Citation
Fine scale genetic structure in bird-dispersed pines

Material Information

Title:
Fine scale genetic structure in bird-dispersed pines evidence for seed dispersal signatures
Creator:
McCune, Ian L
Publication Date:
Language:
English
Physical Description:
54 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Whitebark pine -- Seeds ( lcsh )
Whitebark pine -- Genetics ( lcsh )
Seeds -- Dispersal ( lcsh )
Corvidae -- Behavior ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 49-54).
General Note:
Department of Integrative Biology
Statement of Responsibility:
by Ian L. McCune.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
47916763 ( OCLC )
ocm47916763
Classification:
LD1190.L45 2001m .M32 ( lcc )

Full Text
FINE SCALE GENETIC STRUCTURE IN
BIRD- DISPERSED PINES:
EVIDENCE FOR SEED DISPERSAL SIGNATURES
by
Ian L. McCune
B.A., Drury College, 1993
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Arts
Biology


This thesis for the Masters of Biology
degree by
Ian Lee McCune
has been approved
by
Leo P. Bruederle
Diana F. Tomback

Date
t


McCune, Ian Lee (M.A., Biology)
Fine-Scale Genetic Structure in Bird-Dispersed Pines:
Evidence for Seed Dispersal Signatures
Thesis directed by Associate Professor Leo P. Bruederle
ABSTRACT
Many species of pines (Pinus, Pmaceae) are believed to be bird-dispersed, as
suggested by their seed (large, wingless) and cone morphology. Populations of bird-
dispersed pines exhibit three major growth forms: single-stem tree, single-genet
multi-trunk individual, and multi-genet tree cluster. In whitebark pine (Pinus
albicaulis) populations, there is a high incidence of the cluster growth form, due to
the caching behavior of Clarks Nutcracker (Nucifraga columbmna). This dispersal
method has been shown to lead to a high level of genetic relatedness among
individuals comprising the cluster, and a low level of genetic relatedness for
individuals comprising different clusters, regardless of distance from one another. As
such, evidence exists for a signature in this species that reflects bird-dispersal. For
the purpose of this research, three whitebark pine populations were examined using
starch gel electrophoresis, allozyme analysis, and spatial autocorrelation to identify
dispersal signatures. A single population of southwestern white pine (P.
strobiformis), for which there is evidence suggesting seed dispersal by Clarks
Nutcracker, was similarly analyzed to determine if it exhibited a similar seed
in


dispersal signature. In whitebark pine, there was a high incidence of individuals
comprising multi-stem tree clumps (65.9%, 60.0%, and 75.4%), with 85.7% of the
clumps being multi-genet tree clusters. Additionally, for 18 of 19 polymorphic loci,
individuals comprising a cluster were found to be more closely related genetically
(large positive r-values) than those individuals comprising different clusters (slightly
negative r-values). The individuals comprising different clusters showed random
genetic relatedness within populations. As such, the three populations have a seed
dispersal signature meeting expectations for a bird-dispersed species, which is
strongly supported by statistically significant correlelograms. In southwestern white
pine, 31.3% of the individuals comprised multi-stem tree clumps, with 51.7% of the
clumps being multi-genet tree clusters. Furthermore, for eight of the nine
polymorphic loci, individuals comprising a cluster were found to be more related
genetically than those individuals comprising different clusters. Although there was
evidence of a seed dispersal signature in southwestern white pine, only one locus was
characterized by a significant correlelogram, due in large part to the small sample
size.
This abstract accurately represents the content of the candidate's thesis. I recommend
its publication.
Signed
Leo P. Bruederle
IV


CONTENTS
Figures ........................................... 11
Tables ............................................9
CHAPTER
1. INTRODUCTION....................................1
2. MATERIALS AND METHODS .......................... 8
Study Sites ..................................... 8
Sample Analysis ................................. 10
Statistical Analysis ............................ 14
3. RESULTS .........................................19
Summary Data .................................... 19
Relatedness ......................................22
Spatial Autocorrelation ..........................29
4. DISCUSSION ......................................37
LITERATURE CITED ..................................49
v


FIGURES
Figure
2.1 a-c Site Maps for Pinus albicaulis ......................11
2.2 Site Map for Pinus strobiformis ....................12
3.1 a-c Pairwise Relatedness Regression for Pinus
albicaulis Using a Fixed Distance Class ............26
3.2 a-c Pairwise Relatedness Regression for Pinus
albicaulis Using a Variable Distance Class .........27
3.3 Pairwise Relatedness Regression for Pinus
strobiformis Using a Fixed Distance Class ..........30
3.4 Pairwise Relatedness Regression for Pinus
strobiformis Using a Variable Distance Class .......31
3.5 a-c Spatial Autocorrelograms for Pinus
albicaulis Using a Fixed Distance Class ............33
3.6 a-c Spatial Autocorrelograms for Pinus
albicaulis Using a Variable Distance Class .........34
3.7 Spatial Autocorrelograms for Pinus
strobiformis Using a Fixed Distance Class ..........35
3.8 Spatial Autocorrelograms for Pinus
strobiformis Using a Variable Distance Class .......36
4.1 Percentage of Individuals Comprising
Differing Numbers of Stems for Pinus
albicaulis and P. strobiformis .....................40
4.2 Proportion of Multi-Stemmed Individuals
That Comprise Multi-Genet Tree Clusters ............41
vi


TABLES
Table
2.1 Population summary for Pinus
albicaulis and P. strobiformis..................9
3.1 Allele frequencies for Pinus
albicaulis and P. strobiformis..................20, 21
3.2 Relatedness summary for Pinus albicaulis........24
3.3 Relatedness summary for Pinus strobiformis......28
vii


CHAPTER 1
INTRODUCTION
The evolutionary potential of a population or species over time can be
influenced by the spatial distribution of genetic variability (Dewey and Heywood
1988). Many evolutionary factors, such as genetic drift, gene flow, migration, and
selection, as well as other ecological factors, such as habitat and lifespan (Leonardi
and Menozzi 1996), may affect the genetic structure in a population. Mating systems
and, more specifically, dispersal methods are often the primary reason for spatial
genetic heterogeneity (Maki and Masuda 1994; Leonardi and Menozzi 1996).
Thirteen species of pines are known to have seeds that are dispersed by
members of the Corvidae, which includes the Clarks Nutcracker (Nucifraga
columbiana) and the Eurasian Nutcracker (N. caryoatactes). Those pine species for
which bird-dispersal has been documented include Pinus albicaulis Engelmann, P.
cembra L., P. sibirica Du Tour, P. koraiensis Siebold and Zuccarini. and P. pumila
Regal in subsection Cembra; P. parviflora Siebold and Zuccarini, P.flexilis E. James,
P. armandii Franchet, P. pence Griseb. and P. strobiformis Engelmann in subsection
Strobi; and the pinyons, P. edulis Engelmann
1


and P. monophylla Torrey and Fremont in subsection Cembroides. Other species,
such as P. gerardiana Wall and P. bungeanci Zuccarini, are also documented as
having some bird-dispersal of their seeds. Although the aforementioned pine species
differ with regard to cone morphology, they all have large wingless seeds that
facilitate bird-mediated dispersal of pine seeds (Bruederle et al. 2001) as do a number
of other pine species, although bird-dispersal has not yet been documented for those
species.
Pinus albicaulis (Pinaceae), whitebark pine, is a member of Pinus subgenus
Strobus section Strobi subsection Cembrae (Critchfield and Little 1966). Seed
dispersal in this species is accomplished primarily by the Clarks Nutcracker, which
is the North American congener of the Eurasian Nutcracker. In late summer the
Clark's Nutcracker collects ripe seeds in a sublingual pouch, and caches the seeds for
use in late winter and early spring. The Nutcracker will transport in its pouch up to
150 seeds (Tomback 1982), burying them in caches of one to fifteen seeds under one
to three centimeters of substrate (Tomback and Linhart 1990). These caches may
occur within relatively short distances from maternal trees, but may occur at greater
distances, with some caches being buried as far as twelve kilometers from parent trees
(Tomback 1988). This is in direct contrast to the seeds of wind-dispersed pines,
which are typically dispersed within 120 meters of the parent tree (Tomback and
Schuster 1994). However, seeds in wind-dispersed pine have a leptokurtotic
distribution; that is, they are typically dispersed within close proximity of the parent,
2


