebo 145 77.2
1st-95 Vet Village 103 71.8
2nd-95 Gazebo 122 76.2
2nd-95 Vet Village 113 74.3
3rd-95 Gazebo 68 67.6
3rd-95 Vet Village 70 67.1
1st-96 Gazebo 48 64.5
1st-96 Vet Village 44 59.1
2nd-96 Gazebo 52 67.3
2nd-96 Vet Village 72 69.4
3rd-96 Gazebo 34 64.7
3rd-96 Vet Village 44 70.5
1st-00 Gazebo 32 75.0
1st-00 Vet Village 26 73.1
1st-01 Gazebo 22 63.6
1st-01 Vet Village 43 69.8
1st-02 Gazebo 120 81.7
1st-02 Vet Village 117 81.2
 
  276
CUMULATIVE REFERENCES 
 
Albrecht I, Schulte-Hostedde H, Gibbs L, et al. (2001) Microgeographic genetic structure 
in the yellow-pine chipmunk (Tamias amoenus).  Molecular Ecology, 10, 1625-1631. 
Arbogast BS (1999) Mitochondrial DNA phylogeography of the New World flying 
squirrels (Glaucomys): implications for Pleistocene biogeography. Journal of 
Mammalogy, 80, 142-155.  
Avise JC (2000) History and purview of phylogeography. In: Phylogeography: The 
History and Formation of Species (ed. Avise JC), pp. 3-36. Harvard University Press, 
London. 
Avise JC, Arnold J, Ball RM, Jr et al. (1987) Intraspecific phylogeography: the 
mitochondrial DNA bridge between population genetics and systematics. Annual 
Review of Ecology and Systematics, 18, 489-522. 
Avise JC, Lansman RA, Shade RO (1979) The use of restriction endonucleases to 
measure mitochondrial DNA sequence relatedness in natural populations. I. Population 
structure and evolution in the genus Peromyscus.  Genetics, 92, 279-295. 
Avise JC, Shapira JF, Daniel SW, Aquadro CF, Lansman RA (1983) Mitochondrial DNA 
differentiation during the speciation process in Peromyscus. Molecular Biology and 
Evolution, 1, 38-56. 
  277
Avise JC, Walker D, Johns GC (1998) Speciation durations and Pleistocene effects on 
vertebrate phylogeography. Proceedings of the Royal Society of London, B 267, 1707-
1712. 
Avise JC (1996) Space and time as axes in intraspecific phylogeography. In: Past and 
Future Environmental Changes: The Spatial and Evolutionary Responses of 
Terrestrial Biota (eds Huntly B, Cramer W, Morgan AV, Prentice HC, Allen JRM), 
pp. 381-388. Springer-Verlag, New York. 
Banks MA, Eichert W (2000) WHICHRUN (version 4.1) a computer program for 
population assignment of individuals based on multilocus genotype data. Journal of 
Heredity, 91, 87-89 
Bengtsson BO (1978) Avoid inbreeding: At what cost? Journal of Theoretical Biology, 
73, 439-444. 
Blair WF (1950) Ecological factors in speciation of Peromyscus. Evolution, 4, 253-275. 
Blondel J, Aronson J (1999) Biology and wildlife of the Mediteranean region. Oxford 
University Press, New York. 
Bowen WW (1968) Variation and Evolution of Gulf Coast populations of beach mice, 
Peromyscus polionotus. Bulletin of the Florida State Museum, 12, 1-91. 
Brookfield JFY (1996) A simple new method for estimating null allele frequency from 
heterozygote deficiency.  Molecular Ecology, 56, 45-47. 
Brown WM, George M, Jr, Wilson AC (1979) Rapid evolution of animal mitochondrial 
DNA. Proceedings of the National Academy of Sciences of the USA, 76, 1967-1971. 
Cain SA (1944) Foundations of plant geography.  Harper and Brothers, New York. 
  278
Chesser RK, Smith MH, Brisbin Jr IL (1980) Management and maintenance of genetic 
variability in endangered species.  International Zoo Yearbook, 20, 146-154. 
Chesser RK, Smith MH, Brisbin Jr IL (1980) Management and maintenance of genetic 
variability in endangered species.  International Zoo Yearbook, 20, 146-154. 
Chesser RK (1983) Genetic variability within and among populations of the black-tailed 
prairie dog. Evolution, 37, 320-331. 
Chesser RK (1991a) Gene diversity and female philopatry. Genetics, 127, 437-447. 
Chesser RK (1991b) Influence of gene flow and breeding tactics on gene diversity within 
populations.  Genetics, 129, 573-583. 
Chesser RK (1998) Relativity of Behavioral Interactions in Socially Structured 
Populations. Journal of Mammalogy, 79, 713-724. 
Chirhart SE, Honeycutt RL, Greenbaum IF (2000) Microsatellite markers for the deer 
mouse Peromyscus maniculatus. Molecular Ecology, 9, 1669-1671. 
Christiansen FB (1974) Sufficient conditions for protected polymorphism in a subdivided 
population.  The American Naturalist, 108, 157-166. 
Clement M, Posada D and Crandall K. 2000. TCS: a computer program to estimate gene 
genealogies. Molecular Ecology 9(10): 1657-1660 
Delcourt PA, Delcourt HR (1981) Vegetation for eastern North America: 40,000 yr B. P. 
to the present. In Geobotany II  (ed. Roamns RC), pp. 123-165. Plenum Press, New 
York. 
Dobson FS (1998) Social structure and gene dynamics in mammals.  Journal of 
Mammalogy, 79, 667-670. 
  279
Donoghue JF, Stapor FW, Tanner WF (1998) Discussion of: Otvos EG (1995) Multiple 
Pliocene-Quaternary marine highstands, northeast Gulf coastal plain - fallacies and 
facts. Journal of Coastal Research, 11, 984-1002. Journal of Coastal Research, 14, 
669-674. 
Double MC, Peakall R, Beck NR, Cockburn A (2005) Dispersal, philopatry, and 
infidelity: dissecting local genetic structure in superb fairy-wrens (Malurus cyaneus).  
Evolution, 59, 625-635. 
Ellsworth DL, Honeycutt RL, Silvy NJ, Bickham JW, Klimstra WD (1994) Historical 
biogeography and contemporary patterns of mitochondrial DNA variation in white-
tailed Deer from the Southeastern United States. Evolution, 48, 122-136. 
Emerson BC, Paradis E, Th?baud C (2001) Revealing the demographic histories of 
species using DNA sequences. Trends in Ecology and Evolution, 16, 707-716. 
Epperson BK (1990) Spatial autocorrelation of genotypes under directional selection.  
Genetics, 124, 757-771. 
Epperson BK, Li TQ (1997) Gene dispersal and spatial genetic structure.  Evolution, 51, 
672-681. 
Excoffier L, Smouse P, et Quattro, JM (1992) Analysis of molecular variance inferred 
from metric distances among DNA haplotypes: Application to human mitochondrial 
DNA restriction data. Genetics, 131, 479-491. 
Felsenstein J (1981) Evolutionary trees from DNA sequences: a maximum likelihood 
approach. Journal of Molecular Evolution, 17, 368-376.  
Felsenstein J (1985) Confidence limits on phylogenies: an approach using the bootstrap. 
Evolution, 39, 783-791.  
  280
Fisher RA (1930) The Genetical Theory of Natural Selection. Clarendon Press, Oxford. 
Foltz DW (1981) Genetic Evidence for long-term monogamy in a small rodent, 
Peromyscus polionotus. The American Naturalist, 117, 665-675. 
Ford EB (1954) Problems in the evolution of geographic races. In: Evolution as a process 
(eds. Huxley J, Hardy AC, Ford EB), pp.99-108. Allen and Unwin, London.  
Ford EB (1960) Evolution in process. In: Evolution after Darwin (ed. Sol Tax), pp. 181-
196. University of Chicago Press, Chicago. 
Fu Y-X, Li W-H (1993) Statistical tests of neutrality of mutations. Genetics, 133, 693-
709. 
Fu Y-X (1997) Statistical tests of neutrality of mutations against population growth, 
hitchhiking and background selection. Genetics, 147, 915-925. 
Gates DM (1993) Climate Change and its Biological Consequences. Sinauer Associates, 
Sunderland Massachusetts. 
Gerlach G, Musolf K (2000) Fragmentation of landscape as a cause for genetic 
subdivision in bank voles. Conservation Biology, 14, 1066-1074. 
Gibbs HL, Prior KA, Weatherhead PJ, et al. (1997) Genetic structure of populations of 
the thretened eastern massasauga rattlesnake, Sistrurus c. catenatus: evidence from 
microsatellite DNA markers.  Molecular Ecology, 6, 1123-1132. 
Giles BE, Goudet J (1997) Genetics differentiation in Silene dioica metapopulations: 
estimation of spatio-temporal effects in a successional plant species. American 
Naturalist, 149, 507-526. 
  281
Graham RW, Lundelius EL Jr. (1984) Coevolutionary disequilibrium and Pleistocene 
extinctions. In: Quaternary Extinctions (eds. Martin PS, Klein RG), pp. 223-249. 
University of Arizona Press, Tucson. 
Greenwood PJ (1980) Mating systems, philopatry and dispersal in birds and mammals. 
Animal Behavior, 28, 1140-1162. 
Guo SW, Thompson EA (1992) Performing the exact test of Hardy-Weinberg proportions 
for multiple alleles.  Biometrics, 48, 361-372. 
Haldane JBS (1948) The theory of a cline. Journal of Genetics, 48, 227-284. 
Hall ER (1981) The mammals of North America. New York, John Wiley and Sons. 
Hall TA (1999) BioEdit: a user-friendly biological sequence alignment editor and 
analysis program for Windows 95/98/NT.  Nucleic Acids Symposium Series, 41, 95-98. 
Hamrick JL, Nason JD (1996) Consequences of dispersal in plants. In: Spatial and 
Temporal Aspects of Population Processes. (eds. Rhodes OE, Chesser R, Smith M),  
pp. 203-236. University of Chicago Press, Chicago. 
Haffer J (1969) Speciation in Amazonian forest birds. Science, 165, 131-137. 
Hartl DL, Clark AG (1989) Principles of population genetics. Sinauer, Sunderland 
Massachusetts. 
Hasegawa M, Kishino H, Yano T (1985) Dating of the human-ape splitting by a 
molecular clock of mitochondrial DNA. Journal of Molecular Evolution, 22, 160-174. 
Hays JP, Harrison RG (1992) Variation in mitochondrial DNA and the biogeographic 
history of woodrats (Neotoma) of the Eastern United States. Systematic Biology, 43, 
331-344. 
  282
Hewitt G M (1996) Some genetic consequences of ice ages, and their role in divergence 
and speciation. Biological Journal of the Linnean Society, 58, 247-276. 
Hewitt G M (2000) The genetic legacy of the Quaternary ice ages. Nature, 405, 907-913. 
Hewitt G M (1993) Postglacial distribution and species substructure: lessons from pollen, 
insects and hybrid zones. In: Evolutionary Patterns and Processes (eds. Lees DR, 
Edwards D), pp. 97-123. Academic Press, San Diego, California. 
Hibbard CW (1968) Paleontology, in Biology of Peromyscus (Rodentia). In: Special 
Publication Number 2 (ed. King JA), pp. 27-74.  American Society of Mammalogists. 
Hoffman EA, Blouin MS (2004) Evolutionary history of the Northern leopard frog: 
reconstruction of phylogeny, phylogeography, and historical changes in population 
demography from mitochondrial DNA. Evolution, 58, 145-159. 
Howell AH (1920) Description of a new species of beach mouse from Florida. Journal of 
Mammalogy, 1, 237-240. 
Hudson RR (1989) How often are polymorphic sites due to a single mutation?  
Theoretical Population Biology, 36, 23-33. 
Huelsenbeck JP, Ronquist FR (2001) MrBayes: Bayesian inference for phylogeny. 
Biometrics 17, 754-756. 
Huelsenbeck JP, Bull JJ, Cunningham CW (2002) Combining data in phylogenetic 
analysis. Trends in Evolution and Ecology, 11, 152-158. 
Huxley JS (1943) Evolution; the modern synthesis. Harper and Brothers, New York.  
Jarne P, Lagoda PJL (1996) Microsatellites, from molecules to populations and back. 
Trends in Ecology and Evolution, 11, 424-429. 
  283
Karlin S, Campbell RB (1980) Polymorphism in subdivided populations characterized by 
a major and subordinate demes.  Heredity, 44, 151-168. 
Kimura M (1955) Solution of a process of random genetic drift with a continuous model. 
Proceedings of the National Academy of Science of the USA, 41, 144-150. 
Kimura M, Weiss GH (1964) The stepping stone model of population structure and the 
decrease of genetic correlation with distance. Genetics, 49, 561-576. 
Kretzmann MB, Gemmell NJ, Meyer A (2001) microsatellite analysis of population 
structure  in the endangered Hawaiian monk seal.  Conservation Biology, 15, 457-466. 
Kuhner MK, Yamato J, Felsenstein (1998) Maximum likelihood estimation of population 
growth rates based on the coalescent. Genetics, 149, 429-434. 
Kuhner MK, Yamato J, Felsenstein J (1995) Estimating effective population size and 
mutation rate from sequence data using Metropolis-Hastings sampling. Genetics, 140, 
1421-1430.   
Lanave C, Preparata G, Saccone C, Serio G (1984) A new method for calculating 
evolutionary substitution rates. Journal of molecular evolution, 20, 86-93.  
Levins R (1969) Some demographic and genetic consequences of environmental 
heterogeneity for biological control. Bulletin of the Entomological Society of America, 
15, 237-240. 
Li W-H, Graur D (1991) Fundamentals of Molecular Evolution. Sinauer Associates, 
Sunderland, Massachusetts. 
Maruyama T (1970) Rate of decrease of genetic variability in a subdivided population.  
Biometrika, 57, 299-311. 
  284
Mayr E (1942) Systematics in the origin of species. Columbia University Press. New 
York. 
Mayr E (1954) Change of genetic environment and evolution. In: Evolution as a process  
(eds. Huxley J, Hardy AC, Ford EB), pp. 157-180. Allen and Unwin, London. 
McDonald JH, Kreitman M (1991) Adaptive protien evolution at the Adh locus in 
Drosophila. Nature, 351, 652-654. 
Mullen LM, Hirschmann RJ, Prince KL et al. (2005) Sixty polymorphic microsatellite 
markers for the oldfield mouse developed in Peromyscus polionotus and Peromyscus 
maniculatus.  Molecular Ecology Notes, in press. 
Nei M, Maruyama T, Chakraborty R (1975) The bottleneck effect and genetic variability 
in populations.  Evolution, 29, 1-10. 
Nei M (1987) Molecular Evolutionary Genetics. Columbia University Press, New York. 
Nylander J.A.A. (2002) MrModeltest version 1.1b. http://morphbank.ebc.uu.se 
/systzoo/staff/nylander.html. 
Oli MK, Holler NR, Wooten MC (2001) Viability analysis of endangered Gulf Coast 
beach mice (Peromyscus polionotus) populations.  Biological Conservation, 97, 107-
118. 
Osgood WH (1909) Revision of the mice the American genus Peromyscus polionotus. 
North American Fauna, 28, 1-285. 