with decreasing seed density over distance. Furthermore, there is also expected to be
an inverse relationship between germination and distance from the maternal tree.
This is due to the probability that a seed will fall closer to the parent tree despite any
seed morphology that it may exhibit.
Caches that are not retrieved by the nutcrackers or taken by seed predators
will often germinate forming multi-genet tree clusters (Carsey and Tomback 1994),
the trunks of which may be fused at the base (Linhart and Tomback 1985).
Furthermore, the harvesting and caching of seeds usually involves one maternal tree
(Tomback 1988). As such, the trees within a cluster tend to be more closely related
than those occupying different clusters (Fumier et al. 1987; Tomback and Schuster
1994), often on the order of half to full siblings.
However, a clump may result from other causes, such as lightning, damage
resulting in loss of the leader, or the accumulation of seeds in depressions. In most
cases for wind-dispersed pines, the trunks comprising a tree clump are expected to be
genetically identical, excepting somatic mutation. The term multi-trunked
individual has been applied to describe this growth form.
Whereas wind-dispersed species typically lack the family structure effected by
scatter-hoarding corvids (Linhart et al. 1981; Knowles 1984; Fumier et al. 1987), the
single-stem growth form is expected to predominate (Torick et al. 1996). However, a
multi-trunked individual tree may result for the aforementioned reasons, although the
majority of multi-stemmed individuals are expected to be a single genet.
3


While seed dispersal in P. albicaulis has been well documented, there is
considerably less known about P. strobiformis. Pinus strobiformis is a member of
subgenus Strobus section Strobus subsection Strobi (Critchfield and Little 1966).
Like P.flexilis, another species of subsection Strobi that is known to be dispersed by
the Clarks nutcracker (Linhart and Tomback 1985; Carsey and Tomback 1994), P.
strobiformis has large wingless seeds (Bruederle et al. 2001). However, there has
been little research with regard to seed dispersal in P. strobiformis.
Dispersal signatures are patterns of spatial genetic heterogeneity that are a
result of a species unique life history. Pervious research examining the behavioral
ecology of Clarks Nutcracker and population structure in P. albicaulis have
suggested that a dispersal signature exists for this, and other bird-dispersed pine
species (Linhart and Tomback 1985; Fumier et al. 1987). It is the difference in
dispersal signature between wind and animal-dispersed pines that can be analyzed
effectively through the use of a statistical technique known as spatial autocorrelation.
However, many other statistical methods have been used to elucidate the genetic
structure of various pine species. Included among these are Wrights F-statistics
(Wright 1951). These statistics quantify the reduction in heterozygosity due to
nonrandom mating within populations and within species (Fis and Fit, respectively),
and genetic subdivision for the species (FSt). This last statistic describes genetic
differentiation, specifically. Another statistic, Neis genetic distance (Nei 1973), has
also been used to elucidate spatial genetic structure in a species. The most common
4


method for analyzing statistically the fine-scale spatial patterns in populations is,
however, spatial autocorrelation.
Using spatial autocorrelation, values are obtained to compare genetic
relatedness among individuals at varying distances. Spatial autocorrelation
specifically tests whether an observed value for a variable is independent of values
for variables at other locations (Shapcott 1995). This allows for a determination of
whether or not the genetic relatedness of individuals within a population is dependent
upon their distance from one another. When dependence of one variable upon other
variables in other individuals does occur (ie., genetic relatedness to distance from one
another), spatial autocorrelation is present (Shapcott 1995). Results may be displayed
in the form of a correlelogram, which may display one of five different distributions
for the different distance classes: short distance (low order), with positive or negative
correlation; long distance (high order), with positive or negative correlation; or no
spatial correlation (Sokal and Oden 1978b). When negative short distance correlation
exists, it is often the result of a patchy distribution of environments at a fine scale,
whereas a positive long distance correlation often results from a uniformity of habitat
gradients (Sokal and Oden 1978b). Many populations, however, show a combination
of short distance positive correlation and long distance negative correlation. This
type of correlation could result from clinal variation and selection pressures that differ
among different distance classes. Another possible explanation for decreasing
correlation with distance is the behavior of the agent of dispersal. This decreasing
5


correlation over distance is expected of plants that experience restricted gene flow
(Maki and Masuda 1994). Those species that use insects, birds, and other animals for
seed and pollen dispersal, are expected to exhibit little or no positive correlation,
especially across short distances (Loiselle et al. 1995).
As part of this research, fine scale spatial structure was examined in
populations of P. albicaulis and P. strobiformis in order to assess the ability of spatial
autocorrelation to discriminate dispersal signatures that are the direct result of seed
dispersal by scatter-hoarding corvids. Whereas P. albicaulis is known to be dispersed
by Clarks nutcracker and have the corresponding population genetic structure,
observations by Samano and Tomback (unpublished data) have revealed P.
strobiformis to be dispersed by Clarks nutcracker, at least in the northern part of the
species distribution, where it is expected to display the corresponding dispersal
signature.
Growth form will be examined for each of the populations to determine the
relative proportions of multi-stem and single-stem individuals. For P. albicaulis, a
high proportion of clumps is expected. Furthermore, it is expected that the majority
of tree clumps will be clusters, comprised of genetically distinct individuals
(Tomback et al. 1993). For P. strobiformis, a high incidence of tree clusters is also
expected.
For both pine species, it is expected that there will be a positive correlation
between individuals within a cluster (nearest neighbors), but with no overall spatial
6


autocorrelation among individuals comprising the population. This contrasts with
expectations for populations of wind-dispersed pines, which are expected to
demonstrate a positive low order (nearest neighbors) correlation and a zero to
negative correlation for the high order (distant neighbors).
7


CHAPTER 2
MATERIALS AND METHODS
Study Sites
Populations of P. albicaulis were sampled from three sites: Union Pass and
Island Lake, located in the Shoshone Forest in Wyoming, and Henderson Mountain,
located in the Gallatin National Forest in Montana (Bruederle et al. 1998) (Table 2.1).
Population samples comprised 20, 30, and 31 tree clumps and single individuals, and
a total of 44, 50, and 65 individual tree stems, respectively. Tree clumps contained
between one and six individual stems. A single P. strobiformis population was
sampled from Mount Withington, located in the San Mateo Mountains in the Cibola
National Forest in New Mexico (Table 2.1). This sample comprised 48 single-
stemmed individuals taken from 40 tree sampling sites containing either a single-stem
tree or multi-stem clump, with tree clumps containing between two and three
individual stems.
For each of the study sites, meter tapes were used to establish an x and y
coordinate system, which was used to determine the spatial relationships of trees
within a plot. The plots ranging from 20m by 20m (400m2) to 75m by 120m
(9,000m2). Foliage was sampled from each
8


Table 2.1 Locations and sample sizes of four populations of Pinus albicaulis Engelm.
(Pinaceae) and P. strobiformis Engelm. (Pinaceae) sampled for this study.
Study Site County State Latitude Longitude No. of stems No. of clumps
Henderson Mt. Park MT 43 N 110W 65 31
Island Lake Park WY 43 N 109 W 50 30
Union Pass Freemont WY 42 N 109 W 44 20
Mt. Withington Socorro NM 34 N 107 W 48 40
9


individual upright stem or trunk in an exhaustive sample of each plot. Samples were
packaged individually, kept on ice, and subsequently refrigerated for between two
and five days until protein extraction was performed. The precise location of each
individual stem was plotted on an (X, Y) coordinate system relative to the transacts
(Fig. 2.1 a-c, Fig. 2.2). Aspect and dbh (diameter at breast height) were subsequently
obtained for each individual. For cases in which the individual was not tall enough
for a dbh measurement (i.e., seedlings and saplings), actual height was recorded.
Sample Analysis
For each individual, foliage was ground with a mortar and pestle using liquid
nitrogen to facilitate the process. Soluble enzymatic proteins were then extracted
from the crushed sample using .16 M phosphate buffer pH 7.0, containing germanium
dioxide (0.107M), dimethyldithicarbonic acid (0.018M), polyvinylpyrrolidone
(molecular weight 40,000) (0.001M), sodium borate (0.088M), sodium metabisulfide
(0.017M), ascorbic acid (sodium salt) (0.222M), DMSO (8.8%), 2-phenoxyethanol
(0.60%), and P-mercaptoethanol (0.18%) (Mitton et al. 1979; Cheliak and Pitel, 1984;
Young et al. 1993). The supernatant containing the soluble proteins was absorbed
onto wicks cut from Whatman filter paper (No. 17); wicks were stored at -70C for
subsequent electrophoresis.
For the three P. albicaulis populations, 10.5% (weight to volume) starch gels
(Sigma Chemical Co.) were prepared using gel buffers for each of four gel-electrode
buffer systems. The first, a Tris-citrate gel-electrode buffer system pH 6.7/6.3
10


Fig. 2.1 a-c. Site map of all Pinus albicaulis Engelm. (Pinacceae) individuals sampled
from the: (a) Union Pass Wyoming population (b) Island Lake, Wyoming population,
and (c) Henderson Mountain, Montana population.
20 .
16 14 12 £ 10 - i ... ~

9
6 4 9
A
0 ^ = i
0 5 10 15 20 j
X (m) |
X (m)
11


Fig. 2.2 Site map of all Pinus strobiformis Engelm. (Pinaceae) individuals sampled
from the Mount Withington, New Mexico population.
12


(Selander et al. 1971), was run at constant current (60 mA) for eight hours.
Following electrophoresis and slicing, these were stained for 6-phosphogluconate
dehydrogenase (6-PGD), malate dehydrogenase (MDH), and glyceraldehyde 3-
phosphate dehydrogenase (G3PDH). A sodium borate gel-electrode buffer system,
pH 7 .5/7.6 (Poulik 1957), was run at constant current (45 mA) until the voltage
reached 300 volts (approximately two hours). At this time, the current was switched
to maintain a constant voltage of 300 volts for the remaining eight hours. These gels
were subsequently stained for glutamate dehydrogenase (GDH), peroxidase (PER),
phosphoglucomutase (PGM), shikimic acid dehydrogenase (SDH), and superoxide
dismutase (SOD). A lithium borate gel-electrode buffer system, pH 8.1/8.5 (Cheliak
and Pitel 1984) was run at constant current (50 mA) for six hours and stained for
PGM, aspartate aminotransferase (AAT), and triose phosphate isomerase (TPI). The
fourth system, a histidine-EDTA gel-electrode buffer system, pH 7.0 (Cheliak and
Pitel, 1984) was run at constant current (40 mA) for 12 hours. These gels were
stained for MDH, PGM, G3PDH, isocitrate dehydrogenase (IDH), and malic enzyme
(ME).
In addition to the aforementioned four gel-electrode buffer systems, three
additional morpholine citrate gel-electrode buffer systems, pH 6.1, 7.1, and 8.1 were
used fori5, strobiformis (Rogers et al. 1999). The morpholine citrate gel-electrode
buffer system, pH 6.1 was run at constant current (60 mA) for seven and a half hours
and stained for 6-PGD, MDH, IDH, GDH, and acid phosphatase (ACP). The
13


morpholine citrate gel-electrode buffer system, pH 7.1 was run at constant current (40
mA) for ten hours and stained for 6-PGD, MDH, SDH, IDH, ACP, aldolase (ALD),
glycerate dehydrogenase (GLY), and phosphoglucose isomerase (PG1). The third
morpholine citrate gel-electrode buffer system, pH 8.1, was run at constant current
(40 mA) for 12 hours and subsequently stained for 6-PGD, ALD, MDH, SDH, GLY,
and PGI.
Allozyme data were collected as individual genotypes for each of 19 putative
loci for the P. albicaulis populations and 20 putative loci for the P. strobiformis
population. The allele that migrated most rapidly though the gel was labeled a,
with slower alleles designated correspondingly by ascending alphabetical letters. In
instances where isozymes (multiple loci encoding a single enzyme system) were
detected, the most rapidly migrating isozyme locus was labeled 1, while the more
slowly migrating isozyme loci were assigned ascending numerical values.
Statistical Analysis
Allele frequencies for each of the putative loci were calculated using the
electrophoresis data and subsequently analyzed statistically. Statistical analyses were
performed using SAAP, a spatial autocorrelation analysis program (Wartenberg 1989)
and Relatedness 5.0 (Goodnight and Queller 1989). Microsoft Excel 98 was used for
database management and for summarizing statistics.
14


Using Relatedness 5.0, relatedness (r) between each pair of individuals over
all loci was ascertained as:
(2.1) r = Ix Ik E| (Py-P*)/ £x 2k 2, (Px-P*),
where, Py is the frequency of the allele at locus k for individual y, Px is the frequency
of the allele at locus k for individual x, and P* is the frequency of the allele for the
population (Queller and Goodnight 1989). The result is that relatedness is equal to
the sum of all alleles of all loci for all individuals using the allele frequency for the
population and the discontinuous allele frequency for the individual (i.e., 0.0, 0.5,
1.0).
This resulted in a half-matrix containing relatedness values for each pairing of
individuals. Pair-wise relatedness values were then compared with the Euclidean
distance between that same pair using Microsoft Excel 98. In addition to calculating
average relatedness for the entire population, relatedness values were calculated for
individuals within tree clusters, and for individuals between tree clusters, both as an
average and on a locus-by-locus basis.
For the spatial autocorrelation analysis, possible pairings of individuals, called
joins (Xie and Knowles 1991), were made based on Euclidean distances (Hossaert-
McKey et al. 1996) between each set of individuals, with the total number of joins
being calculated as n(n-l) (Epperson and Allard 1989).
15