Otis DL, Burnham KP, White GC et al. (1978) Statistical inference from capture data on 
closed animal populations.  Wildlife Monographs, 62, 1-135. 
Otvos EG (1995) Multiple Pliocene-Quaternary marine highstands, northeast Gulf 
coastal-plain-fallacies and facts. Journal of Coastal Research, 11, 984-1002. 
  285
Paetkau D, Calvert W, Stirling I, Strobeck C (1995) Microsatellite analysis of population 
structure in polar bears. Molecular Ecology, 4, 347-354.  
Palumbi, JP,  Martin A, Romano S et al. (1991) Simple fool?s guide to PCR.  Department 
of Zoology and Kewalo Marine Laboratory, University of Hawaii, Honolulu, HI. 
Patton JL, Yang SY (1977) Genetic variation in Thomomys bottae pocket gophers: 
macrogeographic patterns.  Evolution, 31, 697-720. 
Peakall R, Smouse PE (2005) GenAlEx 6: Genetic Analysis in Excel. Population genetic 
software for teaching and research.  Australian National University, Canberra, 
Australia.  http://www.anu.edu.au/BoZo/GenAlEx/
Peakall R, Ruibal M, Lindenmayer DB (2003) Spatial autocorrelation analysis offers new 
insights into gene flow in the Australian bush rat, Rattus fuscipes.  Evolution, 57, 
1182-1195. 
Petit JR, Jouzel J, Raynaud D (1999) Climate and atmospheric history of the past 420,000 
years from the Vostok ice core, Antarctica. Nature, 399, 429-436. 
Pfenninger M, Posada D (2002) Phylogeographic history of the land snail Candidula 
unifasciata (Helicellinae, Stylommatophora): fragmentation, corridor migration, and 
secondary contact.  Evolution, 56, 1776-1788. 
Pielou EC (1991) After the ice age: return of life to glaciated North America. University 
of Chicago Press, Chicago, Illinois. 
Posada D, Crandall KA (1998) MODELTEST: testing the model of DNA substitution. 
Bioinformatics, 14, 817-818. 
  286
Posada D, Crandall KA, Templeton AR (2000) GeoDis: a program for the cladistic 
nested analysis of the geographical distribution of genetic haplotypes. Molecular 
Ecology, 9, 487-488. 
Prince KL, Glenn TC, Dewey MJ (2002) Cross-species amplification among 
peromyscines of new microsatellite DNA loci from the oldfield mouse (Peromyscus 
polionotus subgriseus) Molecular Ecology Notes, 2, 133-136. 
Raymond M, Rousset F (1995) GENEPOP, version 3.4: an updated version of 
GENEPOP (version 1.2), population genetics software for exact tests and ecumenism.  
Journal of Herdity, 86, 248-249. 
Rohlf FJ, Schnell GD (1971) An investigation of the isolation-by-distance model.  The 
American Naturalist, 19, 473-511. 
Root JJ, Black WC IV, Calisher CH et al. (2003) Analyses of gene flow among 
populations of deer mice (Peromyscus polionotus) at sites near hantavirus pulmonary 
syndrome case-patient residences.  Journal of Wildlife Diseases, 39, 287-298. 
Ross KG, Krieger MJB, Shoemaker DD, et al. (1997) Hierarchical analysis of genetic 
structure in native fire ant populations: results from three classes of molecular 
markers.  Genetics, 147, 643-655. 
Rozas J, S?nchez-DelBarrio JC, Messeguer X, Rozas R (2003) DnaSP, DNA 
polymorphism analysis by the coalescent and other methods. Bioinformatics, 19, 2496-
2497. 
Ruez DR, Jr (2001) Early Irvingtonian (latest Pliocene) rodents from Inglis 1C, Citrus 
County, Florida. Journal of Vertebrate Paleontology, 21, 153-171. 
SAS Institute (2000) SAS System for Windows v.9.10.  SAS Institute, Cary, NC, USA. 
  287
Sbisa E, Tanzariello F, Reyes A, et al. (1997) Mammalian mitochondrial D-loop region 
structural analysis: identification of new conserved sequences and their functional and 
evolutionary implications. Gene, 205, 125-140. 
Schneider S, Roessli D, Excoffier L (2000) ARLEQUIN v. 2.0 A software for population 
genetic data analysis. University of Geneva: Genetics and Biometry Laboratory. 
Scribner KT, Chesser RK (1993) Environmental and demographic correlates of spatial 
and seasonal genetic structure in the eastern cottontail (Sylvilagu floridanus).  Journal 
of Mammalogy, 74, 1026-1044. 
Selander RK (1970a) Biochemical polymorhpism in populations of the house mouse and 
old-field mouse. Symposium of the Zoological Society of London, 26, 73-91. 
Selander RK (1970b) Behavior and genetic variation in natural populations. American 
Zoologist, 10, 53-66. 
Selander RK, Smith MH, Yang SY, Johnson WE, Gentry JB (1971) Biochemical 
polymorphism and systematics in the genus Peromyscus. I. Variation in the oldfield 
mouse (Peromyscus polionotus). Studies in Genetics, 6, 49-90.  
Slatkin M (1976) The rate and spread of an advantageous allele in a subdivided 
population. In: Population Genetics and Ecology (eds Karlin S, Nevo E), pp. 767-780. 
Academic Press. New York. 
Slatkin M (1977) Gene flow and genetic drift in a species subject to frequent local 
extinctions. Theoretical Population Biology, 12, 253-262. 
Slatkin M (1985) Gene flow in natural population.  Annual review of Ecology and 
Systematics, 16, 393-430. 
  288
Smouse PE, Long JC (1992) Matrix correlation analysis in anthropology and genetics.  
Yearbook of Physical Anthropology, 35, 187-213. 
Smouse PE, Peakall R (1999) Spatial autocorrelation analysis of individual multiallele 
and multilocus genetic structure, Heredity, 82, 561-773. 
Smouse PE, Long JC, Sokal RR (1986) Multiple regression and correlation extensions of 
the Mantel test of matrix correspondence.  Systematic Zoology, 35, 627-632. 
Sober E (1988) Reconstructing the past: Parsimony, Evolution, and Inference. MIT Press, 
Cambridge, Mass. 
Sokal RR, Oden NL (1978a) Spatial autocorrelation in biology. 1. Methodology. 
Biological Journal of the Linnean Society, 10, 199-228. 
Sokal RR, Oden NL (1978b) Spatial autocorrelation in biology. 2. Some biological 
implications and four applications of evolutionary and ecological interest.  Biological 
Journal of the Linnean Society, 10, 229-249. 
Sokal RR, Wartenberg DE (1983) A test of spatial autocorrelation analysis using an 
isolation-by-distance model.  Genetics, 105, 219-237. 
Sokal RR, Jacquez GL, Wooten MC (1989) Spatial autocorrelation analysis of migration 
and selection.  Genetics, 121, 845-855. 
Sumner FB (1926) An analysis of geographic variation in mice of the Peromyscus 
polionotus group from Florida and Alabama. Journal of Mammalogy, 7, 149-184. 
Sumner FB (1929) An analysis of a concrete case of intergradation between two 
subspecies. Proceedings of the National Academy of Sciences of the USA, 15, 11-120. 
  289
Swilling RS Jr, Wooten MC, Holler NR, Lynn WJ (1998) Population dynamics of 
Alabama beach mice (Peromyscus polionotus ammobates) following hurricane Opal.  
The American Midland Naturalist, 140, 287-298. 
Swilling RS Jr, Wooten MC (2000) Subadult dispersal in a monogamous species: the 
Alabama beach mouse (Peromyscus polionotus ammobates). Journal of Mammalogy, 
83, 252-259. 
Swofford DL (2002) PAUP*: phylogenetic analysis using parsimony (* and other 
methods). Ver. 4.0b10. Sinauer Associates, Sunderland, MA. 
Tajima F (1989) Statistical method for testing the neutral mutation hypothesis by DNA 
polymorphism. Genetics, 123, 585-595. 
Templeton AR (1998) Nested clade analyses of phylogeographic data: testing hypotheses 
about gene flow and population history.  Molecular Ecology, 7. 381-397. 
Templeton AR, Boerwinkle E, Sing CF (1987) A cladistic analysis of phenotypic 
associations with haplotypes inferred from restriction endonuclease mapping. I. Basic 
theory and an anlysis of alcohol dehydrogenase activity in Drosophila.  Genetics 117, 
343-351. 
Templeton AR, Crandall A, Sing CF (1992) A cladistic analysis of phenotypic 
associations with haplotypes inferred from restriction endonuclease mapping and DNA 
sequence data. III. Cladogram estimation.  Genetics 132, 619-633. 
Templeton AR, Sing CF (1993) A cladistic analysis of phenotypic associations with 
haplotypes inferred from restriction endonuclease mapping. IV. Nested analysis with 
cladogram uncertainty and recombination.  Genetics, 134, 659-669. 
  290
Templeton AR, Routman E, Phillips CA (1995) Seperating population structure from 
population history: a cladistic analysis of the geographical distribution of 
mitochondrial DNA haplotypes in the tigar salamander, Ambystoma tigrinum. 
Genetics, 140, 767-782. 
Thompson,J.D., Gibson,T.J., Plewniak,F., Jeanmougin,F. and Higgins,D.G. (1997) The 
ClustalX windows interface: flexible strategies for multiple sequence alignment aided 
by quality analysis tools. Nucleic Acids Research, 24:4876-4882. 
Tiedemann R, Hardy O, Vekemans X, Milinkovitch MC (2000) Higher impact of female 
than male migration on population structure in large mammals. Molecular Ecology, 9, 
1159-1163. 
Turner ME, Stephens JC, Anderson WW (1982) Homozygosity and patch structure in 
plant populations as a result of nearest-neighbor pollination. Proceedings of the 
National Academy of Science of the USA, 79, 203-207. 
van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P (2004) MICRO-CHECKER: 
software for identifying and correcting genotyping errors in microsatellite data.  
Molecular Ecology Notes, 4, 535-538. 
van Staaden MJ, Michener GR, Chesser RK (1996) Spatial analysis of microgeographic 
genetic structure in Richardson?s ground squirrels.  Canadian Journal of Zoology, 
74,1187-1195. 
Veith M, Kosuch J, Vences M (2003) Climatic oscillations triggered post-Messinian 
speciation of Western Palearctic brown frogs (Amphibia, Ranidae). Molecular 
Phylogenetics and Evolution, 26, 310-327. 
  291
Vigilant LA, Stoneking M, Harpending H, Hawkes K, Wilson AC (1991) African 
populations and the evolution of mitochondrial DNA.  Science, 253, 1503-1507. 
Wahlund S (1928) Composition of populations from the perspective of the theory of 
heredity. Hereditas, 11, 65-105 (in German). 
Walker D, Avise JC (1998) Principles of phylogeography as illustrated by freshwater and 
terrestrial turtles in the Southeastern United States. Annual Review of Ecology and 
Systematics, 29, 23-58. 
Wanless HR (1989) The inundation of our coastlines Past, present and future with a focus 
on south Florida, Sea Frontiers, September-October, 264-271. 
Wanless HR, Parkinson RW, Tedesco LP (1994) Sea level control on stability of 
Everglades wetlands. In Everglades: The Ecosystem and Its Restoration, (eds. Davis S, 
Ogden J) pp. 199-223. St. Lucie Press, Delray Beach, Florida. 
Waser J (1993) Sex, mating Systems, Inbreeding, and Outbreeding. In: The Natural 
History of Inbreeding and Outbreeding (ed. Thornhill NW) pp. 1-13. University of 
Chicago Press, Chicago. 
Waser PM, Elliot LF (1991) Dispersal and genetic structure in Kangaroo rats.  Evolution, 
45, 935-943. 
Waser PM, Strobeck C (1998) Genetic signatures of interpopulation dispersal.  Trends in 
Ecology and Evolution, 13, 43-44. 
Watterson GA (1975) On the number of segregating sites in genetic models without 
recombination. Theoretical Population Biology, 7, 256-276. 
Webb SD (1974) Chronology of Florida Pleistocene mammals, In: Pleistocene Mammals 
of Florida (ed. Webb SD), pp. 5-31. University Presses of Florida, Gainesville. 
  292
Webb SD, Wilkins KT (1984) Historical biogeography of Florida Pleistocene Mammals.  
In: Special Publication of Carnegie Museum of Natural History (Genoways HH, 
Dawson), pp. 370-383. Pittsburgh, PA. 
Williams CL, Serfass TL, Cogan R, et al. (2002) Microsatellite variation in the 
reintroduced Pennsylvania elk herd.  Molecular Ecology, 11, 1299-1310. 
Winker CD, Howard JD (1977) Plio-Pleistocene paleogeography of the Florida Gulf 
Coast interpreted from relict shorelines. Transactions-Gulf Coast Association of 
Geological Societies, XXVII, 409-420. 
Wooten MC, Holler NR (1999) Genetic analysis within and among natural populations of 
beach mice. Final Report, United States Fish and Wildlife Service. 
Wooten MC, Scribner KT, Krehling JT (1999) Isolation and characterization of 
microsatellite loci from the endangered beach mouse Peromyscus polionotus.  
Molecular Ecology, 8, 167-168. 
Wright S (1931) Evolution in Mendelian populations. Genetics, 16, 97-159.  
Wright S (1940) The statistical consequences of Mendelian heredity in relation to 
speciation. In: The new systematics (Huxley J), pp. 161-183. Clarendon Press, Oxford. 
Wright S (1943) Isolation by distance. Genetics, 28, 114-138. 
Wright S (1951) The genetical structure of populations. Annals of Eugenics, 15, 323-354. 
Wright S (1969) Evolution and the genetics of populations, In: The theory of gene 
frequencies, Vol 2. University of Chicago Press, Chicago. 
Wright S (1978) Evolution and the genetics of populations, In: Variability within and 
among natural populations, Vol 4. University of Chicago Press, Chicago. 
  293
Yang Z (1996) Among-site rate variation and its impact on phylogenetic analyses. Trends 
in Ecology and Evolution, 11, 367-372. 
Zamudio KR, Savage WK (2003) Historical isolation, range expansion, and secondary 
contact of two highly divergent mitochondrial lineages in spotted salamanders 
(Ambystoma maculatum). Evolution, 57, 1631-1652. 
Zhang XS, Wang J, Hill WG (2004) Redistribution of gene frequency and changes of 
genetic variation following a bottleneck in population size.  Genetics, 167, 1475-1492. 
 294
APPENDICES 
Appe
ndix 1
  
Allele freque
n
cies for each loc
u
s 
by tra
ppi
ng 
peri
od from
 G
azebo, 
Vet Vill
age a
n
d co
m
b
ined grids
 from
 a population 
of Alabam
a
 beac
h  
mice (
Peromyscu
s
 po
lio
no
tu
s a
mmo
ba
tes
).
 