Morans I is the actual correlation for an allele for a specific locus at a given
distance class in terms of similarity to other alleles in that same distance class.
Morans I-value was calculated as:
(2.2) I k= n Z jZj W (Yj- Y*) (Yr Y*)/ W Z, (Y,-Y*):
(Xie and Knowles 1991), where 1^ is Morans I for distance class k, Yj is the
frequency of the allele for individual i, Y* is the mean frequency of the allele for the
population, and W is the assigned weighting (Epperson 1990). Values range from 1
to +1, with a value of zero resulting from an absence of correlation, and large
negative and positive values having strongly negative and positive autocorrelation,
respectively (Sokal and Oden 1978a). This statistic, in particular, was used due to
the small number of individuals being sampled (Geburek 1993).
In order to develop Morans I values, distance classes were created for each of
the populations. Two methods of distance classification were employed, resulting in
fixed distance and variable distance (Leonardi and Menozzi 1996). For the fixed
distance method, each class comprised a set distance (e.g., 0-lm, l-2m, 2-3m, etc.).
The variable distance method involved dividing the individual pairings and placing an
even number of pairs in each distance class (e.g., 0-lm, l-3.5m, 3.5-8m, etc.).
For each join formed using a Euclidian distance, a distance class, either fixed
or variable, was assigned for use in determining Moran's I values. Due to the
presence of tree clusters, individuals within a cluster comprised the first distance class
for both the fixed and variable distance classification. The total number of joins for
16


the population determined the number of distance classes, with the determining factor
being retention of enough joins within the variable distance classification to allow for
statistical significance. For each diallelic locus, one allele was omitted from the
analysis to reduce statistical dependence on the data (Xie and Knowles 1991). Each
locus comprised of three or more alleles was analyzed using the most common allele,
with all remaining alleles being considered the same allele for this analysis. Each
individual was then given a numeric value for each locus, which was assigned based
upon the frequency of the populations most frequent allele at that locus. In this
analysis, homozygous individuals containing two copies of the most frequent allele
were assigned a numerical value of 1.0 for that locus. Homozygous individuals
containing zero copies of the most frequent allele were assigned a numerical value of
0.0, and heterozygous individuals containing one copy of the most frequent allele
were assigned a numerical value of 0.5 for that locus (Maki and Masuda 1994).
Once Morans I was calculated for each distance class, the values were
combined to create a correlelogram for each locus, which was used to give a visual
description of the spatial distribution for that given locus (Sokal and Oden 1978a).
Morans I-values for the correlelogram may be compared to an expected I-value
expressed as:
(2.3) E(I)= -(N-l)"',
where N is the number of individuals (Maki and Masuda 1994). The expected 1 value
will approximate zero for any population for which a substantial number of
17


individuals are included in the analysis. This is representative of spatial randomness.
The statistical significance of the I-values was determined by calculating the standard
normal deviate (SND) for each of the values, with the SND being equal to the
difference between the expected and actual genotypic pairs in each distance class.
This value was then standardized for the standard deviation of that distance class
(Leonardi and Menozzi 1996) and tested for significance using a Chi-square test
(Cliff and Ord 1973). All I-values and their corresponding statistical significance
were calculated using SAAP (Maki and Masuda 1994; Perry and Knowles 1991;
Knowles et al. 1992).
18


CHAPTER 3
RESULTS
Summary Data
In the three populations of P. albicaulis, 13 enzyme systems were encoded by
19 putative loci: AAT-1, AAT-2, ALD, GDH, G3PDH-1, G3PDH-2, IDH, MDH-1,
MDH-2, PER, PGD-1, PGD-2, PGM-1, PGM-2, SDH, SOD-1, SOD-2, TPI, and
UDP-2. Three additional loci (MDH-3, PER-2, and PER-3) stained inconsistently
and were eliminated from further analysis. Five loci were polymorphic in each of the
three populations: MDH-1, PER-1, PGD-1, SDH, and TPI (no criterion). Four
additional loci were found to be polymorphic in at least one population: AAT-1,
AAT-2, ALD, and PGD-2 (Table 3.1).
In the Union Pass population, nine of 11 tree clumps, each containing between
two and six stems, were found to contain genetically distinct individuals. In the
Island Lake and Henderson Mountain populations, eight of ten and 13 of 14 tree
clumps, respectively, were found to be comprised of genetically distinct individuals.
Of the five tree clumps within the three populations found to be comprised of
genetically identical individuals using these techniques, three comprised two stems,
one comprised three stems, and the other comprised four stems. All clumps
containing five or more trunks were found to comprise distinct individuals.
19


Table 3.1 Allele frequencies at 23 loci for the Union Pass (Wyoming), Island Lake
(Wyoming), and Henderson Mountain (Montana) populations of Pinus albicaulis
Engelm. (Pinaceae), and the Mount Withington (New Mexico) population of P.
strobiformis Engelm. (Pinaceae). Population numbers correspond to those in Table
2.1. See text for allozyme nomenclature.
Pinus albicaulis Pinus strobiformis
Locus allele Union Pass Island Lake Henderson Mt. Withington
AAT-1 a 0.000 NA 0.115 1.000
b 1.000 NA 0.885 0.000
AAT-2 a 0.116 0.010 0.000 0.000
b 0.884 0.990 1.000 1.000
ALD a 1.000 1.000 0.992 1.000
b 0.000 0.000 0.008 0.000
GDH b 1.000 1.000 1.000 1.000
G3PDH-1 a 1.000 1.000 1.000 1.000
G3PDH-2 a 1.000 1.000 1.000 1.000
IDH b 1.000 1.000 1.000 1.000
MDH-1 a 0.384 0.660 0.592 1.000
b 0.616 0.340 0.407 0.000
MDH-2 a 1.000 1.000 1.000 1.000
MDH-3 a NA NA NA 1.000
PER-1 a 0.000 0.000 0.008 0.000
b 0.407 0.360 0.415 0.802
c 0.593 0.640 0.577 0.198
PER-2 a NA NA NA 0.021
b NA NA NA 0.979
PGD-1 a 0.233 0.359 0.250 0.448
b 0.000 0.000 0.000 0.375
c 0.767 0.641 0.750 0.177
PGD-2 a 1.000 1.000 1.000 1.000
b 0.000 0.000 0.000 0.000
PGI-1 a NA NA NA 0.281
b NA NA NA 0.719
PGM-1 a 1.000 1.000 1.000 0.979
b 0.000 0.000 0.000 0.021
PGM-2 b 1.000 1.000 1.000 1.000
20


Table 3.1 (Cont.)
Locus allele Pinus albicaulis Pinus strobiformis
Union Pass Island Lake Henderson Mt. Withington
SDH-1 a 0.035 0.310 0.385 0.073
b 0.314 0.690 0.615 0.875
c 0.651 0.000 0.000 0.052
SDH-2 a NA NA NA 0.052
b NA NA NA 0.948
TPI a 0.023 0.100 0.038 0.000
b 0.977 0.900 0.962 1.000
SOD-1 a 1.000 1.000 NA NA
SOD-2 a 1.000 1.000 NA NA
UDP-2 a NA 1.000 NA NA
21


The maximum distances separating individuals in the Union Pass, Island Lake, and
Henderson Mountain populations were 24.93 m, 17.03 m, and 33.9 m, respectively.
For the Mount Withington population of P. strobiformis, 11 enzyme systems
were encoded by 20 putative loci: AAT-1, AAT-2, ALD, GDH, G3PDH-1, G3PDH-
2, IDH, MDH-1, MDH-2, MDH-3, PER-1, PER-2, PGD-1, PGD-2, PGI-1, PGM-1,
PGM-2, SDH-1, SDH-2, and TPI. Five additional enzyme systems were omitted
from further analysis due to weak activity and inconsistent banding: ACP, GLY, ME,
SOD, and UDP. Nine loci were found to be polymorphic within this population:
AAT-2, PER-1, PER-2, PGD-1, PGM-1, PGM-2, PGI-1, SDH-1, and SDH-2 (no
criterion).
The Mount Withington population contained seven tree clumps, with six of
the seven clumps comprising two stems, and the remaining clump comprising three
stems. Of seven clumps, four were found to be comprised of genetically distinct
individuals. All three clumps for which no genetic differentiation was found
comprised only two stems. The Mount Withington population was less dense than
the other populations sampled, with the maximum distance separating two individuals
being 114.86 m.
Relatedness
Average relatedness (r) was slightly negative for the Union Pass, Island Lake,
and Henderson Mountain populations of P. albicaulis (-0.0233, -0.0218, and -0.0165,
respectively), approximating zero. Two analyses were performed to examine the
22


relationship between distance and relatedness for the tree clumps. For the first
analysis, average relatedness was calculated for individuals within a clump and for all
individuals excepting those within a clump. In addition to the relatedness values that
were obtained for each polymorphic locus, an average r-value was calculated for the
population for each classification. Eighteen of the 19 polymorphic loci across the
three whitebark pine populations were found to display a higher relatedness within
tree clumps than between them. Only the PER-1 locus for the Island Lake population
of P. albicaulis had a lower r-value for individuals within clumps than between
clumps (-0.0613 and -0.0197, respectively). All other loci had significantly positive
r-values, when comparing individuals within clumps (ranging from 0.2840 to 1.0),
and slightly negative values, when comparing individuals between clumps (ranging
from -0.0156 to -0.3740). The average r-value for each of the populations across all
individuals was slightly negative (ranging from -0.0165 to -0.0233), although these
values were found to be slightly more positive than average relatedness when
comparing individuals between clumps (Table 3.2).
The second analysis, which incorporated relatedness, involved calculating a
pair-wise relatedness value for each possible pairing of individuals as a function of
the distance between the individuals comprising the pairing. Fixed and variable
distance classes were created for each population. The number of distance classes
created for each population was dependent upon the total number of pairings within
the population. All pairings of individuals within a tree clump were first grouped
23