 Lo
cu
s 
A
llele 
Feb-
95
 
A
p
r
-
95
 
Ju
n-9
5
 
Feb-
96
 
A
p
r
-
96
 
Ju
n-9
6
 
Feb-
97
 
A
p
r
-
98
 
Feb-
99
 
Mar
-
0
0
 
N
o
v
-
00
 
Mar
-
0
1
 
N
o
v
-
01
 
Mar
-
0
2
 
Ppa
-0
1 
15
2 
0.
00
4
0 
0.
01
8
0 
0.
03
3
0 
0.
04
3
0 
0.
06
7
0 
0.
04
4
0 
- 
- 
- 
- 
- 
- 
- 
- 
 
18
4 
0.
26
9
0 
0.
26
8
0 
0.
25
8
0 
0.
25
5
0 
0.
27
9
0 
0.
32
4
0 
0.
38
9
0 
0.
62
5
0 
0.
56
3
0 
0.
37
5
0 
0.
37
5
0 
0.
37
5
0 
0.
41
7
0 
0.
35
7
0 
 
18
6 
0.
72
7
0 
0.
71
4
0 
0.
70
8
0 
0.
70
2
0 
0.
65
4
0 
0.
63
2
0 
0.
61
1
0 
0.
37
5
0 
0.
43
8
0 
0.
62
5
0 
0.
62
5
0 
0.
62
5
0 
0.
58
3
0 
0.
64
3
0 
Ppa
-1
2
 
22
9 
- 
- 
- 
- 
- 
0.
01
5
0
 
- 
- 
- 
- 
- 
- 
- 
- 
 
23
4 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
00
5
0
 
 
23
7 
- 
- 
- 
- 
- 
0.
01
5
0
 
- 
- 
- 
- 
- 
- 
- 
- 
 
24
2 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
00
5
0
 
 
24
4 
0.
34
9
0 
0.
35
7
0 
0.
37
1
0 
0.
39
4
0 
0.
41
3
0 
0.
29
4
0 
0.
31
3
0 
0.
42
9
0 
0.
12
5
0 
0.
30
6
0 
0.
45
5
0 
0.
47
7
0 
0.
20
8
0 
0.
24
5
0 
 
24
8 
0.
52
5
0 
0.
50
4
0 
0.
48
5
0 
0.
42
6
0 
0.
40
4
0 
0.
47
1
0 
0.
37
5
0 
0.
28
6
0 
0.
87
5
0 
0.
62
9
0 
0.
45
5
0 
0.
45
5
0 
0.
75
0
0 
0.
64
1
0 
 
25
0 
0.
01
1
0 
0.
00
8
0 
- 
- 
- 
- 
0.
06
3
0 
0.
07
1
0 
- 
- 
- 
- 
- 
- 
 
25
2 
0.
11
6
0 
0.
13
0
0 
0.
14
4
0 
0.
18
1
0 
0.
18
3
0 
0.
20
6
0 
0.
25
0
0 
0.
21
4
0 
- 
0.
06
5
0 
0.
09
1
0 
0.
06
8
0 
0.
04
2
0 
0.
10
5
0 
Ppa
-4
6
 
22
2 
- 
- 
- 
- 
- 
- 
- 
- 
0.
25
0
0
 
- 
- 
- 
- 
- 
 
22
6 
- 
- 
- 
- 
- 
- 
- 
0.
11
1
0 
0.
25
0
0 
- 
- 
- 
- 
- 
 
22
9 
0.
28
9
0 
0.
26
3
0 
0.
34
0
0 
0.
52
1
0 
0.
43
3
0 
0.
36
8
0 
0.
61
1
0 
0.
22
2
0 
0.
06
3
0 
0.
40
6
0 
0.
22
7
0 
0.
29
5
0 
0.
29
2
0 
0.
30
4
0 
 
23
1 
0.
59
8
0 
0.
67
9
0 
0.
60
4
0 
0.
36
5
0 
0.
48
1
0 
0.
52
9
0 
0.
38
9
0 
0.
61
1
0 
0.
37
5
0 
0.
53
1
0 
0.
77
3
0 
0.
63
6
0 
0.
62
5
0 
0.
63
7
0 
 
23
5 
0.
11
3
0 
0.
05
8
0 
0.
05
7
0 
0.
11
5
0 
0.
08
7
0 
0.
08
8
0 
- 
0.
05
6
0 
0.
06
3
0 
0.
06
3
0 
- 
0.
06
8
0 
0.
08
3
0 
0.
05
8
0 
 
23
9 
- 
- 
- 
- 
- 
0.
01
5
0
 
- 
- 
- 
- 
- 
- 
- 
- 
PO
3-
85 
22
7 
0.
05
2
0 
0.
04
5
0 
0.
05
2
0 
0.
02
1
0 
0.
03
8
0 
0.
04
4
0 
- 
- 
- 
0.
03
2
0 
- 
0.
04
5
0 
- 
0.
02
9
0 
 
22
9 
0.
00
7
0 
0.
00
8
0 
0.
00
7
0 
- 
0.
01
0
0 
0.
02
9
0 
- 
- 
- 
- 
- 
- 
- 
- 
 
23
1 
0.
57
3
0 
0.
58
3
0 
0.
53
7
0 
0.
52
1
0 
0.
63
5
0 
0.
57
4
0 
0.
55
6
0 
0.
68
8
0 
0.
75
0
0 
0.
48
4
0 
0.
30
0
0 
0.
36
4
0 
0.
72
7
0 
0.
54
6
0 
 
23
3 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
01
6
0
 
- 
- 
- 
- 
 
23
5 
0.
31
8
0 
0.
32
2
0 
0.
36
6
0 
0.
30
2
0 
0.
24
0
0 
0.
27
9
0 
0.
16
7
0 
0.
25
0
0 
0.
25
0
0 
0.
35
5
0 
0.
65
0
0 
0.
40
9
0 
0.
18
2
0 
0.
25
4
0 
 
23
7 
0.
04
9
0 
0.
04
1
0 
0.
03
7
0 
0.
15
6
0 
0.
07
7
0 
0.
07
4
0 
0.
27
8
0 
0.
06
3
0 
- 
0.
11
3
0 
0.
05
0
0 
0.
18
2
0 
0.
09
1
0 
0.
17
1
0 
 
295
Lo
cu
s 
A
llele 
Feb-
95
 
A
p
r
-
95
 
Ju
n-9
5
 
Feb-
96
 
A
p
r
-
96
 
Ju
n-9
6
 
Feb-
97
 
A
p
r
-
98
 
Feb-
99
 
Mar
-
0
0
 
N
o
v
-
00
 
Mar
-
0
1
 
N
o
v
-
01
 
Mar
-
0
2
 
PO
3-
68 
26
3 
- 
- 
- 
- 
- 
0.
01
5
0
 
- 
- 
- 
- 
- 
- 
- 
- 
 
27
0 
0.
35
8
0 
0.
34
0
0 
0.
32
3
0 
0.
31
5
0 
0.
31
7
0 
0.
27
9
0 
0.
12
5
0 
0.
31
3
0 
0.
75
0
0 
0.
33
9
0 
0.
36
4
0 
0.
34
1
0 
0.
37
5
0 
0.
27
4
0 
 
27
2 
0.
00
4
0 
0.
00
4
0 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
 
27
6 
0.
12
1
0 
0.
09
7
0 
0.
08
5
0 
0.
10
9
0 
0.
11
5
0 
0.
11
8
0 
0.
18
8
0 
0.
31
3
0 
- 
0.
24
2
0 
0.
27
3
0 
0.
18
2
0 
0.
08
3
0 
0.
22
6
0 
 
27
8 
0.
12
8
0 
0.
14
3
0 
0.
16
2
0 
0.
19
6
0 
0.
20
2
0 
0.
19
1
0 
0.
37
5
0 
0.
25
0
0 
- 
0.
29
0
0 
0.
09
1
0 
0.
15
9
0 
0.
20
8
0 
0.
19
7
0 
 
28
0 
0.
05
0
0 
0.
04
6
0 
0.
05
4
0 
0.
09
8
0 
0.
08
7
0 
0.
10
3
0 
0.
12
5
0 
- 
- 
- 
- 
0.
02
3
0 
0.
04
2
0 
0.
04
7
0 
 
28
2 
0.
07
1
0 
0.
09
2
0 
0.
09
2
0 
0.
04
3
0 
0.
08
7
0 
0.
08
8
0 
0.
06
3
0 
0.
06
3
0 
- 
0.
08
1
0 
0.
22
7
0 
0.
22
7
0 
0.
25
0
0 
0.
22
6
0 
 
28
4 
0.
27
0
0 
0.
27
7
0 
0.
28
5
0 
0.
23
9
0 
0.
19
2
0 
0.
20
6
0 
0.
12
5
0 
0.
06
3
0 
0.
25
0
0 
0.
04
8
0 
0.
04
5
0 
0.
06
8
0 
0.
04
2
0 
0.
03
0
0 
PO-
2
5 
14
3 
- 
- 
- 
- 
- 
0.
01
5
0
 
- 
- 
- 
- 
- 
- 
- 
- 
 
14
5 
- 
- 
- 
- 
- 
0.
02
9
0
 
- 
- 
- 
- 
- 
- 
- 
- 
 
14
7 
0.
32
4
0 
0.
33
1
0 
0.
31
1
0 
0.
31
9
0 
0.
34
3
0 
0.
36
8
0 
0.
38
9
0 
0.
37
5
0 
0.
37
5
0 
0.
66
1
0 
0.
25
0
0 
0.
28
6
0 
0.
12
5
0 
0.
36
9
0 
 
14
9 
0.
11
4
0 
0.
12
0
0 
0.
09
1
0 
0.
06
4
0 
0.
14
7
0 
0.
07
4
0 
0.
22
2
0 
0.
18
8
0 
0.
12
5
0 
0.
06
5
0 
0.
30
0
0 
0.
14
3
0 
0.
20
8
0 
0.
15
3
0 
 
15
1 
0.
05
2
0 
0.
03
7
0 
0.
02
3
0 
0.
09
6
0 
0.
04
9
0 
0.
05
9
0 
0.
05
6
0 
0.
06
3
0 
0.
12
5
0 
0.
12
9
0 
0.
15
0
0 
0.
02
4
0 
0.
04
2
0 
0.
13
1
0 
 
15
3 
- 
- 
0.
00
8
0 
0.
01
1
0 
0.
01
0
0 
0.
01
5
0 
- 
- 
- 
- 
- 
- 
- 
- 
 
16
3 
0.
01
0
0 
0.
01
2
0 
- 
- 
- 
0.
01
5
0 
- 
- 
- 
- 
- 
- 
0.
04
2
0 
0.
00
4
0 
 
16
9 
0.
50
0
0 
0.
50
0
0 
0.
56
8
0 
0.
51
1
0 
0.
45
1
0 
0.
42
6
0 
0.
33
3
0 
0.
37
5
0 
0.
37
5
0 
0.
14
5
0 
0.
30
0
0 
0.
54
8
0 
0.
58
3
0 
0.
34
3
0 
Pm
l
-
03 
23
7 
0.
59
2
0 
0.
58
2
0 
0.
58
5
0 
0.
55
0
0 
0.
74
1
0 
0.
66
0
0 
0.
50
0
0 
0.
44
4
0 
0.
40
0
0 
0.
85
9
0 
0.
63
6
0 
0.
72
7
0 
0.
54
2
0 
0.
61
4
0 
 
24
9 
- 
- 
- 
- 
- 
0.
02
0
0
 
- 
- 
- 
- 
- 
- 
- 
- 
 
25
1 
0.
13
0
0 
0.
12
6
0 
0.
10
6
0 
0.
20
0
0 
0.
10
3
0 
0.
06
0
0 
0.
18
8
0 
0.
11
1
0 
- 
0.
03
1
0 
0.
18
2
0 
0.
04
5
0 
- 
0.
05
7
0 
 
25
3 
0.
00
5
0 
0.
00
5
0 
0.
01
1
0 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
 
25
5 
0.
16
3
0 
0.
13
7
0 
0.
17
0
0 
0.
15
0
0 
0.
01
7
0 
0.
06
0
0 
0.
12
5
0 
0.
22
2
0 
0.
40
0
0 
0.
04
7
0 
0.
09
1
0 
0.
02
3
0 
0.
04
2
0 
0.
05
7
0 
 
25
7 
- 
- 
- 
- 
0.
01
7
0 
0.
04
0
0 
- 
- 
- 
- 
- 
- 
- 
- 
 
25
9 
0.
10
9
0 
0.
14
8
0 
0.
12
8
0 
0.
10
0
0 
0.
12
1
0 
0.
16
0
0 
0.
18
8
0 
0.
22
2
0 
0.
20
0
0 
0.
06
3
0 
0.
09
1
0 
0.
20
5
0 
0.
41
7
0 
0.
27
3
0 
 
296
Lo
cu
s 
A
llele 
Feb-
95
 
A
p
r
-
95
 
Ju
n-9
5
 
Feb-
96
 
A
p
r
-
96
 
Ju
n-9
6
 
Feb-
97
 
A
p
r
-
98
 
Feb-
99
 
Mar
-
0
0
 
N
o
v
-
00
 
Mar
-
0
1
 
N
o
v
-
01
 
Mar
-
0
2
 
Pm
l
-
11 
25
6 
0.
21
8
0 
0.
19
9
0 
0.
24
2
0 
0.
34
4
0 
0.
28
4
0 
0.
31
8
0 
0.
50
0
0 
0.
50
0
0 
0.
62
5
0 
0.
34
4
0 
0.
36
4
0 
0.
25
0
0 
0.
45
8
0 
0.
32
9
0 
 
25
8 
- 
- 
- 
- 
- 
0.
01
5
0
 
- 
- 
- 
- 
- 
- 
- 
- 
 
26
0 
0.
06
7
0 
0.
05
9
0 
0.
06
3
0 
0.
03
1
0 
0.
02
9
0 
0.
03
0
0 
- 
0.
06
3
0 
0.
12
5
0 
0.
03
1
0 
- 
0.
04
5
0 
- 
0.
04
3
0 
 
26
2 
0.
25
4
0 
0.
25
0
0 
0.
22
7
0 
0.
17
7
0 
0.
21
6
0 
0.
18
2
0 
0.
16
7
0 
0.
18
8
0 
0.
12
5
0 
0.
23
4
0 
0.
31
8
0 
0.
36
4
0 
0.
16
7
0 
0.
23
9
0 
 
26
4 
0.
35
6
0 
0.
36
4
0 
0.
33
6
0 
0.
33
3
0 
0.
34
3
0 
0.
34
8
0 
0.
22
2
0 
0.
25
0
0 
0.
12
5
0 
0.
34
4
0 
0.
18
2
0 
0.
20
5
0 
0.
20
8
0 
0.
19
7
0 
 
26
6 
0.
05
6
0 
0.
06
8
0 
0.
08
6
0 
0.
09
4
0 
0.
04
9
0 
0.
04
5
0 
0.
11
1
0 
- 
- 
- 
- 
- 
- 
0.
09
8
0 
 