Table 3.2 Summary of average relatedness among tree clumps and between
clumps for 19 loci for the Union Pass (Wyoming), Island Lake (Wyoming),
and Henderson Mountain (Montana) populations of Pinus albicaulis Engelm
(Pinaceae).
Site Locus Overall R Within clump R Between clump R
Union Pass -0.0233 0.6949 -0.0596
PER-I 0.6254 -0.0478
PGD-1 0.7708 -0.0565
SDH-1 0.6282 -0.0648
MDH-1 0.5102 -0.0458
TP1 1 -0.0798
AAT-2 0.8308 -0.0765
Island Lake -0.0218 0.4592 -0.0343
PER-1 -0.0613 -0.0197
PGD-1 0.3641 -0.0331
SDH-1 0.3815 -0.0290
MDH-1 0.5646 -0.0373
TP1 0.4697 -0.0340
AAT-2 1 -0.0464
Henderson Mt. -0.0165 0.4838 -0.0310
PER-I 0.3681 -0 0252
PGD-1 0.4667 -0.0285
SDH-I 0.2840 -0.0267
* MDH-1 0.6533 -0 0356
TPI 0.6977 -0.0414
AAT-1 0.5824 -0.0310
24


together within the first distance classification. For the variable distance
classification, each distance class was determined by the remainder of all possible
pairings being divided equally to ensure that each distance class would contain a
minimum 150 pairings of individuals. The number of distance classes used for the
variable distance classification was subsequently used in the development of the fixed
distance classification, with all non-within clump pairings being placed in the
distance class corresponding to their distance of separation between individuals. This
resulted in fewer than 150 pairings in some of the individual distance classes. The
distance classes that were created for this analysis were subsequently used for the
spatial autocorrelation that was performed. With each pairing given equal weighting,
average relatedness was calculated for each distance class for each of the populations.
Under both distance classification methods, relatedness among individuals within
clumps was higher than for any other distance class for each of the populations (Fig.
3.1 a-c, Fig. 3.2 a-c).
For the Mount Withington population of P. strobiformis, eight of nine
polymorphic loci also displayed higher relatedness among individuals within tree
clumps, than between them. The single exception, SDH-2, showed a considerably
lower w ithin clump relatedness than between clump relatedness (-0.4769 and -0.0143,
respectively) (Table 3.3). Average relatedness for the entire population was still
higher than that obtained among individuals between clumps. When examining
average relatedness for each distance classification, average relatedness within tree
25


Fig. 3.1 a-c Pairwise relatedness as a function of distance class for the: (a) Union
Pass, Wyoming (b) Island Lake, Wyoming, and (c) Henderson Mountain, Montana,
Pinus albicaulis Engelm. (Pinaceae) populations, using a fixed distance classification,
excepting distance class one, which consists of pairings occurring within tree clumps.
26


Fig. 3.2 a-c Pairwise relatedness as a function of distance class for the: (a) Union
Pass, Wyoming, (b) Island Lake, Wyoming, and (c) Henderson Mountain, Montana,
Pinus albicaulis Engelm. (Pinaceae) populations, using a variable distance
classification, excepting distance class one, which consists of pairings occurring
within tree clumps.
27


Table 3.3 Summary of average relatedness among tree clumps and between tree
clumps for six loci for the Mount Withington, New Mexico population of Pinus
strobiformis Engelm. (Pinaceae).
Site Locus Overall R Within clump R Between clump R
Mt. Withington -0.0213 0.4641 -0.0248
PER-1 0.5407 -0.0244
PGD-1 0.5152 -0.0266
PGM-I 1.00 -0.0300
SDH-2 -0.4769 -0.0143
PGI-1 0.2672 -0.0231
PER-2 1.00 -0.0256
28


clumps was found to be considerably higher than for all distance classifications
comprising non-clumped pairings of individuals (Fig. 3.3, Fig. 3.4)
Spatial Autocorrelation
Distance classes that were created for the relatedness analysis were
subsequently used for the spatial autocorrelation analysis. Seven, six and eight loci
were analyzed for the Union Pass, Island Lake, and Henderson Mountain populations
of P. albicaulis, respectively. For all three populations, MDH-1, PER, PGD-1, SDH,
and TPI were analyzed. For the Union Pass population, six of the seven loci (AAT-2
included) using both distance classification methods had statistically significant
positive values (p<.01) for the within clump pairings of individuals; only TPI did not
have a highly positive I-value. While many of the indiv idual I-values for a respective
distance class were found to have statistical significance, of those values, none had
more highly positive Morans I-values than the I-values for the within clump distance
classes. Each of the correlelograms, excepting TPI, also had statistically significant
correlelogram p-values (p<0.05) suggesting that the correlelogram may accurately
describe the spatial relationship for that locus. For the Island Lake population, four of
the six loci (MDH-1, PGM-2, SDH, and TPI) had statistically significant positive I-
values for the within clump pairings of individuals. The same four loci also had
statistically significant correlelogram p-values. The Henderson Mountain population
was found to have highly statistically significant positive I-values for the within
clump pairings of individuals for seven of the eight loci examined. All I-values for
29


Fig. 3.3 Pairwise relatedness as a function of distance class for the Mount
Withington, New Mexico, Pinus strobiformis Engelm. (Pinaceae) population, using a
fixed distance classification, excepting distance class one, which consists of pairings
occurring within tree clumps.
30


Fig. 3.4 Pairwise relatedness as a function of distance class for the Mount
Withington, New Mexico, Pinus strobiformis Engelm. (Pinaceae) population, using a
variable distance classification, excepting distance class one, which consists of
pairings occurring within tree clumps.
31


the within clump distance classes were more positive than any other I-values for any
of the other distance classes, with the exception of ALD. Excepting SDH with a
fixed distance class, the correlelogram p-values were statistically significant (p<0.05)
for all loci analyzed for the population (Fig. 3.5 a-c, Fig. 3.6 a-c).
Spatial autocorrelation was performed on six loci for the Mount Withington
population of P. strobiformis: PER-1, PER-2, PGD-1, PGI-1, PGM-1, and SDH-2.
However, only PGD-1 was found to have a highly statistically significant positive I-
value for the within clump distance class. The SDH-2 locus had a considerably
negative l-value (-0.33) for the within clump distance class, although this value was
still not considered significant, even at p<0.05. This negative number does, however,
correlate with the negative value obtained when examining r-values for within clump
pairings for SDH-2. Only PGD-1 and PGI-2 (with fixed distance classes) had
statistically significant correlelogram p-values (p<0.05) (Fig. 3.7, Fig. 3.8).
32