26
8 
0.
04
9
0 
0.
05
9
0 
0.
04
7
0 
0.
02
1
0 
0.
07
8
0 
0.
06
1
0 
- 
- 
- 
0.
04
7
0 
0.
13
6
0 
0.
13
6
0 
0.
16
7
0 
0.
09
4
0 
Pm
l
-
04 
22
2 
0.
34
3
0 
0.
33
5
0 
0.
32
0
0 
0.
35
4
0 
0.
33
7
0 
0.
38
2
0 
0.
27
8
0 
0.
22
2
0 
0.
50
0
0 
0.
43
8
0 
0.
36
4
0 
0.
34
1
0 
0.
33
3
0 
0.
31
5
0 
 
22
6 
0.
63
1
0 
0.
64
6
0 
0.
68
0
0 
0.
58
3
0 
0.
63
3
0 
0.
54
4
0 
0.
61
1
0 
0.
72
2
0 
0.
50
0
0 
0.
56
3
0 
0.
06
3
6 
0.
65
9
0 
0.
62
5
0 
0.
66
8
0 
 
22
8 
0.
02
5
0 
0.
01
9
0 
- 
0.
06
3
0 
0.
03
1
0 
0.
04
4
0 
0.
11
1
0 
0.
05
6
0 
- 
- 
- 
- 
0.
04
2
0 
0.
01
7
0 
 
23
0 
- 
- 
- 
- 
- 
0.
02
9
0
 
- 
- 
- 
- 
- 
- 
- 
- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
PO-
7
1 
23
7 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
00
4
0
 
 
24
1 
0.
01
8
0 
0.
02
5
0 
0.
03
0
0 
- 
- 
- 
- 
0.
06
3
0 
- 
0.
09
3
0 
0.
09
1
0 
0.
02
4
0 
- 
0.
00
9
0 
 
24
3 
0.
43
3
0 
0.
40
6
0 
0.
46
2
0 
0.
52
1
0 
0.
51
0
0 
0.
51
7
0 
0.
83
3
0 
0.
56
3
0 
0.
75
0
0 
0.
66
7
0 
0.
40
9
0 
0.
54
8
0 
0.
54
5
0 
0.
59
3
0 
 
24
5 
0.
29
9
0 
0.
29
1
0 
0.
29
5
0 
0.
24
5
0 
0.
29
4
0 
0.
25
9
0 
0.
08
3
0 
0.
18
8
0 
0.
12
5
0 
0.
05
6
0 
0.
18
2
0 
0.
21
4
0 
0.
40
9
0 
0.
27
0
0 
 
24
7 
0.
01
1
0 
0.
01
2
0 
0.
00
8
0 
- 
- 
- 
- 
0.
06
3
0 
0.
12
5
0 
0.
05
6
0 
0.
04
5
0 
0.
04
8
0 
- 
- 
 
24
9 
0.
05
6
0 
0.
07
8
0 
0.
04
5
0 
0.
08
5
0 
0.
07
8
0 
0.
06
9
0 
- 
- 
- 
0.
05
6
0 
0.
13
6
0 
0.
11
9
0 
- 
0.
05
8
0 
 
25
1 
0.
15
5
0 
0.
16
0
0 
0.
12
9
0 
0.
11
7
0 
0.
08
8
0 
0.
13
8
0 
0.
08
3
0 
0.
12
5
0 
- 
0.
07
4
0 
0.
13
6
0 
0.
02
4
0 
- 
0.
02
7
0 
 
25
3 
0.
02
5
0 
0.
02
9
0 
0.
03
0
0 
0.
03
2
0 
0.
02
9
0 
0.
01
7
0 
- 
- 
- 
- 
- 
0.
02
4
0 
0.
04
5
0 
0.
03
1
0 
 
25
8 
0.
00
4
0
 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
 
26
2 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
00
9
0
 
Pm
l
-
06 
13
3 
0.
47
9
0 
0.
46
3
0 
0.
41
8
0 
0.
42
7
0 
0.
53
1
0 
0.
60
9
0 
0.
05
6
0 
0.
25
0
0 
0.
50
0
0 
0.
21
9
0 
0.
33
3
0 
0.
43
2
0 
0.
45
8
0 
0.
36
3
0 
 
14
8 
0.
17
5
0 
0.
15
3
0 
0.
16
4
0 
0.
17
7
0 
0.
11
2
0 
0.
14
1
0 
0.
33
3
0 
0.
18
8
0 
0.
10
0
0 
0.
25
0
0 
0.
11
1
0 
0.
18
2
0 
0.
16
7
0 
0.
23
1
0 
 
15
2 
0.
34
6
0 
0.
38
4
0 
0.
41
8
0 
0.
39
6
0 
0.
35
7
0 
0.
25
0
0 
0.
61
1
0 
0.
56
3
0 
0.
40
0
0 
0.
53
1
0 
0.
55
6
0 
0.
38
6
0 
0.
37
5
0 
0.
40
6
0 
 
297
Vet Villag
e
 
 Lo
cu
s 
A
llele 
Feb-
95
 
A
p
r
-
95
 
Ju
n-9
5
 
Feb-
96
 
A
p
r
-
96
 
Ju
n-9
6
 
Feb-
97
 
A
p
r
-
98
 
Mar
-
0
0
 
N
o
v
-
00
 
Mar
-
0
1
 
N
o
v
-
01
 
Mar
-
0
2
 
Ppa
-0
1 
15
2 
- 
- 
- 
- 
0.
01
4
0 
- 
- 
0.
02
3
0 
- 
- 
- 
- 
- 
 
18
4 
0.
34
5
0 
0.
37
4
0 
0.
41
9
0 
0.
36
4
0 
0.
27
1
0 
0.
29
1
0 
0.
41
1
0 
0.
47
7
0 
0.
44
2
0 
0.
25
0
0 
0.
20
9
0 
0.
20
0
0 
0.
20
9
0 
 
18
6 
0.
65
5
0 
0.
62
6
0 
0.
58
1
0 
0.
63
6
0 
0.
71
5
0 
0.
70
9
0 
0.
58
9
0 
0.
50
0
0 
0.
55
8
0 
0.
75
0
0 
0.
79
1
0 
0.
80
0
0 
0.
79
1
0 
Ppa
-1
2
 
23
4 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
00
9
0
 
 
24
4 
0.
45
4
0 
0.
44
4
0 
0.
45
5
0 
0.
34
9
0 
0.
35
9
0 
0.
38
1
0 
0.
53
6
0 
0.
36
4
0 
0.
42
0
0 
0.
31
3
0 
0.
38
5
0 
0.
27
9
0 
0.
21
2
0 
 
24
8 
0.
32
1
0 
0.
33
3
0 
0.
33
6
0 
0.
32
6
0 
0.
33
1
0 
0.
35
7
0 
0.
37
5
0 
0.
56
8
0 
0.
44
0
0 
0.
43
8
0 
0.
39
7
0 
0.
50
0
0 
0.
42
9
0 
 
25
0 
0.
02
0
0 
0.
00
9
0 
0.
00
7
0 
0.
02
3
0 
- 
0.
02
4
0 
- 
- 
- 
- 
- 
- 
- 
 
25
2 
0.
16
8
0 
0.
15
7
0 
0.
17
9
0 
0.
30
2
0 
0.
30
3
0 
0.
23
8
0 
0.
08
9
0 
0.
06
8
0 
0.
14
0
0 
0.
25
0
0 
0.
21
8
0 
0.
22
1
0 
0.
35
0
0 
 
25
4 
0.
03
6
0 
0.
05
6
0 
0.
02
2
0 
- 
0.
00
7
0 
- 
- 
- 
- 
- 
- 
- 
- 
Ppa
-
4
6 
22
9 
0.
38
3
0 
0.
41
0
0 
0.
41
9
0 
0.
47
7
0 
0.
45
1
0 
0.
36
0
0 
0.
23
2
0 
0.
31
8
0 
0.
34
6
0 
0.
25
0
0 
0.
34
9
0 
0.
31
4
0 
0.
27
4
0 
 
23
1 
0.
58
5
0 
0.
55
9
0 
0.
55
1
0 
0.
45
5
0 
0.
46
5
0 
0.
53
5
0 
0.
73
2
0 
0.
59
1
0 
0.
63
5
0 
0.
62
5
0 
0.
58
1
0 
0.
64
3
0 
0.
69
7
0 
 
23
5 
0.
03
2
0 
0.
03
2
0 
0.
02
9
0 
0.
06
8
0 
0.
08
3
0 
0.
10
5
0 
0.
03
6
0 
0.
09
1
0 
0.
01
9
0 
0.
12
5
0 
0.
07
0
0 
0.
04
3
0 
0.
03
0
0 
PO
3-
85 
22
7 
0.
13
7
0 
0.
11
9
0 
0.
11
4
0 
0.
13
6
0 
0.
10
4
0 
0.
13
1
0 
0.
23
2
0 
0.
18
2
0 
0.
11
5
0 
0.
18
8
0 
0.
03
6
0 
0.
05
6
0 
0.
03
0
0 
 
22
9 
0.
02
5
0 
0.
03
5
0 
0.
04
3
0 
0.
01
1
0 
0.
02
8
0 
0.
03
6
0 
- 
- 
- 
- 
- 
- 
- 
 
23
1 
0.
40
7
0 
0.
42
5
0 
0.
43
6
0 
0.
42
0
0 
0.
47
2
0 
0.
42
9
0 
0.
42
9
0 
0.
31
8
0 
0.
23
1
0 
0.
50
0
0 
0.
57
1
0 
0.
43
1
0 
0.
45
7
0 
 
23
3 
- 
0.
00
4
0 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
 
23
5 
0.
30
4
0 
0.
31
4
0 
0.
32
1
0 
0.
27
3
0 
0.
27
1
0 
0.
32
1
0 
0.
25
0
0 
0.
27
3
0 
0.
55
8
0 
0.
31
3
0 
0.
28
6
0 
0.
27
8
0 
0.
28
9
0 
 
23
7 
0.
12
7
0 
0.
10
2
0 
0.
08
6
0 
0.
15
9
0 
0.
12
5
0 
0.
08
3
0 
0.
08
9
0 
0.
22
7
0 
0.
09
6
0 
- 
0.
10
7
0 
0.
23
6
0 
0.
22
4
0 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
298
Lo
cu
s 
A
llele 
Feb-
95
 
A
p
r
-
95
 
Ju
n-9
5
 
Feb-
96
 
A
p
r
-
96
 
Ju
n-9
6
 
Feb-
97
 
A
p
r
-
98
 
Mar
-
0
0
 
N
o
v
-
00
 
Mar
-
0
1
 
N
o
v
-
01
 
Mar
-
0
2
 
PO
3-
68 
27
0 
0.
18
3
0 
0.
18
3
0 
0.
19
3
0 
0.
12
5
0 
0.
21
5
0 
0.
16
3
0 
0.
22
2
0 
0.
25
0
0 
0.
61
5
0 
0.
62
5
0 
0.
52
3
0 
0.
44
3
0 
0.
44
9
0 
 
27
2 
- 
0.
00
4
0 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
 
27
6 
0.
12
9
0 
0.
14
3
0 
0.
15
0
0 
0.
22
7
0 
0.
20
8
0 
0.
27
9
0 
0.
29
6
0 
0.
38
6
0 
0.
09
6
0 
0.
12
5
0 
0.
26
7
0 
0.
25
7
0 
0.
29
9
0 
 
27
8 
0.
21
3
0 
0.
21
9
0 
0.
15
7
0 
0.
25
0
0 
0.
16
7
0 
0.
17
4
0 
0.
24
1
0 
0.
20
5
0 
0.
09
6
0 
0.
06
3
0 
0.
04
7
0 
0.
17
1
0 
0.
10
3
0 
 
28
0 
0.
06
4
0 
0.
04
9
0 
0.
07
1
0 
0.
03
4
0 
0.
06
9
0 
0.
07
0
0 
0.
03
7
0 
- 
- 
0.
06
3
0 
0.
02
3
0 
0.
04
3
0 
0.
01
7
0 
 
28
2 
0.
06
4
0 
0.
09
8
0 
0.
07
9
0 
0.
08
0
0 
0.
09
7
0 
0.
09
3
0 
0.
03
7
0 
0.
11
4
0 
0.
19
2
0 
0.
06
3
0 
0.
11
6
0 
0.
08
6
0 
0.
13
2
0 
 
28
4 
0.
34
7
0 
0.
30
4
0 
0.
35
0
0 
0.
28
4
0 
0.
24
3
0 
0.
22
1
0 
0.
16
7
0 
0.
04
5
0 
- 
0.
06
3
0 
0.
02
3
0 
- 
- 
PO-
2
5 
14
5 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
03
1
0 
0.
00
5
0 
 
14
7 
0.
36
0
0 
0.
35
4
0 
0.
40
7
0 
0.
46
5
0 
0.
42
0
0 
0.
42
7
0 
0.
46
4
0 
0.
43
2
0 
0.
43
8
0 
0.
42
9
0 
0.
45
3
0 
0.
31
3
0 
0.
44
2
0 
 
14
9 
0.
11
0
0 
0.
09
3
0 
0.
07
1
0 
0.
02
3
0 
0.
05
8
0 
0.
11
0
0 
0.
05
4
0 
0.
11
4
0 
0.
25
0
0 
0.
07
1
0 
0.
12
5
0 
0.
20
3
0 
0.
11
7
0 
 
15
1 
0.
03
5
0 
0.
05
3
0 
0.
05
0
0 
0.
03
5
0 
0.
02
9
0 
0.
06
1
0 
0.
07
1
0 
- 
0.
08
3
0 
0.
35
7
0 
0.
07
8
0 
0.
14
1
0 
0.
20
4
0 
 
15
3 
- 
0.
00
4
0 
0.
00
7
0 
0.
02
3
0 
0.
02
2
0 
0.
03
7
0 
- 
- 
- 
- 
- 
- 
- 
 
16
9 
0.
49
5
0 
0.
49
6
0 
0.
46
4
0 
0.
45
3
0 
0.
47
1
0 
0.
36
6
0 
0.
41
1
0 
0.
45
5
0 
0.
22
9
0 
0.
14
3
0 
0.
34
4
0 
0.
31
3
0 
0.
23
3
0 
Pm
l-03 
23
5 
- 
- 
- 
- 
- 
0.
01
2
0
 
- 
- 
- 
- 
- 
- 
- 
 
23
7 
0.
55
8
0 
0.
58
0
0 
0.
54
3
0 
0.
47
7
0 
0.
43
7
0 
0.
52
3
0 
0.
44
6
0 
0.
34
1
0 
0.
53
8
0 
1.
00
0
0 
0.
67
4
0 
0.
73
5
0 
0.
70
3
0 
 
25
1 
0.
24
8
0 
0.
20
8
0 
0.
19
3
0 
0.
29
1
0 
0.
28
9
0 
0.
18
6
0 
0.
17
9
0 
0.
18
2
0 
0.
03
8
0 
- 
0.
07
0
0 
0.
04
4
0 
0.
05
6
0 
 