Fig. 3.5 a-c Morans I as a function of distance class for: (a) PGD-1, AAT-2, MDH-1,
TPI, SDH-1, and PER-1 for the Union Pass, Wyoming (b) PER-1, MDH-1, SDH-1,
PGD-1, and TPI for the Island Lake, Wyoming and (c) for PER-1, AAT-1, MDH-1,
SDH-1, TPI, ALD, and PGD-1 for the Henderson Mountain, Montana Pinus
albicaulis Engelm. (Pinaceae) populations using a fixed distance classification,
excepting class one, which consists of pairings occuring within tree clumps.
Distance class
-PGD-1
AAT-2
MDH-1
-TPI
-SDH-1
-PER-1
33


Fig. 3.6 a-c Morans I as a function of distance class for: PGD-1, AAT-2, MDH-1,
TPI, SDH-1, and PER-1 for the Union Pass Wyoming (b) PER-1, MDH-1, SDH-1,
PGD-1, and TPI for the Island Lake, Wyoming, and (c) PER-1, AAT-1, MDH-1,
SDH-1, TPI, ALD, and PGD-1 for the Henderson Mountain, Montana, Pinus
albicaulis Engelm. (Pinaceae) populations, using a variable distance classification,
excepting class one, which consists of pairings occurring within tree clumps.
1 00
i
(C)
Distance Class
-PER
AAT-1
MDH-1
SDH
TPI
ALD
PGD-1
PGM-2
-0 40
34


Fig. 3.7 Morans I as a function of distance class for PER-1, PGD-1, PGM-1, SDH-2,
PGI-1, and PER-2 for the Mount Withington, New Mexico, Pinus strobiformis
Engelm. (Pinaceae) population, using a fixed distance classification, excepting class
one, which consists of pairings occurring within tree clumps.
35


Fig. 3.8 Morans I as a function of distance class for PER-1, PGD-1, PGM-1, SDH-2,
PGI-1, and PER-2 for the Mount Withington, New Mexico, Pinus strobiformis
Engelm. (Pinaceae) population, using a variable distance classification, excepting
class one, which consists of pairings occurring within tree clumps.
36


CHAPTER 4
DISCUSSION
Populations of plants are expected to exhibit spatial genetic heterogeneity due,
in part, to variously restricted gene flow. The seeds of wind-dispersed species are
typically dispersed leptokurtotically, with the highest proportion falling in close
proximity to the source. As such, there is an expected decreasing correlation of
relatedness between individuals with increasing distance (Shapcott 1995), although
this is not always true due to other factors, such as inbreeding depression (Xie and
PCnowles 1991) and obligate outcrossing (Epperson and Allard 1989). In contrast,
dispersal by birds, is expected to result in a spatial pattern, or signature, that is
characteristic for the species.
The method of dispersal and the effect on family structure is particularly
evident within pine species. For example, within species of pines that are bird-
dispersed, populations often have a high proportion of multi-genet tree clusters
(Linhart and Tomback 1985; Tomback et al. 1993; Torrick et al. 1996). In contrast,
there has been little documented evidence of this family structure occurring within
wind-dispersed pines.
Difference in dispersal method results not only in differing spatial structures
in populations, but also plays a role in determining growth form in a population.
37


While wind-dispersed pine species are expected to consist primarily of single stem
individuals, with gradually decreasing correlation of relatedness over distance, this is
not anticipated for species exhibiting animal-dispersal of seeds. Species dispersed by
caching/scatter-hoarding animals or birds, in particular, are expected to have a
considerably higher percentage of multi-stem clumps within a population (Linhart
and Tomback 1985; Tomback et al. 1993; Torick et al. 1996), as well as multi-genet
clusters. This contrasts with the relatively low proportion of tree multi-stem clumps
occurring within wind-dispersed pine populations. While bird-dispersed pines
contain clusters that are multi-genet, it is expected that the individuals comprising the
tree clusters will be more closely related genetically with each other than those
individuals comprising different clusters. Furthermore, genetic differentiation among
individuals of different clusters provides the bulk of the genetic differentiation within
the population (Bruederle et al. 2001). Finally, it is anticipated that whereas
individuals comprising a tree cluster will show a high level of positive spatial
correlation, individuals in all other distance classes should show no to slightly
negative spatial correlation (Maki and Masuda 1994).
As has been previously demonstrated for P. albicaulis (Linhart and Tomback
1985; Tomback et al. 1993; Bruederle et al., 2001), tree clumps are common in
whitebark pine (37.5%, 33.3%, and 46.7% of individuals being multi-stemmed for the
Union Pass, Island Lake, and Henderson Mountain populations, respectively).
38


Additionally, these clumps comprise a large number of stems each (29, 30, and 49 for
the Union Pass, Island Lake, and Henderson Mountain populations, respectively),
representing a large percentage of the individual stems for these populations (65.9 %,
60.0%, and 75.4%, respectively) (Fig. 4.1). Furthermore, a large percentage (85.7%)
of multi-stemmed individuals are comprised of genetically distinct individuals (Fig.
4.2), and, as such, are tree clusters. When tree clumps comprised of only two stems
are removed from the analysis, the percentage of multi-stemmed individuals that are
multi-genet, and therefore clusters, increases to 92.0%; this supports previous
findings for the species (Linhart and Tomback 1985; Tomback et al. 1993; Torick et
al. 1996). Clearly the most common growth form in P. albicaulis is the multi-genet
tree cluster. Conversely, multi-stemmed individuals resulting from the branching of a
single genetic individual are not particularly common in subalpine whitebark pine
populations.
As seed caches buried by Clarks Nutcrackers often derive from a common
maternal tree (Tomback 1988; Tomback and Schuster 1994), the expected level of
genetic relatedness between individuals within these clusters should remain higher
than the overall genetic relatedness between individuals among different tree clusters
and that of the population as a whole. Spatial autocorrelation analysis for the three
populations of P. albicaulis revealed a dispersal signature strongly supporting this
hypothesis (Fig. 3.7). Of the 21 loci examined in the three populations, 17 (81%)
39


Fig. 4.1 Percentage of individuals comprising varying numbers of stems for the
Union Pass (Wyoming), Island Lake (Wyoming), and Henderson Mountain
(Montana) populations of Pinus albicaulis Engelm. (Pinaceae), and the Mount
Withington (New Mexico) population of P. strobiformis Engelm. (Pinaceae).
w 90%
Q.
| | 80%
! 70%
I
1 £
5 60%
| 50%
E
2 40%
g 30%
o 20%
8 10%
0)
a o%

iii .
XL
1 2 3 4 5 6
number of stems per "individual"
(Union Pass
Island Lake
Henderson j
Mt Withington j
I
I
40


Fig. 4.2 Proportion of multi-stemmed individuals that are multi-genet as a function
of number of stems present summarized over all sampled populations of Pinus
albicaulis Engelm. (Pinaceae) and P. strobiformis Engelm. (Pinaceae).
41


were found to have a highly significant (p< .01) positive correlation for individuals
within a cluster. For three of the four loci for which this was not the case, the p-value
for the associated correlelogram was not statistically significant (p< .05) and, as a
result, these loci were not reliable in providing an accurate description of the
population and were subsequently eliminated from further discussion. The only locus
that did not have a high within cluster value, while maintaining a significant
correllogram p-value, was the Henderson Mountain ALD locus (-0.02). This value,
which was slightly negative, is similar to the expected Morans I-value for a given
distance class, assuming random mating. This particular deviation from the expected
Morans I-value is likely due to the low level of genetic differentiation in the
population for that locus; the frequency of the most frequent allele was 0.992. This
low level of polymorphism resulted in insufficient differentiation to calculate
Morans I-values differing from expected (expected I-value ~ 0.0) and resulted in a
small range of I-values for the locus over all distance classes (-0.08 to 0.02).
In nutcracker-dispersed pines, within cluster relatedness should be high, while
the remaining distance classes would be expected to show little to no relatedness,
regardless of the physical distance between them. This was found to be the case
when analyzing the data from the three P. albicaulis populations. While there were
numerous I-values that were found to be significant for each of the correlograms,
there was no overall positive or negative trend with increasing distance.
42


While the statistical significance was not calculated for a regression of overall
relatedness versus distance, the analyses using Relatedness 5.0 (Goodnight and
Quellar 1989) seem to support the results from the spatial autocorrelation analysis.
Once again, for each of the three populations, overall relatedness (r) for individuals
within a cluster was found to be considerably higher than that between individuals in
different clusters. The average r-value for the within cluster pairs ranged between
0.2928 and 0.6990, while the highest average r-value for any of the distance classes
comparing between cluster pairings was 0.0128, with values ranging as low as -
0.3037. This analysis, in addition to the spatial autocorrelation analysis, reveals a
dispersal signature for P. albicaulis that meets expectations, and provides a
framework in which to examine and discuss spatial genetic heterogeneity in P.
strobiformis, as well as other pine species.
In the Mount Withington population of P. strobiformis, tree clumps comprised
only 17.5% of the population. In other words, there was one half to a third as many
clumps as was present in the populations of P. albicaulis. Furthermore, fewer than
half of the stems in the Mount Withington population (31.3%) comprised a tree
clump. Generally speaking, the tree clump growth form appears to be less frequent in
southwestern white pine from Mount Withington, San Mateo Mountains, and
comprises fewer stems. The proportion of tree clumps within the population may,
however, vary from population to population, due to geographical differences
43


throughout the range of the species. Only a single tree clump was comprised of three
trunks, the remainder comprising two trunks. Only 57.1% of the multi-stem clumps
were multi-genet and, therefore, tree clusters (50% of two-stemmed clumps and the
single three-stemmed tree clump); the low frequency of large numbers of stems
within each clump is likely a contributing factor, thereby restricting the utility of
these data. This is slightly different from the whitebark pine populations, in which
66% of the two-stemmed individuals were found to be multi-genet clusters.
Complicating factors in understanding the significance of these results include
markedly different stand and cluster densities; the southwestern white pine population
had a considerably lower stand density and frequency of multi-stem individuals in
comparison to the whitebark pine populations. These densities were a full power
lower at the Mount Withington P. strobiformis site (stand density of 0.006
individuals/m2 and tree clump density of 0.001 tree clumps/m2) than at the Island
Lake, Union Pass, and Henderson Mountain P. albicaulis sites (stand densities of
0.222, 0.110, and 0.072 individuals/m2, respectively, and tree clump densities of
0.044, 0.016, and 0.028 tree clumps/m2, respectively).
Of the six loci analyzed in the Mount Withington southwestern white pine
population using spatial autocorrelation, only one I-value was found to be statistically
significant for the distance class comprising within cluster pairs of individuals (0.94
for PGD-1). Furthermore, only PGD-1 and PER-1 (using the fixed distance method)
44


were found to have significant correlelogram p-values (p<0.05). These data do not
provide overwhelming support implicating dispersal by scatter-hoarding corvids or
other scatter-hoarding animals in southwestern white pine. However, when looking
at average r-values for the population, the pattern is similar to that for populations of
P. albicaulis, with the average within cluster r-value for southwestern white pine
being 0.4225, while the average r-value for the between cluster distance classes
ranging between 0.0883 and -0.3204. As such, these data support expectations for
seed dispersal involving scatter-hoarding corvids. The smaller sample size for the
tree clump form and decreased stand density may explain the inability to obtain
statistical significance for the within clump spatial autocorrelation analysis.
Other factors need to be considered when interpreting the results for P.
strobiformis. For example, while it is known that Clarks Nutcracker is responsible
for seed dispersal in whitebark pine (Benkman et al. 1984), it has only recently been
demonstrated for southwestern white pine from a single site in Colorado (S. Samano
and D.F. Tomback, unpublished data). Further uncertainty surrounds the dynamic
nature of dispersal method over the geographical range of southwestern white pine.
Clarks nutcrackers, the potential dispersers, are not syntopic with the southwestern
white pine in the southern portion of the pines distribution (Tomback 1988).
Furthermore, the dynamic behaviors of seed dispersers, differing habitats and
corresponding selection pressures, and differing role of potential seed predators
45


suggest a complex and dynamic relationship that undoubtedly influences fine scale
structure in this species throughout its range.
Another factor limiting the effectiveness of the statistical analyses,
specifically with regard to southwestern whitebark pine, is the relatively low
percentage of tree clumps within populations and the lower average number of trunks
per tree clump. The latter results in a reduced likelihood of identifying tree clusters
due to the probability of any two indiv iduals having different genotypes based on
random mating and allele frequencies at polymorphic loci. More importantly for this
analysis, the low proportion of tree clumps in a population results in fewer possible
pairings that are able to be used for the within cluster distance class, thereby
decreasing the resolution for identifying statistically significant data. This inability to
obtain statistically significant results occurs at a greatly higher level, with the fewer
tree cluster samples obtained. This paucity of within cluster pairings prevents an
accurate depiction of spatial relationships at the finest scale.
A final factor that should be considered when describing spatial genetic
heterogeneity in P. strobiformis is the difference in spatial relationship among
different enzyme systems. While some loci appeared to show a spatial relationship
similar to that found in P. albicaulis (PGD-1, PER-1), other loci do not meet
expectations for either a wind or animal dispersed species (PGM-1, SDH-2, PGI-1,
PER-2). Again, this is likely due to low sample sizes precluding an accurate
46


depiction of a biologically real phenomenon. For example, for three of the four
enzyme systems revealing no discernible dispersal signatures, the frequency of the
most common allele was extremely high (p>0.948), excepting PG1-1, which had
allele frequencies closer to that of PGD-1 and PER-1 (p= 0.719). Alternatively, other
evolutionary forces, such as selection, may be acting on these loci, effecting the
observed patterns.
Further investigations should take the aforementioned into consideration when
examining fine-scale spatial structure in southwestern white pine. Three steps should
be taken to ensure an accurate, quantitative description of fine scale spatial structure
in P. strobiformis. First, large sample sizes are necessary to reveal any statistical
significance, should it occur. While the data in the sampled P. strobiformis
population did not yield many statistically significant I-values for the within cluster
distance classes, the values themselves may be significant, given a larger sample size.
The larger sample size would, given an equivalent proportion of tree clumps, provide
enough within clump pairings to obtain statistical significance, if it existed.
Secondly, populations having a higher frequency of the multi-stem growth form
should be examined. While it is important to sample typical populations to
characterize this phenomenon, inclusion of a population comprising a large number of
multi-stemmed individuals would allow one to better assess the role, if any, of
scatter-hoarding corvids in determining population genetic structure in southwestern
47


white pine. Admittedly, this may infer little about the species as a whole. Finally, in
order to provide an accurate estimate of fine scale structure, and evidence for
dispersal signatures in P. strobiformis and other species, it is necessary to analyze
populations across the entire range of the species. Any reliable geographic variation
in dispersal signature would be a fascinating finding, with many implications
concerning ecology and natural selection.
48