25
5 
0.
10
2
0 
0.
10
6
0 
0.
15
7
0 
0.
16
3
0 
0.
18
3
0 
0.
20
9
0 
0.
28
6
0 
0.
15
9
0 
0.
11
5
0 
- 
0.
02
3
0 
0.
01
5
0 
- 
 
25
7 
- 
- 
- 
- 
0.
00
7
0 
0.
01
2
0 
- 
- 
- 
- 
- 
- 
- 
 
25
9 
0.
09
2
0 
0.
10
6
0 
0.
10
7
0 
0.
07
0
0 
0.
08
5
0 
0.
05
8
0 
0.
08
9
0 
0.
31
8
0 
0.
30
8
0 
- 
0.
23
3
0 
0.
20
6
0 
0.
24
1
0 
Pm
l
-
11 
25
6 
0.
34
5
0 
0.
40
7
0 
0.
38
6
0 
0.
32
6
0 
0.
39
3
0 
0.
35
7
0 
0.
44
6
0 
0.
36
4
0 
0.
17
3
0 
0.
18
8
0 
0.
20
0
0 
0.
30
3
0 
0.
21
9
0 
 
25
8 
- 
0.
00
4
0 
0.
00
7
0 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
 
26
0 
0.
02
9
0 
0.
04
4
0 
0.
03
6
0 
0.
05
8
0 
0.
07
1
0 
0.
06
0
0 
0.
01
8
0 
- 
0.
03
8
0 
- 
- 
- 
0.
00
9
0 
 
26
2 
0.
24
3
0 
0.
19
9
0 
0.
22
1
0 
0.
16
3
0 
0.
15
0
0 
0.
21
4
0 
0.
17
9
0 
0.
20
5
0 
0.
46
2
0 
0.
37
5
0 
0.
30
0
0 
0.
34
8
0 
0.
40
6
0 
 
26
4 
0.
20
4
0 
0.
20
4
0 
0.
22
1
0 
0.
19
8
0 
0.
20
7
0 
0.
27
4
0 
0.
26
8
0 
0.
34
1
0 
0.
26
9
0 
0.
37
5
0 
0.
40
0
0 
0.
22
7
0 
0.
25
4
0 
 
26
6 
0.
13
1
0 
0.
10
6
0 
0.
09
3
0 
0.
23
3
0 
0.
11
4
0 
0.
04
8
0 
0.
08
9
0 
0.
09
1
0 
0.
03
8
0 
- 
0.
03
8
0 
0.
07
6
0 
0.
05
8
0 
 
26
8 
0.
04
9
0 
0.
03
5
0 
0.
03
6
0 
0.
02
3
0 
0.
06
4
0 
0.
04
8
0 
- 
- 
0.
01
9
0 
0.
06
3
0 
0.
06
3
0 
0.
04
5
0 
0.
05
4
0 
 
299
Lo
cu
s 
A
llele 
Feb-
95
 
A
p
r
-
95
 
Ju
n-9
5
 
Feb-
96
 
A
p
r
-
96
 
Ju
n-9
6
 
Feb-
97
 
A
p
r
-
98
 
Mar
-
0
0
 
N
o
v
-
00
 
Mar
-
0
1
 
N
o
v
-
01
 
Mar
-
0
2
 
Pm
l
-
04 
22
2 
0.
16
5
0 
0.
20
4
0 
0.
16
4
0 
0.
23
9
0 
0.
26
4
0 
0.
26
1
0 
0.
19
6
0 
0.
27
3
0 
0.
24
0
0 
0.
56
3
0 
0.
45
3
0 
0.
31
9
0 
0.
43
5
0 
 
22
4 
- 
- 
- 
- 
0.
00
7
0 
0.
01
1
0 
0.
01
8
0 
- 
0.
02
0
0 
- 
0.
01
2
0 
- 
0.
00
4
0 
 
22
6 
0.
76
2
0 
0.
72
6
0 
0.
76
4
0 
0.
69
3
0 
0.
66
7
0 
0.
71
6
0 
0.
71
4
0 
0.
70
5
0 
0.
74
0
0 
0.
43
8
0 
0.
53
5
0 
0.
68
1
0 
0.
56
1
0 
 
22
8 
0.
07
3
0 
0.
07
1
0 
0.
07
1
0 
0.
06
8
0 
0.
06
3
0 
0.
01
1
0 
0.
07
1
0 
0.
02
3
0 
- 
- 
- 
- 
- 
PO-
7
1 
24
1 
0.
01
7
0 
0.
00
5
0 
0.
00
8
0 
- 
0.
00
7
0 
0.
01
2
0 
0.
03
6
0 
- 
0.
08
3
0 
0.
06
3
0 
0.
04
7
0 
0.
02
8
0 
0.
04
3
0 
 
24
3 
0.
37
2
0 
0.
37
2
0 
0.
35
0
0 
0.
32
9
0 
0.
39
9
0 
0.
31
0
0 
0.
46
4
0 
0.
55
0
0 
0.
37
5
0 
0.
37
5
0 
0.
31
4
0 
0.
44
4
0 
0.
38
3
0 
 
24
5 
0.
37
8
0 
0.
41
3
0 
0.
42
5
0 
0.
28
0
0 
0.
28
3
0 
0.
32
1
0 
0.
21
4
0 
0.
05
0
0 
0.
31
3
0 
0.
06
3
0 
0.
30
2
0 
0.
30
6
0 
0.
27
4
0 
 
24
7 
0.
04
1
0 
0.
02
0
0 
0.
03
3
0 
0.
04
9
0 
0.
01
4
0 
0.
02
4
0 
- 
0.
07
5
0 
0.
08
3
0 
0.
06
3
0 
0.
03
5
0 
0.
01
4
0 
0.
01
3
0 
 
24
9 
0.
06
4
0 
0.
06
6
0 
0.
06
7
0 
0.
18
3
0 
0.
15
9
0 
0.
14
3
0 
0.
12
5
0 
0.
10
0
0 
0.
10
4
0 
0.
31
3
0 
0.
27
9
0 
0.
20
8
0 
0.
26
1
0 
 
25
1 
0.
10
5
0 
0.
09
2
0 
0.
08
3
0 
0.
08
5
0 
0.
08
7
0 
0.
11
9
0 
0.
10
7
0 
0.
22
5
0 
0.
04
2
0 
0.
12
5
0 
0.
02
3
0 
- 
- 
 
25
3 
0.
01
7
0 
0.
02
6
0 
0.
02
5
0 
0.
04
9
0 
0.
02
9
0 
0.
06
0
0 
0.
05
4
0 
- 
- 
- 
- 
- 
0.
01
7
0 
 
26
2 
0.
00
6
0 
0.
00
5
0 
0.
00
8
0 
0.
02
4
0 
0.
02
2
0 
0.
01
2
0 
- 
- 
- 
- 
- 
- 
0.
00
9
0 
Pm
l
-
06 
13
3 
0.
45
0
0 
0.
46
5
0 
0.
43
6
0 
0.
44
3
0 
0.
43
8
0 
0.
51
2
0 
0.
53
6
0 
0.
54
5
0 
0.
28
0
0 
0.
18
8
0 
0.
17
4
0 
0.
29
2
0 
0.
20
9
0 
 
14
8 
0.
21
3
0 
0.
21
7
0 
0.
20
7
0 
0.
22
7
0 
0.
18
8
0 
0.
17
9
0 
0.
10
7
0 
0.
09
1
0 
0.
08
0
0 
0.
06
3
0 
0.
14
0
0 
0.
05
6
0 
0.
09
4
0 
 
15
2 
0.
33
7
0 
0.
31
9
0 
0.
35
7
0 
0.
33
0
0 
0.
37
5
0 
0.
31
0
0 
0.
35
7
0 
0.
36
4
0 
0.
64
0
0 
0.
75
0
0 
0.
68
6
0 
0.
65
3
0 
0.
69
7
0 
 
300
Com
bined gri
ds 
 Loc
u
s 
Al
l
e
l
e
 
1st
-
95 
2n
d
-9
5
 
3r
d-
9
5
1st
-
96
2n
d
-9
6
3r
d-
9
6
1st
-
97
1st
-
98
1st
-
99 
1st
-
00
2n
d
-0
0
1st
-
01
2n
d
-0
1
1st
-
02
Ppa
-0
1 
15
2 
0.
00
2
2 
0.
00
9
0 
0.
01
5
6
0.
02
2
0
0.
03
6
3 
0.
01
9
5
- 
0.
01
6
7
- 
- 
- 
- 
- 
- 
 
18
4 
0.
30
1
7 
0.
32
0
6 
0.
34
3
8
0.
30
7
7
0.
27
4
2 
0.
30
5
2
0.
40
5
4
0.
51
6
7
0.
56
3
0 
0.
40
5
2
0.
31
2
5 
0.
26
1
9
0.
26
0
4 
0.
28
3
9
 
18
6 
0.
69
6
1 
0.
67
0
4 
0.
64
0
6
0.
67
0
3
0.
68
9
5 
0.
67
5
3
0.
59
4
6
0.
46
6
7
0.
43
8
0 
0.
59
4
8
0.
68
7
5 
0.
73
8
1
0.
73
9
6 
0.
71
6
1
Ppa
-1
2
 
22
9 
- 
- 
- 
- 
- 
0.
00
6
6
- 
- 
- 
- 
- 
- 
- 
- 
 
23
7 
- 
- 
- 
- 
- 
0.
00
6
6
- 
- 
- 
- 
- 
- 
- 
- 
 
23
4 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
00
6
7
 
24
2 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
00
2
2
 
24
4 
0.
39
1
7 
0.
39
8
7 
0.
41
3
5
0.
37
2
2
0.
38
2
1 
0.
34
2
1
0.
48
6
1
0.
37
9
3
0.
12
5
0 
0.
35
7
1
0.
39
4
7 
0.
41
8
0
0.
27
0
8 
0.
22
8
7
 
24
8 
0.
44
1
7 
0.
42
2
9 
0.
40
9
8
0.
37
7
8
0.
36
1
8 
0.
40
7
9
0.
37
5
0
0.
50
0
0
0.
87
5
0 
0.
54
4
6
0.
44
7
4 
0.
41
8
0
0.
56
2
5 
0.
53
3
6
 
25
0 
0.
01
4
6 
0.
00
8
8 
0.
00
3
8
0.
01
1
1
- 
0.
01
3
2
0.
01
3
9
0.
01
7
2
- 
- 
- 
- 
- 
- 
 
25
2 
0.
13
7
5 
0.
14
3
2 
0.
16
1
7
0.
23
8
9
0.
25
2
0 
0.
22
3
7
0.
12
5
0
0.
10
3
4
- 
0.
09
8
2
0.
15
7
9 
0.
16
3
9
0.
16
6
7 
0.
22
8
7
 
25
4 
0.
01
4
6 
0.
02
6
4 
0.
01
1
3
- 
0.
00
4
1 
- 
- 
- 
- 
- 
- 
- 
- 
- 
Ppa
-4
6
 
22
2 
- 
- 
- 
- 
- 
- 
- 
- 
0.
25
0
0
 
- 
- 
- 
- 
- 
 
22
6 
- 
- 
- 
- 
- 
- 
- 
0.
03
2
3
0.
25
0
0 
- 
- 
- 
- 
- 
 
22
9 
0.
33
5
1 
0.
34
2
2 
0.
38
4
3
0.
50
0
0
0.
44
3
5 
0.
36
3
6
0.
32
4
3
0.
29
0
3
0.
06
3
0 
0.
37
9
3
0.
23
6
8 
0.
33
0
8
0.
30
2
1 
0.
28
9
0
 
23
1 
0.
59
1
6 
0.
61
4
1 
0.
57
4
4
0.
40
7
6
0.
47
1
8 
0.
53
2
5
0.
64
8
6
0.
59
6
8
0.
37
5
0 
0.
57
7
6
0.
71
0
5 
0.
60
0
0
0.
64
5
8 
0.
66
6
7
 
23
5 
0.
07
3
3 
0.
04
3
7 
0.
04
1
3
0.
09
2
4
0.
08
4
7 
0.
09
7
4
0.
02
7
0
0.
08
0
6
0.
06
3
0 
0.
04
3
1
0.
05
2
6 
0.
06
9
2
0.
05
2
1 
0.
04
4
3
 
23
9 
- 
- 
- 
- 
- 
0.
00
6
5
- 
- 
- 
- 
- 
- 
- 
- 
PO
3-
85 
22
7 
0.
08
7
8 
0.
08
1
2 
0.
08
3
9
0.
07
6
1
0.
07
6
6 
0.
09
2
1
0.
17
5
7
0.
13
3
3
- 
0.
07
0
2
0.
08
3
3 
0.
03
9
1
0.
04
1
7 
0.
02
9
7
 
22
9 
0.
01
4
3 
0.
02
1
4 
0.
02
5
5
0.
00
5
4
0.
02
0
2 
0.
03
2
9
- 
- 
- 
- 
- 
- 
- 
- 
 
23
1 
0.
50
4
1 
0.
50
6
4 
0.
48
5
4
0.
47
2
8
0.
54
0
3 
0.
49
3
4
0.
45
9
5
0.
41
6
7
0.
75
0
0 
0.
36
8
4
0.
38
8
9 
0.
50
0
0
0.
51
0
4 
0.
50
2
1
 
23
3 
0.
31
2
2 
0.
00
2
1 
- 
- 
- 
- 
- 
- 
- 
0.
00
8
8
- 
- 
- 
- 
 
23
5 
- 
0.
31
8
4 
0.
34
3
1
0.
28
8
0
0.
25
8
1 
0.
30
2
6
0.
22
9
7
0.
26
6
7
0.
25
0
0 
0.
44
7
4
0.
50
0
0 
0.
32
8
1
0.
25
0
0 
0.
27
1
2
 
23
7 
0.
08
1
6 
0.
07
0
5 
0.
06
2
0
0.
15
7
6
0.
10
4
8 
0.
07
8
9
0.
13
5
1
0.
18
3
3
- 
0.
10
5
3
0.
02
7
8 
0.
13
2
8
0.
19
7
9 
0.
19
7
0
PO
3-
68 
26
3 
- 
- 
- 
- 
- 
0.
00
6
5
- 
- 
- 
- 
- 
- 
- 
- 
 
27
0 
0.
28
5
1 
0.
26
4
1 
0.
25
5
6
0.
22
2
2
0.
25
8
1 
0.
21
4
3
0.
20
0
0
0.
26
6
7
0.
75
0
0 
0.
46
4
9
0.
47
3
7 
0.
46
1
5
0.
42
7
1 
0.
36
1
1
 
27
2 
0.
00
2
1 
0.
00
4
3 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
 
27
6 
0.
12
4
0 
0.
11
9
0 
0.
11
8
5
0.
16
6
7
0.
16
9
4 
0.
20
7
8
0.
27
1
4
0.
36
6
7
- 
0.
17
5
4
0.
21
0
5 
0.
23
8
5
0.
21
8
8 
0.
26
2
8
 
27
8 
0.
16
3
2 
0.
17
9
7 
0.
15
9
3
0.
22
2
2
0.
18
1
5 
0.
18
1
8
0.
27
1
4
0.
21
6
7
- 
0.
20
1
8
0.
07
8
9 
0.
08
4
6
0.
17
7
1 
0.
14
9
6
 
28
0 
0.
05
5
8 
0.
04
7
6 
0.
06
3
0
0.
06
6
7
0.
07
6
6 
0.
08
4
4
0.
05
7
1
- 
- 
- 
0.
02
6
3 
0.
02
3
1
0.
04
1
7 
0.
03
2
1
 
28
2 
0.
06
8
2 
0.
09
5
2 
0.
08
5
2
0.
06
1
1
0.
09
2
7 
0.
09
0
9
0.
04
2
9
0.
10
0
0
- 
0.
13
1
6
0.
15
7
9 
0.
15
3
8
0.
12
5
0 
0.
17
9
5
 
28
4 
0.
30
1
7 
0.
29
0
0 
0.
31
8
5
0.
26
1
1
0.
22
1
8 
0.
21
4
3
0.
15
7
1
0.
05
0
0
0.
25
0
0 
0.
02
6
3
0.
05
2
6 
0.
03
8
5
0.
01
0
4 
0.
01
5
0
 
301
Loc
u
s 
Al
l
e
l
e
 
1st
-
95 
2n
d
-9
5
 
3r
d-
9
5
1st
-
96
2n
d
-9
6
3r
d-
9
6
1st
-
97
1st
-
98
1st
-
99 
1st
-
00
2n
d
-0
0
1st
-
01
2n
d
-0
1
1st
-
02
PO-
2
5 
14
3 
- 
- 
- 
- 
- 
0.
00
6
7
- 
- 
- 
- 
- 
- 
- 
- 
 
14
5 
- 
- 
- 
- 
- 
0.
01
3
3
- 
- 
- 
- 
- 
- 
0.
02
0
8 
0.
00
2
3
 
14
7 
0.
33
8
8 
0.
34
1
9 
0.
36
0
3
0.
38
8
9
0.
38
7
5 
0.
40
0
0
0.
44
5
9
0.
41
6
7
0.
37
5
0 
0.
56
3
6
0.
32
3
5 
0.
38
6
8
0.
27
0
8 
0.
40
2
7
 
14
9 
0.
11
2
2 
0.
10
6
8 
0.
08
0
9
0.
04
4
4
0.
09
5
8 
0.
09
3
3
0.
09
4
6
0.
13
3
3
0.
12
5
0 
0.
14
5
5
0.
20
5
9 
0.
13
2
1
0.
20
8
3 
0.
13
5
7
 
15
1 
0.
04
4
9 
0.
04
4
9 
0.
03
6
8
0.
06
6
7
0.
03
7
5 
0.
06
0
0
0.
06
7
6
0.
01
6
7
0.
12
5
0 
0.
10
9
1
0.
23
5
3 
0.
05
6
6
0.
11
4
6 
0.
16
5
2
 
15
3 
- 
0.
00
2
1 
0.
00
7
4
0.
01
6
7
0.
01
6
7 
0.
02
6
7
- 
- 
- 
- 
- 
- 
- 
- 
 
16
3 
0.
00
6
1 
0.
00
6
4 
- 
- 
- 
0.
00
6
7
- 
- 
- 
- 
- 
- 
0.
01
0
4 
0.
00
2
3
 
16
9 
0.
49
8
0 
0.
49
7
9 
0.
51
4
7
0.
48
3
3
0.
46
2
5 
0.
39
3
3
0.
39
1
9
0.
43
3
3
0.
37
5
0 
0.
18
1
8
0.
23
5
3 
0.
42
4
5
0.
37
5
0 
0.
29
1
9
Pm
l-03 
23
5 
- 
- 
- 
- 
- 
0.
00
7
4
- 
- 
- 
- 
- 
- 
- 
- 
 
23
7 
0.
57
4
4 
0.
58
0
9 
0.
55
9
8
0.
49
0
6
0.
52
5
0 
0.
57
3
5
0.
45
8
3
0.
37
1
0
0.
40
0
0 
0.
71
5
5
0.
78
9
5 
0.
69
2
3
0.
68
7
5 
0.
66
4
2
 
24
9 
- 
- 
- 
- 
- 
0.
00
7
4
- 
- 
- 
- 
- 
- 
- 
- 
 
25
1 
0.
19
2
3 
0.
17
1
6 
0.
15
8
1
0.
27
3
6
0.
23
5
0 
0.
13
9
7
0.
18
0
6
0.
16
1
3
- 
0.
03
4
5
0.
10
5
3 
0.
06
1
5
0.
03
1
3 
0.
05
6
4
 
25
3 
0.
00
2
6 
0.
00
2
5 
0.
00
4
3
- 
0.
13
5
0 
0.
15
4
4
0.
25
0
0
0.
17
7
4
0.
40
0
0 
0.
07
7
6
0.
05
2
6 
0.
02
3
1
0.
02
0
8 
0.
02
4
5
 
25
5 
0.
13
0
8 
0.
12
0
1 
0.
16
2
4
0.
16
0
4
0.
01
0
0 
0.
02
2
1
- 
- 
- 
- 
- 
- 
0.
01
0
4 
- 
 
25
9 
0.
10
0
0 
0.
12
5
0 
0.
11
5
4
0.
07
5
5
0.
09
5
0 
0.
09
5
6
0.
11
1
1
0.
29
0
3
0.
20
0
0 
0.
17
2
4
0.
05
2
6 
0.
22
3
1
0.
25
0
0 
0.
25
4
9
Pm
l
-
11 
25
6 
0.
27
1
4 
0.
30
0
9 
0.
31
7
2
0.
33
5
2
0.
34
7
1 
0.
34
0
0
0.
45
9
5
0.
40
0
0
0.
62
5
0 
0.
26
7
2
0.
28
9
5 
0.
21
7
7
0.
34
3
8 
0.
27
5
1
 
25
8 
- 
0.
00
2
2 
0.
00
3
7
- 
- 
0.
00
6
7
- 
- 
- 
- 
- 
- 
- 
- 
 
26
0 
0.
05
1
0 
0.
05
1
9 
0.
04
8
5
0.
04
4
0
0.
05
3
7 
0.
04
6
7
0.
01
3
5
0.
01
6
7
0.
12
5
0 
0.
03
4
5
- 
0.
01
6
1
- 
0.
02
6
2
 
26
2 
0.
24
9
0 
0.
22
5
1 
0.
22
3
9
0.
17
0
3
0.
17
7
7 
0.
20
0
0
0.
17
5
7
0.
20
0
0
0.
12
5
0 
0.
33
6
2
0.
34
2
1 
0.
32
2
6
0.
30
2
1 
0.
32
1
0
 
26
4 
0.
29
1
8 
0.
28
5
7 
0.
27
6
1
0.
26
9
2
0.
26
4
5 
0.
30
6
7
0.
25
6
8
0.
31
6
7
0.
12
5
0 
0.
31
0
3
0.
26
3
2 
0.
33
0
6
0.
22
9
2 
0.
22
4
9
 
26
6 
0.
08
7
8 
0.
08
6
6 
0.
08
9
6
0.
15
9
3
0.
08
6
8 
0.
04
6
7
0.
09
4
6
0.
06
6
7
- 
0.
01
7
2
- 
0.
02
4
2
0.
05
2
1 
0.
07
8
6
 
26
8 
0.
04
9
0 
0.
04
7
6 
0.
04
1
0
0.
02
2
0
0.
07
0
2 
0.
05
3
3
- 
- 
- 
0.
03
4
5
0.
10
5
3 
0.
08
8
7
0.
07
2
9 
0.
07
4
2
Pm
l
-
04 
22
2 
0.
26
0
2 
0.
26
7
1 
0.
23
6
6
0.
29
8
9
0.
29
3
4 
0.
31
4
1
0.
21
6
2
0.
25
8
1
0.
50
0
0 
0.
35
0
9
0.
44
7
4 
0.
41
5
4
0.
32
2
9 
0.
37
3
9
 
22
4 
- 
- 
- 
- 
0.
00
4
1 
0.
00
6
4
0.
01
3
5
- 
- 
0.
00
8
8
- 
0.
00
7
7
- 
0.
00
2
1
 
22
6 
0.
69
2
3 
0.
68
7
2 
0.
72
5
2
0.
63
5
9
0.
65
2
9 
0.
64
1
0
0.
68
9
2
0.
70
9
7
0.
50
0
0 
0.
64
0
4
0.
55
2
6 
0.
57
6
9
0.
66
6
7 
0.
61
5
4
 
22
8 
0.
04
7
5 
0.
04
5
7 
0.
03
8
2
0.
06
5
2
0.
04
9
6 
0.
02
5
6
0.
08
1
1
0.
03
2
3
- 
- 
- 
- 
0.
01
0
4 
0.
00
8
5
 
23
0 
- 
- 
- 
- 
- 
0.
01
2
8
- 
- 
- 
- 
- 
- 
- 
- 
 
302
Loc
u
s 
Al
l
e
l
e
 
1st
-
95 
2n
d
-9
5
 
3r
d-
9
5
1st
-
96
2n
d
-9
6
3r
d-
9
6
1st
-
97
1st
-
98
1st
-
99 
1st
-
00
2n
d
-0
0
1st
-
01
2n
d
-0
1
1st
-
02
PO-
7
1 
23
7 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
0.
00
2
2
 
24
1 
0.
01
7
5 
0.
01
5
9 
0.
01
9
8
- 
0.
00
4
2 
0.
00
7
0
0.
02
9
4
0.
01
7
9
- 
0.
08
8
2
0.
07
8
9 
0.
03
9
1
0.
02
0
8 
0.
02
6
3
 
24
3 
0.
41
0
1 
0.
39
0
9 
0.
40
8
7
0.
43
1
8
0.
44
5
8 
0.
39
4
4
0.
52
9
4
0.
55
3
6
0.
75
0
0 
0.
52
9
4
0.
39
4
7 
0.
39
0
6
0.
46
8
8 
0.
48
6
8
 
24
5 
0.
32
8
9 
0.
34
5
5 
0.
35
7
1
0.
26
1
4
0.
28
7
5 
0.
29
5
8
0.
19
1
2
0.
08
9
3
0.
12
5
0 
0.
17
6
5
0.
13
1
6 
0.
27
3
4
0.
33
3
3 
0.
27
1
9
 
24
7 
0.
02
1
9 
0.
01
5
9 
0.
01
9
8
0.
02
2
7
0.
00
8
3 
0.
01
4
1
- 
0.
07
1
4
0.
12
5
0 
0.
06
8
6
0.
05
2
6 
0.
03
9
1
0.
01
0
4 
0.
00
6
6
 
24
9 
0.
05
9
2 
0.
07
2
7 
0.
05
5
6
0.
13
0
7
0.
12
5
0 
0.
11
2
7
0.
10
2
9
0.
07
1
4
- 
0.
07
8
4
0.
21
0
5 
0.
22
6
6
0.
15
6
3 
0.
16
0
1
 
25
1 
0.
13
6
0 
0.
12
9
5 
0.
10
7
1
0.
10
2
3
0.
08
7
5 
0.
12
6
8
0.
10
2
9
0.
19
6
4
- 
0.
05
8
8
0.
13
1
6 
0.
02
3
4
- 
0.
01
3
2
 
25
3 
0.
02
1
9 
0.
02
7
3 
0.
02
7
8
0.
03
9
8
0.
02
9
2 
0.
04
2
3
0.
04
4
1
- 
- 
- 
- 
0.
00
7
8
0.
01
0
4 
0.
02
4
1
 
25
8 
0.
00
2
2
 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
- 
 
26
2 
0.
00
2
2 
0.
00
2
3 
0.
00
4
0
0.
01
1
4
0.
01
2
5 
0.
00
7
0
- 
- 
- 
- 
- 
- 
- 
0.
00
8
8
Pm
l
-
06 
13
3 
0.
46
7
2 
0.
46
3
7 
0.
42
7
0
0.
43
4
8
0.
47
5
2 
0.
55
4
1
0.
41
8
9
0.
46
6
7
0.
50
0
0 
0.
24
5
6
0.
26
4
7 
0.
26
1
5
0.
33
3
3 
0.
28
6
3
 
14
8 
0.
19
0
6 
0.
18
3
8 
0.
18
6
1
0.
20
1
1
0.
15
7
0 
0.
16
2
2
0.
16
2
2
0.
11
6
7
0.
10
0
0 
0.
17
5
4
0.
08
8
2 
0.
15
3
8
0.
08
3
3 
0.
16
2
4
 
15
2 
0.
34
2
2 
0.
35
2
6 
0.
38
6
9
0.
36
4
1
0.
36
7
8 
0.
28
3
8
0.
41
8
9
0.
41
6
7
0.
40
0
0 
0.
57
8
9
0.
64
7
1 
0.
58
4
6
0.
58
3
3 
0.
55
1
3
 
303
Appendix 2  Exact tests of Hardy-Weinberg equilibrium for each locus based on the 
Markov chain method for Gazebo grid, Vet Village grid and the combined grids. 
 