LITERATURE CITED
BENKMAN, C.W., R.P. BALD A, AND C.C. SMITH. 1984. Adaptations for seed
dispersal and the compromises due to seed predation in limber pine. Ecology
65:632-642.
BRUEDERLE, L.P., D.L. ROGERS, K.V. KRUTOVSKII, AND D.V. POLITOV.
2001. Population Genetics and Evolutionary Implications. In D. F. Tomback,
S. F. Amo, and R. E. Keane (eds.) WhitebarkPine Communities: Ecology
and Restoration, 137-153. Island Press, Washington, D. C. USA.
-------, D.F. TOMBACK, K.K. KELLY, AND R.C. HARDWICK. 1998.
Population genetic structure in a bird-dispersed pine, Pinus albicaulis
(Pinaceae). Canadian Journal of Botany 76: 83-90.
CARSEY, K.S. AND D.F. TOMBACK. 1994. Growth form distribution and genetic
relationships in tree clusters of Pinus flexilis, a bird-dispersed pine.
Oceologia 98:402-411.
CHELIAK, W.M. AND J.A. PITEL. 1984. Techniques for starch gel electrophoresis
of enzymes from forest tree species. Canadian National Forestry Service Pi
X 42.
CLIFF, A.D. AND J.K. ORD. 1973. Spatial Autocorrelation. 175pp. London.
49


CRITCHFIELD, W.B. AND E.L. LITTLE. 1966. Geographic distribution of the
pines of the world. U.S. Department of Agriculture No. 1991.
DEWEY, S.E. AND J.S. HEYWOOD. 1988. Spatial genetic structure in a
population of Psychorta nervosa. I. Distribution of genotypes. Evolution 42:
834-838.
EPPERSON, B.K. 1990. Spatial autocorrelation of genotypes under directional
selection. Genetics 124:757-771.
-------AND R.W. ALLARD. 1989. Spatial autocorrelation analysis of the
distribution of genotypes within populations of lodgepole pine. Genetics 121:
369-377.
FURNIER, G.R., P. KNOWLES, M.A. CLYDE, AND B.P. DANCIK. 1987. Effects
of Avian seed dispersal on the genetic structure of whitebark pine populations.
Evolution 41:607-612.
GEBUREK, T. 1993. Are genes randomly distributed over space in mature
populations of sugar maple (Acer saccharum March.)? Annals of Botany 71:
217-222.
GOODNIGHT, K.F., AND D.C. QUELLER. Relatedness 5.0. The Keck Center for
Computational Biology, Rice University, Houston TX.
HOSSART-MCKEY, M., M. VALERO, D. MAGDA, M. JARRY, J. CUGUEN,
AND P. VERNET. 1996. The evolving genetic history of a population of
50


Lathyrus sylvestris: Evidence from temporal and spatial genetic structure.
Evolution 50: 1808-1821.
KNOWLES, P. 1984. Genetic variability among and within closely spaced
populations of lodgepole pine. Canadian Journal of Genetic Cytology 26:
177-184.
--------, D.J. PERRY, AND H.A. FOSTER. 1992. Spatial genetic structure in two
tamarack [Larix laricina (Du Roi) K. Koch] populations with differing
establishment histories. Evolution 46: 572-576.
LEONARDI, S. AND P. MENOZZI. 1996. Spatial structure of genetic variability in
natural stands of Fagus sylvatica L. (beech) in Italy. Heredity 77: 359-368.
LINHART, Y.B., J.B. MITTON, K.B. STURGEON, AND M.L. DAVIS. 1981.
Genetic variation in space and time in a population of ponderosa pine.
Heredity 46: 407-426.
--------AND D.F. TOMBACK. 1985. Seed dispersal by nutcrackers causes multi-
trunk growth form in pines. Oecologia 67:107-110.
LOISELLE, B.A., V.L. SORK, J. NASON, AND C. GRAHAM. 1995. Spatial
genetic structure of a tropical understory shrub, Psychotria officinalis
(Rubiaceae). American Journal of Botany 82: 1420-1425.
51


MAKI, M. AND M. MASUDA. 1994. Spatial genetic structure within two
populations of a self-incompatible perennial, Chionographis japonica var.
japonica (Liliaceae). Journal of Plant Research 107:283-287.
MITTON, J.B., Y.B. LINHART, J.L. HAMRICK, AND J.S. BECKMAN. 1979.
Observations of the genetic structure and mating system of ponderosa pine in
the Colorado front range. Theoretical Applications of Genetics 51: 5-13.
NEI, M. 1973. Analysis of gene diversity in subdivided populations. Proceedings of
the National Academy of Sciences USA 70:3321-3323.
PERRY, D.J. AND P. KNOWLES. 1991. Spatial genetic structure within three
sugar maple (Acer saccharum Marsh.) stands. Heredity 66:137-142.
POULIK, M.B. 1957. Starch gel electrophoresis in a discontinuous system of buffer.
Nature. 180: 1477-1479.
QUELLER, D.C. AND K.F. GOODNIGHT. 1989. Estimating relatedness using
genetic markers. Evolution 43: 258-275.
ROGERS, D.L., C. I. MILLAR, AND R. D. WESTRFALL. 1999. Fine-scale
genetic structure of whitebark pine (Pinus albicaulis): associations with
watershed and growth form. Evolution 53:74-90.
SELANDER, R.K., M.H. SMITH, S.Y. YANG, AND J.B. GENTRY. 1971.
Biochemical systematics in the genus Peromyscus. I. Variation in the old
field mouse. Stud. Genetics VI 7103:49-90.
52


SHAPCOTT, A. 1995. The spatial genetic structure in natural populations of the
Australian temperate rainforest tree Atherosperma moschatum (Labill.)
(Monimiaceae). Heredity 74: 28-38.
SOKAL, R.R. AND N.L. ODEN. 1978. Spatial autocorrelation in biology 1.
Methodology. Biological Journal of the Linnean Society 10: 199-228.
--------AND-----------. 1978. Spatial autocorrelation in biology 2. Some
biological implications and four applications of evolutionary and ecoligical
interest. Biological Journal of the Linnean Society 10: 229-249.
TOMBACK. D.F. 1982. Dispersal of whitebark pine sees by Clarks Nutcracker: a
mutualism hypothesis. Journal of Animal Ecology 51: 451-467.
--------. 1988. Nutcracker-pine mutualisms: multi-trunk trees and seed size. In
Ouellet, H. (ed.), Acta XJX Congressus Internationalis Ornithologici, Vol. 1.
Ottowa, Ontario: University of Ottowa Press, 518-527.
--------, F.K. HOLTMEIER, H. MATTES, K.S. CARSEY, AND M.L. POWELL.
1993. Tree clusters and growth form distribution in Pinus cembra, a bird-
dispersed pine. Arctic and Alpine Research Vol. 25, No. 4, 374-381.
--------AND Y.B. LINHART. 1990. The evolution of bird-dispersed pines.
Evolutionary Ecology 4: 185-219.
--------AND W.S.F. SCHUSTER. 1994. Genetic population structure and growth
form distribution in bird-dispersed pines. Proc.-lnternational Workshop on
53


Subalpine Stone Pines and Their Enviropnment U.S. Department of
Agriculture. July 1994.
TORICK, L.L., D.F. TOMBACK, AND R. ESPINOZA. 1996. Occurrence of multi-
genet tree clusters in wind-dispersed pines. American Midi. Nat. 136:262-
266.
WARTENBERG, D. 1989. SAAP------------- Spatial Autocorrelation Analysis Program.
University of Medicine and Dentistry of New Jersey, Piscataway, NJ.
WRIGHT, S. 1951. The genetical structure of populations. Ann. Eugenics 15:323-
354.
XIE, C.Y. AND P. KNOWLES. 1991. Spatial genetic structure within natural
populations of jack pine {Pinus banksiana). Canadian Journal of Botany
69:547-551.
YOUNG, A.G., S.J. WARWICKW, AND H.G. MERRIAMM. 1993. Genetic
variation and structure at three spatial scales for Acer saccharum (sugar
maple) in Canada and the implications for conservation. Canadian Journal of
Forest Resources 23:2568-2578.
54