 Feb-95  Apr-95  Jun-95  Feb-96  
Locus P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 0.2288 0.0104 0.4896 0.0072 0.5848 0.0057 0.3343 0.0060
Ppa-12 0.2065 0.0097 0.3433 0.0103 0.1007 0.0034 0.1624 0.0042
Ppa-46 0.4577 0.0066 0.3382 0.0060 1.0000 0.0000 0.5810 0.0058
PO3-85 0.9699 0.0026 0.9462 0.0043 0.2097 0.0102 0.7855 0.0059
PO3-68 0.5631 0.0165 0.1711 0.0131 0.3816 0.0109 0.1778 0.0085
PO-25 0.6123 0.0131 0.1725 0.0099 0.4500 0.0136 0.9837 0.0017
Pml-03 0.6292 0.0106 0.0030 0.0008 0.0072 0.0019 0.3409 0.0078
Pml-11 0.7762 0.0098 0.9454 0.0041 0.2777 0.0104 0.2638 0.0120
Pml-04 0.3704 0.0082 0.7154 0.0067 0.0794 0.0021 0.5474 0.0050
PO-71 0.0228 0.0100 0.4276 0.0168 0.5623 0.0171 0.5719 0.0108
Pml-06 0.5402 0.0076 0.4787 0.0066 0.9165 0.0020 0.8098 0.0028
         
 Apr-96  Jun-96  Feb-97  Apr-98  
 P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 0.8144 0.0029 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000
Ppa-12 0.8170 0.0031 0.1498 0.0105 0.0694 0.0038 0.0507 0.0031
Ppa-46 0.4150 0.0053 0.2037 0.0074 1.0000 0.0000 0.0535 0.0039
PO3-85 0.3576 0.0115 0.1377 0.0082 0.6089 0.0035 0.0819 0.0032
PO3-68 0.2950 0.0100 0.4190 0.0132 0.8349 0.0085 0.6304 0.0073
PO-25 0.9784 0.0017 0.2544 0.0177 0.9244 0.0029 0.5183 0.0073
Pml-03 0.4475 0.0129 0.3929 0.0172 0.0060 0.0011 0.0608 0.0035
Pml-11 0.3516 0.0111 0.1080 0.0098 0.1637 0.0050 0.1467 0.0054
Pml-04 0.5692 0.0043 0.0006 0.0003 0.0343 0.0021 0.5272 0.0054
PO-71 0.3274 0.0092 0.8708 0.0061 1.0000 0.0000 0.7904 0.0087
Pml-06 0.2463 0.0054 0.0668 0.0030 0.6433 0.0046 1.0000 0.0000
         
         
         
         
         
 304
         
 Feb-99  Mar-00  Nov-00  Mar-01  
 P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 0.1419 0.0017 0.2688 0.0030 1.0000 0.0000 1.0000 0.0000
Ppa-12   0.0867 0.0031 0.1474 0.0042 0.2960 0.0044
Ppa-46 0.0643 0.0054 0.4778 0.0052 0.0375 0.0011 1.0000 0.0000
PO3-85 1.0000 0.0000 0.1409 0.0089 1.0000 0.0000 0.7804 0.0050
PO3-68 1.0000 0.0000 0.1632 0.0074 0.6754 0.0067 0.8412 0.0070
PO-25 1.0000 0.0000 0.1627 0.0066 0.0767 0.0032 0.0578 0.0040
Pml-03 1.0000 0.0000 0.2473 0.0082 0.3758 0.0083 0.6636 0.0092
Pml-11 0.4225 0.0096 0.4980 0.0088 0.9551 0.0019 0.5460 0.0087
Pml-04 1.0000 0.0000 0.4949 0.0030 1.0000 0.0000 1.0000 0.0000
PO-71 1.0000 0.0000 0.0785 0.0090 0.1473 0.0084 0.5519 0.0180
Pml-06 0.6134 0.0043 0.0739 0.0026 1.0000 0.0000 0.2811 0.0046
         
 Nov-01  Mar-02      
 P-val S. E. P-val S. E.     
Ppa-01 0.5512 0.0018 1.0000 0.0000     
Ppa-12 1.0000 0.0000 0.1109 0.0102     
Ppa-46 0.5440 0.0045 0.9230 0.0025     
PO3-85 0.0880 0.0029 0.3258 0.0079     
PO3-68 0.9029 0.0061 0.0579 0.0055     
PO-25 0.7880 0.0093 0.4546 0.0144     
Pml-03 1.0000 0.0000 0.0337 0.0024     
Pml-11 0.4736 0.0062 0.2997 0.0111     
Pml-04 1.0000 0.0000 0.1799 0.0068     
PO-71 0.3735 0.0045 0.4160 0.0255     
Pml-06 0.6127 0.0037 0.2393 0.0065     
  
    
  
 
 
 
 
 
 
 
 305
 
 
Vet Village 
 
Mar-95 
  
Apr-95
  
Jun-95 
  
Mar-96 
 
Locus P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 0.3787 0.0044 0.6855 0.0032 0.3293 0.0044 0.3337 0.0031
Ppa-12 0.4883 0.0113 0.6210 0.0105 0.1742 0.0095 0.1240 0.0060
Ppa-46 0.8974 0.0028 0.9564 0.0017 0.9322 0.0019 0.6720 0.0046
PO3-85 0.6941 0.0088 0.4593 0.0137 0.4320 0.0099 0.5999 0.0102
PO3-68 0.2067 0.0106 0.0141 0.0030 0.0312 0.0033 0.7798 0.0079
PO-25 0.3299 0.0085 0.7419 0.0100 0.6670 0.0115 0.5865 0.0123
Pml-03 0.2089 0.0066 0.5165 0.0081 0.0540 0.0038 0.2572 0.0063
Pml-11 0.1602 0.0100 0.8883 0.0075 0.5782 0.0155 0.0372 0.0038
Pml-04 0.8792 0.0027 0.8793 0.0029 1.0000 0.0000 0.9108 0.0022
PO-71 0.3897 0.0228 0.4188 0.0194 0.8770 0.0137 0.3458 0.0125
Pml-06 0.6656 0.0062 0.3584 0.0073 0.4557 0.0060 0.3145 0.0051
         
 Apr-96  Jun-96  Mar-97  Feb-98  
 P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 0.7587 0.0047 0.4516 0.0027 0.7045 0.0018 0.6671 0.0054
Ppa-12 0.9211 0.0039 0.2764 0.0081 0.8975 0.0021 0.6325 0.0045
Ppa-46 0.1819 0.0050 0.6590 0.0048 0.7852 0.0043 0.4997 0.0047
PO3-85 0.1805 0.0084 0.5789 0.0085 0.6562 0.0065 0.0868 0.0031
PO3-68 0.7815 0.0070 0.4180 0.0098 0.5078 0.0100 0.1422 0.0063
PO-25 0.9261 0.0044 0.9372 0.0036 0.2076 0.0060 0.5295 0.0051
Pml-03 0.9822 0.0017 0.2323 0.0126 0.7001 0.0054 0.6606 0.0056
Pml-11 0.5217 0.0109 0.5828 0.0112 0.1891 0.0076 0.1986 0.0047
Pml-04 0.9325 0.0034 0.3136 0.0105 0.8638 0.0043 0.1658 0.0054
PO-71 0.7827 0.0164 0.5339 0.0200 0.2859 0.0122 0.3059 0.0100
Pml-06 0.4677 0.0066 0.0895 0.0032 1.0000 0.0000 0.6678 0.0037
         
         
         
         
         
         
         
 306
         
         
         
         
         
 Mar-00  Dec-00  Mar-05  Oct-05  
 P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 0.2260 0.0024 0.3849 0.0022 0.6590 0.0016 0.1072 0.0020
Ppa-12 1.0000 0.0000 1.0000 0.0000 0.7099 0.0038 0.5790 0.0039
Ppa-46 0.1284 0.0052 0.0804 0.0028 0.3597 0.0052 0.4289 0.0062
PO3-85 0.6830 0.0056 0.7797 0.0031 0.4003 0.0084 0.1579 0.0048
PO3-68 0.6758 0.0063 0.2386 0.0201 0.5548 0.0134 0.4579 0.0105
PO-25 0.3779 0.0050 0.7362 0.0052 0.0940 0.0036 0.0193 0.0022
Pml-03 0.8657 0.0041   0.5356 0.0077 1.0000 0.0000
Pml-11 0.7286 0.0095 0.8127 0.0070 0.6413 0.0076 0.7186 0.0068
Pml-04 0.0090 0.0013 0.0475 0.0011 0.8668 0.0034 0.0609 0.0016
PO-71 0.2945 0.0118 0.0025 0.0011 0.2970 0.0121 0.3734 0.0118
Pml-06 0.6542 0.0028 0.3972 0.0066 0.4161 0.0047 0.5388 0.0050
         
         
Feb-02  
 P-val S. E.       
Ppa-01 0.5824 0.0035       
Ppa-12 0.0714 0.0073       
Ppa-46 0.0388 0.0024       
PO3-85 0.0692 0.0045       
PO3-68 0.8749 0.0058       
PO-25 0.0000 0.0000       
Pml-03 0.6490 0.0061       
Pml-11 0.1683 0.0109       
Pml-04 0.0930 0.0075       
PO-71 0.4676 0.0173       
Pml-06 0.1981 0.0059       
 
 
 
 
 
 
 
 
 307
 
 
 
 
Combined grids 
 
 1st-95  2nd-95  3rd-95  1st-96  
Locus P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 0.1230 0.0106 0.6417 0.0079 0.3458 0.0088 0.7392 0.0043
Ppa-12 0.2794 0.0127 0.2323 0.0114 0.0224 0.0033 0.0846 0.0056
Ppa-46 0.4917 0.0081 0.6616 0.0067 0.9685 0.0013 0.7227 0.0056
PO3-85 0.7769 0.0106 0.6798 0.0141 0.6057 0.0104 0.5558 0.0122
PO3-68 0.7693 0.0153 0.4678 0.0166 0.0703 0.0061 0.3431 0.0101
PO-25 0.4322 0.0128 0.1720 0.0120 0.7719 0.0085 0.8471 0.0067
Pml-03 0.5185 0.0145 0.0889 0.0076 0.0017 0.0008 0.1138 0.0046
Pml-11 0.1208 0.0081 0.7553 0.0134 0.2308 0.0151 0.1759 0.0117
Pml-04 0.5349 0.0084 0.5381 0.0068 0.4287 0.0077 0.7405 0.0046
PO-71 0.0487 0.0091 0.8589 0.0137 0.5346 0.0213 0.0475 0.0061
Pml-06 0.9758 0.0014 0.8759 0.0043 0.7371 0.0060 0.7062 0.0052
         
 2nd-96  3rd-96  1st-97  1st-98  
 P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 1.0000 0.0000 0.0618 0.0072 0.7317 0.0016 0.8522 0.0034
Ppa-12 0.8708 0.0059 0.0498 0.0070 0.3148 0.0080 0.0476 0.0030
Ppa-46 0.0459 0.0034 0.2800 0.0102 0.5259 0.0059 0.0378 0.0027
PO3-85 0.8266 0.0071 0.3061 0.0107 0.6720 0.0064 0.0057 0.0008
PO3-68 0.3691 0.0103 0.1436 0.0107 0.5292 0.0109 0.3076 0.0080
PO-25 0.9861 0.0014 0.3174 0.0195 0.4224 0.0075 0.3496 0.0078
Pml-03 0.5597 0.0117 0.0991 0.0106 0.6035 0.0067 0.5068 0.0056
Pml-11 0.1380 0.0098 0.3515 0.0155 0.5305 0.0094 0.2378 0.0088
Pml-04 0.8849 0.0062 0.0009 0.0005 0.8039 0.0061 0.5311 0.0056
PO-71 0.2874 0.0188 0.3864 0.0216 0.2978 0.0115 0.2423 0.0125
Pml-06 0.4187 0.0067 0.0110 0.0009 0.6121 0.0049 0.7764 0.0034
         
         
         
         
         
 308
         
         
         
         
         
         
         
 1st-00  2nd-00  1st-01  2nd-01  
 P-val S. E. P-val S. E. P-val S. E. P-val S. E. 
Ppa-01 1.0000 0.0000 0.5875 0.0019 1.0000 0.0000 0.7078 0.0017
Ppa-12 0.3559 0.0054 0.4690 0.0045 0.4950 0.0053 0.8102 0.0030
Ppa-46 0.2171 0.0049 0.0169 0.0015 0.5061 0.0046 0.7635 0.0042
PO3-85 0.4645 0.0127 0.9207 0.0029 0.3234 0.0071 0.1101 0.0050
PO3-68 0.1870 0.0069 0.6540 0.0115 0.6084 0.0113 0.5726 0.0128
PO-25 0.8311 0.0042 0.1031 0.0033 0.1869 0.0063 0.0166 0.0027
Pml-03 0.4665 0.0088 0.1072 0.0065 0.8196 0.0066 1.0000 0.0000
Pml-11 0.6800 0.0113 0.8773 0.0031 0.6046 0.0124 0.5246 0.0084
Pml-04 0.0362 0.0036 0.3509 0.0021 0.7617 0.0050 0.0681 0.0044
PO-71 0.0211 0.0036 0.0068 0.0012 0.2366 0.0165 0.1224 0.0109
Pml-06 0.4605 0.0051 0.4438 0.0047 0.5308 0.0054 0.5698 0.0048
         
         
 1st-02        
 P-val S. E.       
Ppa-01 0.5269 0.0051       
Ppa-12 0.0002 0.0002       
Ppa-46 0.3938 0.0092       
PO3-85 0.4740 0.0120       
PO3-68 0.1666 0.0117       
PO-25 0.0006 0.0006       
Pml-03 0.0786 0.0045       
Pml-11 0.7342 0.0119       
Pml-04 1.0000 0.0000       
PO-71 0.4960 0.0248       
Pml-06 0.3584 0.0088       
 
 309
 310
Appendix 3  Spatial autocorrelation analysis for each trapping period on Gazebo, Vet 
Village and the combined grids after Peakall et al. (2003).  The distance classes used  
in the analysis are 20, 50, 100 and 250 m for Gazebo and Vet Village grids.  The  
distance classes used for the combined grids were 20, 50, 100, 200, 320 and 800 m.   
The number of pairwise comparisons (n), the correlation coefficient (r), 95% CI upper 
(U) and lower bounds (L), bootstrap resampling upper (Ur) and lower (Lr) bounds  
about r, the probability based on a one-tailed test (P) and the estimated x-intercept.   
For distance classes where significance was not found the x-intercept was made zero. 
 
Feb-95     Apr-95    
D 20 50 100 250 20 50 100 250
n 127 658 1733 6576 74 397 1152 4617
r 0.131 0.061 0.031 0.008 0.063 0.046 0.022 0.008
U 0.042 0.018 0.011 0.003 0.048 0.021 0.012 0.004
L -0.038 -0.017 -0.009 -0.003 -0.046 -0.018 -0.011 -0.003
P 0.001 0.001 0.001 0.001 0.008 0.001 0.002 0.002
Ur 0.178 0.082 0.043 0.014 0.111 0.069 0.034 0.015
Lr 0.083 0.044 0.020 0.002 0.012 0.022 0.010 0.002
Int 95 150 193 348 196 215 252 346
         
Jun-95     Feb-96    
D 20 50 100 250 20 50 100 250
n 24 130 401 1395 24 84 274 806
r -0.025 -0.009 0.010 0.006 0.123 0.053 0.025 0.004
U 0.083 0.035 0.018 0.008 0.088 0.048 0.025 0.009
L -0.080 -0.032 -0.017 -0.006 -0.081 -0.041 -0.022 -0.007
P 0.723 0.678 0.134 0.063 0.005 0.017 0.027 0.153
Ur 0.079 0.022 0.030 0.017 0.223 0.104 0.054 0.020
Lr -0.131 -0.044 -0.010 -0.006 0.035 -0.002 -0.007 -0.011
Int 0 0 0 0 107 0 0 0
         
 
 
 
 
 
 
 
 
 
     
 
 
 
 
 
 
 
 
   
 
 
Apr-96 
 
 
Jun-96 
D 20 50 100 250 20 50 100 250
n 13 92 241 926 7 56 147 451
r 0.074 -0.007 0.008 0.003 -0.078 -0.005 0.014 -0.004
U 0.145 0.047 0.027 0.009 0.146 0.053 0.032 0.008
L -0.123 -0.046 -0.023 -0.007 -0.143 -0.047 -0.025 -0.007
P 0.131 0.612 0.246 0.186 0.855 0.562 0.159 0.857
Ur 0.233 0.042 0.039 0.019 0.034 0.055 0.048 0.014
Lr -0.095 -0.059 -0.022 -0.012 -0.193 -0.061 -0.02 -0.022
Int 0 0 0 0 0 0 0 0
         
         
Feb-97     Apr-98    
D 20 50 100 250 20 50 100 250
n 3 4 5 25 1 3 14 29
r 0.415 0.273 0.339 -0.056 0.281 0.147 -0.011 -0.010
U 0.414 0.312 0.245 0.055 0.517 0.183 0.067 0.023
L -0.211 -0.216 -0.209 -0.053 -0.329 -0.177 -0.053 -0.023
P 0.021 0.039 0.007 0.986 0.071 0.056 0.645 0.820
Ur 0.455 0.443 0.527 0.053 0.128 0.276 0.065 0.046
Lr 0.342 -0.026 0.071 -0.157 0.128 0.064 -0.080 -0.065
Int 48 0 173 0 0 0 0 0
         
         
Feb-99     Mar-00    
D 20 50 100 250 20 50 100 250
n 4 6 9 32 14 55 150 433
r 0.277 0.033 -0.021 0.006 0.091 0.107 0.018 0.010
U 0.774 0.623 0.410 0.073 0.135 0.070 0.035 0.007
L -0.330 -0.196 -0.169 -0.069 -0.122 -0.058 -0.030 -0.007
P 0.154 0.277 0.375 0.385 0.085 0.006 0.139 0.006
Ur 0.954 0.529 0.449 0.231 0.237 0.186 0.064 0.035
Lr -0.394 -0.447 -0.451 -0.192 -0.057 0.038 -0.033 -0.014
Int 0 0 0 0 0 88 0 0
         
 
 
 
     
 
 
   
 311
 
 
 
 
Nov-00 
 
 
 
 
Mar-01 
D 20 50 100 250 20 50 100 250
n 2 5 14 50 5 20 57 186
r 0.230 0.001 -0.015 0.001 0.061 0.111 0.083 0.015
U 0.342 0.250 0.120 0.017 0.198 0.093 0.047 0.011
L -0.317 -0.173 -0.099 -0.018 -0.176 -0.075 -0.039 -0.011
P 0.092 0.448 0.568 0.464 0.244 0.013 0.002 0.007
Ur 0.093 0.356 0.138 0.073 0.287 0.209 0.141 0.043
Lr 0.093 -0.256 -0.139 -0.071 -0.149 0.009 0.026 -0.015
Int 0 0 0 0 0 204 218 0
         
         
Nov-01     Mar-02    
D 20 50 100 250 20 50 100 250
n 3 5 15 55 138 707 1711 5766
r 0.312 0.225 0.093 0.030 0.127 0.058 0.021 0.006
U 0.186 0.164 0.090 0.018 0.037 0.016 0.010 0.002
L -0.168 -0.154 -0.084 -0.016 -0.035 -0.015 -0.008 -0.002
P 0.001 0.001 0.017 0.001 0.001 0.001 0.001 0.001
Ur 0.381 0.323 0.172 0.071 0.176 0.075 0.031 0.012
Lr 0.247 0.135 0.019 -0.013 0.081 0.039 0.010 0.000
Int 78 128 219 0 85 95 216 297
 
 
Vet Village grid 
Mar-95     Apr-95    
D 20 50 100 250 20 50 100 250
n 56 376 1005 3739 55 398 1234 4276
r 0.202 0.091 0.034 0.001 0.073 0.040 0.022 0.004
U 0.059 0.021 0.012 0.004 0.057 0.021 0.011 0.004
L -0.053 -0.019 -0.010 -0.003 -0.054 -0.019 -0.010 -0.003
P 0.001 0.001 0.001 0.196 0.011 0.001 0.001 0.012
Ur 0.286 0.117 0.048 0.008 0.143 0.066 0.035 0.011
Lr 0.117 0.064 0.019 -0.006 0.008 0.013 0.008 -0.002
Intercept 77 99 186 0 80 147 198 0
         
 312
 
 
 
 
 
Jun-95 
     
 
 
 
 
Mar-96 
   
D 20 50 100 250 20 50 100 250
n 19 156 479 1649 15 81 209 663
r 0.019 0.031 0.004 0.002 0.029 0.046 0.025 0.011
U 0.099 0.033 0.018 0.006 0.115 0.049 0.027 0.009
L -0.099 -0.032 -0.016 -0.005 -0.105 -0.043 -0.024 -0.008
P 0.353 0.035 0.326 0.174 0.268 0.029 0.038 0.013
Ur 0.152 0.071 0.023 0.013 0.186 0.103 0.058 0.027
Lr -0.099 -0.004 -0.015 -0.009 -0.117 -0.013 -0.012 -0.007
Intercept 0 0 0 0 0 0 0 0
         
Apr-96     Jun-96    
D 20 50 100 250 20 50 100 250
n 25 157 457 1610 9 66 202 653
r -0.010 0.030 0.027 0.007 -0.007 0.011 0.009 0.002
U 0.077 0.033 0.017 0.006 0.124 0.047 0.022 0.007
L -0.077 -0.030 -0.015 -0.005 -0.111 -0.039 -0.020 -0.006
P 0.585 0.041 0.005 0.015 0.539 0.307 0.208 0.323
Ur 0.067 0.064 0.046 0.017 0.108 0.057 0.033 0.015
Lr -0.100 -0.003 0.009 -0.003 -0.115 -0.032 -0.016 -0.012
Intercept 0 0 225 0 0 0 0 0
         
         
Mar-97     Feb-98    
D 20 50 100 250 20 50 100 250
n 5 28 86 261 5 16 48 186
r 0.287 0.183 0.084 0.010 0.083 0.013 0.058 0.030
U 0.235 0.093 0.045 0.017 0.225 0.124 0.066 0.016
L -0.194 -0.078 -0.039 -0.012 -0.212 -0.115 -0.058 -0.011
P 0.009 0.001 0.003 0.097 0.232 0.419 0.041 0.001
Ur 0.517 0.297 0.135 0.041 0.400 0.142 0.125 0.065
Lr 0.008 0.068 0.026 -0.019 -0.117 -0.078 -0.015 -0.004
Intercept 106 121 174 0 0 0 0 0
 
 
     
 
   
 313
 314
Mar-00 Dec-00 
D 20 50 100 250 20 50 100 250
n 4 33 104 252 2 3 4 22
r 0.112 0.005 0.054 0.013 -0.197 -0.007 -0.066 -0.021
U 0.258 0.084 0.040 0.015 0.366 0.269 0.267 0.037
L -0.240 -0.083 -0.034 -0.010 -0.318 -0.168 -0.181 -0.037
P 0.192 0.452 0.004 0.043 0.896 0.508 0.681 0.846
Ur 0.309 0.122 0.109 0.048 0.149 0.320 0.180 0.082
Lr -0.146 -0.100 -0.003 -0.021 0.149 -0.352 -0.280 -0.120
Intercept 0 0 0 0 0 0 0 0
 
 
        
Mar-01     Oct-01    
D 20 50 100 250 20 50 100 250
n 43 154 261 752 12 79 195 552
r 0.197 0.069 0.050 0.009 0.109 0.110 0.036 0.002
U 0.070 0.034 0.025 0.007 0.121 0.046 0.023 0.006
L -0.058 -0.028 -0.020 -0.006 -0.117 -0.040 -0.021 -0.005
P 0.001 0.002 0.001 0.004 0.044 0.001 0.006 0.197
Ur 0.267 0.114 0.080 0.024 0.235 0.167 0.068 0.021
Lr 0.127 0.029 0.021 -0.008 -0.013 0.053 0.005 -0.015
Intercept 60 126 164 0 0 94 172 0
        
Feb-02     
 
   
D 20 50 100 250  
n 133 821 2234 6012    
r 0.22 0.075 0.026 0.002  
U 0.037 0.014 0.007 0.002    
L -0.036 -0.013 -0.007 -0.002  
P 
Ur 
0.001 
0.267 
0.001 
0.095 
0.001
0.036
0.028
0.008
Lr 0.17 0.056 0.017 -0.004   
Intercept 0 99 175 0   
   
 
 
 
 
 
 
 
 315
 
Combined grids 
 
1st-95   
D 20 50 100 200 320 800
n 183 1034 2738 7586 12753 16562
r 0.146 0.074 0.045 0.026 0.022 0.016
U 0.027 0.011 0.007 0.004 0.003 0.002
L -0.025 -0.011 -0.006 -0.003 -0.002 -0.002
P 0.001 0.001 0.001 0.001 0.001 0.001
Ur 0.181 0.087 0.052 0.031 0.026 0.019
Lr 0.109 0.059 0.037 0.022 0.019 0.013
Intercept 341 506 524 639 704 1166
   
2nd-95   
D 20 50 100 200 320 800
n 129 795 2386 6778 10918 14643
r 0.096 0.072 0.048 0.036 0.028 0.023
U 0.038 0.014 0.007 0.004 0.003 0.002
L -0.033 -0.014 -0.007 -0.004 -0.002 -0.002
P 0.001 0.001 0.001 0.001 0.001 0.001
Ur 0.140 0.088 0.056 0.041 0.032 0.026
Lr 0.053 0.054 0.039 0.031 0.024 0.020
Intercept 400 539 576 701 777 1175
   
3rd-95   
D 20 50 100 200 320 800
n 43 286 880 2377 3708 5151
r 0.013 0.029 0.026 0.030 0.026 0.020
U 0.058 0.024 0.013 0.008 0.005 0.004
L -0.057 -0.025 -0.012 -0.007 -0.004 -0.003
P 0.324 0.010 0.001 0.001 0.001 0.001
Ur 0.093 0.053 0.040 0.039 0.032 0.025
Lr -0.068 0.004 0.013 0.021 0.019 0.014
Intercept 398 449 672 674 775 1172
   
 
 
  
 
 
1st-96 
D 20 50 100 200 320 800
n 39 165 483 1176 1741 2204
r 0.156 0.117 0.087 0.069 0.068 0.057
U 0.068 0.031 0.016 0.012 0.009 0.007
L -0.068 -0.031 -0.017 -0.010 -0.007 -0.006
P 0.002 0.001 0.001 0.001 0.001 0.001
Ur 0.240 0.159 0.111 0.083 0.078 0.067
Lr 0.081 0.073 0.065 0.057 0.057 0.048
Intercept 336 506 512 740 768 1181
   
2nd-96   
D 20 50 100 200 320 800
n 38 249 698 2036 3100 4075
r 0.037 0.033 0.044 0.035 0.027 0.020
U 0.068 0.025 0.013 0.008 0.006 0.005
L -0.059 -0.026 -0.014 -0.007 -0.055 -0.004
P 0.128 0.007 0.001 0.001 0.001 0.001
Ur 0.111 0.060 0.060 0.043 0.034 0.026
Lr -0.041 0.005 0.028 0.024 0.020 0.014
Intercept 291 396 458 518 614 1178
   
3rd-96   
D 20 50 100 200 320 800
n 16 122 349 919 1275 1535
r -0.029 0.015 0.025 0.012 0.010 0.009
U 0.087 0.033 0.017 0.011 0.008 0.007
L -0.090 -0.030 -0.017 -0.009 -0.007 -0.005
P 0.741 0.170 0.003 0.011 0.011 0.006
Ur 0.045 0.053 0.045 0.025 0.020 0.018
Lr -0.105 -0.022 0.006 0.000 0.000 0.000
Intercept 113 185 438 450 1207 1196
   
   
   
 316
 
 
 
1st-97 
  
D 20 50 100 200 320 800
n 8 32 91 239 354 429
r 0.410 0.227 0.119 0.027 0.018 0.018
U 0.160 0.080 0.046 0.022 0.015 0.011
L -0.142 -0.074 -0.040 -0.019 -0.011 -0.008
P 0.001 0.001 0.001 0.008 0.014 0.004
Ur 0.524 0.329 0.173 0.061 0.043 0.041
Lr 0.219 0.098 0.061 -0.008 -0.006 -0.005
Intercept 109 130 181 750 778 1110
   
1st-98   
D 20 50 100 200 320 800
n 6 19 62 168 257 267
r 0.109 0.043 0.038 0.035 0.006 0.004
U 0.176 0.100 0.055 0.025 0.017 0.017
L -0.164 -0.087 -0.045 -0.019 -0.012 -0.011
P 0.116 0.179 0.070 0.006 0.199 0.261
Ur 0.330 0.134 0.093 0.069 0.035 0.032
Lr -0.071 -0.047 -0.021 0.000 -0.023 -0.021
Intercept 80 245 248 284 347 1180
   
1st-00   
D 20 50 100 200 320 800
n 18 88 254 612 778 821
r 0.113 0.138 0.095 0.065 0.051 0.047
U 0.116 0.060 0.032 0.018 0.016 0.015
L -0.107 -0.050 -0.024 -0.033 -0.010 -0.009
P 0.029 0.001 0.001 0.001 0.001 0.001
Ur 0.227 0.202 0.133 0.086 0.069 0.068
Lr -0.019 0.072 0.056 0.044 0.033 0.030
Intercept 219 272 298 389 564 1202
   
   
 317
 318
   
2nd-00   
D 20 50 100 200 320 800
n 2 8 18 68 80 83
r 0.248 -0.038 -0.017 0.016 0.020 0.017
U 0.320 0.168 0.105 0.048 0.046 0.050
L -0.285 -0.145 -0.090 -0.037 -0.032 -0.029
P 0.071 0.663 0.626 0.209 0.154 0.174
Ur 0.065 0.223 0.106 0.087 0.072 0.079
Lr 0.065 -0.234 -0.132 -0.041 -0.032 -0.034
Intercept 31 197 364 870 434 1230
   
1st-01   
D 20 50 100 200 320 800
n 48 174 318 745 1107 1134
r 0.224 0.107 0.089 0.039 0.025 0.024
U 0.060 0.029 0.023 0.012 0.009 0.009
L -0.052 -0.025 -0.018 -0.010 -0.006 -0.006
P 0.001 0.001 0.001 0.001 0.001 0.001
Ur 0.286 0.144 0.114 0.055 0.038 0.037
Lr 0.157 0.067 0.062 0.022 0.012 0.010
Intercept 169 176 294 389 602 1161
   
2nd-01   
D 20 50 100 200 320 800
n 15 84 210 526 695 696
r 0.232 0.188 0.111 0.078 0.078 0.077
U 0.125 0.055 0.028 0.012 0.010 0.010
L -0.117 -0.044 -0.023 -0.010 -0.006 -0.006
P 0.001 0.001 0.001 0.001 0.001 0.001
Ur 0.394 0.250 0.146 0.099 0.094 0.094
Lr 0.071 0.133 0.078 0.057 0.060 0.060
Intercept 255 958 917 836 1014 1123
   
   
  
 319
   
1st-02 
D 20 50 100 200 320 800
n 271 1528 3945 9745 13290 13926
r 0.208 0.114 0.067 0.048 0.041 0.040
U 0.025 0.011 0.006 0.004 0.003 0.003
L -0.023 -0.010 -0.006 -0.003 -0.002 -0.002
P 0.001 0.001 0.001 0.001 0.001 0.001
Ur 0.243 0.126 0.075 0.052 0.045 0.043
Lr 0.173 0.100 0.060 0.043 0.037 0.036
Intercept 372 373 806 591 751 1201