p: msmi:

1'

an:

mam&w:u

t...

i

1%.
31

335...?
. SIQEXF

T1
1
. A
sham". mu
Ir :tF‘.“
a 5 , 3

“$31".

 

 

 

 

 

 

 

I.
2.. 4..
t 3 3.1!.

.w 1...!!!
...t!7..l.(
w.

THE-Sis

«7 [OD

 

MichiSlam State

 

J'Itliitliiiiiilll

LIBRARY

University

 

 

This is to certify that the

dissertation entitled

Double-stranded RNA mediated recovery of
American chestnut populations:
A demographic analysis

presented by

Anita L. Davelos

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Botany and Plant Pathology

2/

Date d6 $47 7? /

 

 

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

_—_.—n- -rE

 

4 -_ — _.______ MAH_,__._rfi._.
V ' v —v _

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

d§72233023u36

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 mm.w5-p.14

 

Ax!" r “' V

l: I- 5“

I!
u.‘U..L .1

 

 

DOUBLE-STRANDED RNA MEDIATED RECOVERY OF AMERICAN CHESTNUT
POPULATIONS: A DEMOGRAPHIC ANALYSIS

By

Anita L. Davelos

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Botany and Plant Pathology

1 999

 

 

 

.
'p‘ -o,
l\§..‘~.h
1'”; he'-
“4., vb“
.
\" . ..<
-h-.~“
:
l‘ A ‘
I“ ‘.
”"ni‘n
N." ~. .3-
t.£l )

5‘
3y—
e"..‘.~o.
“‘5 I
1
‘ e
I 1 ‘I
bk.“ \.
\

ABSTRACT
DOUBLE-STRANDED RNA MEDIATED RECOVERY OF AMERICAN
CHESTNUT POPULATIONS: A DEMOGRAPHIC ANALYSIS
By

Anita L. Davelos

Cryphonectria parasitica, the chestnut blight pathogen, can cause serious
reductions of the survival and reproduction in American chestnuts (Castanea dentata).
This pathogenic fungus can be infected with a cytoplasmic hyperparasite (a double-
stranded (ds) RNA) which debilitates the pathogen and reduces its virulence. In host
populations infected with hyperparasitized fungi, trees can respond to infection and
recover. However, no work has determined if recovery of individual trees translates into
recovery at the population level. The main objectives of this study are: ( l) evaluate the
effects of C. parasitica on trees of varying size; (2) determine how disease alters host
demographics; and (3) evaluate the extent of dsRNA mediated recovery in American
chestnut populations.

Inoculation and natural infection studies revealed that small individuals succumb
to infection regardless of the virulence of the inoculant. The presence of dsRNA appears
to delay death for medium to large sized branches.

Matrix projection models compared the finite rate of population increase (A) and
size distributions of healthy, recovering, and non-recovering populations of the American

chestnut. Disease reduced population growth rates of non-recovering. Further,

 

mam--

‘b_<:r‘

I - ‘IM

   

-.-. '

r, .
13w...

» ‘.‘.~
Hy‘.‘

 

V
'vn.
' .3
M"“&\
.
., ‘.,
L‘".I
p. .
“‘.:.<.
b. .
A ~“
' n
.a .
A
t“‘ ‘
<

retrogressions in size of large individuals are observed in diseased populations but not in
healthy ones resulting in an increased frequency of small to mid-sized trees in non-
recovering populations. Transition matrices for recovering populations contained
characteristics of both healthy and non-recovering populations, and population grth
rates tended to be slightly lower than growth rates found in healthy populations.

Sensitivity and elasticity analyses indicate that dsRNA should be introduced onto
1-10 cm dbh trees in non-recovering populations to have the largest impact on population
growth rates. The G/L/F (Growth/Longevity/Fecundity) elasticity ratio, used to detect
stressed populations in conservation biology, did not detect the effects of chestnut blight
epidemics. Since pathogen infections do not materially affect survivorship, epidemics
have little effect on the G/L/F portrayal of a population.

Studies on the effects of disease on seedling survival and growth revealed that the
disease status of the adult population generally affects emergence and survival of
seedlings with a trend for recovering populations to perform best. Disease status does not
influence final seedling size; rather, final seedlings size was influenced by population.

This study emphasizes the need to examine not only effects of infection on
individuals but also on populations. Matrix projection models are an effective tool for
predicting population growth rates and examining the relative contributions of different
life history stages to population growth rates. This information allows effective

management strategies to be developed.

To my father

ACKNOWLEDGMENTS

To my advisor, Andy Jarosz, I express my deepest gratitude for his
encouragement, guidance, good humor, and willingness to challenge me throughout the
course of this work. He has been a wonderful colleague and supportive fi'iend. I also
thank Andy for exposing me to some fine dining experiences, especially Dinghy’s and the
Cabbage Shed, and for not giving me the “one more plot look” more often.

I thank my other committee members, Dennis Fulbright, Richard Lenski, Bryan
Epperson, and Jim Hancock, for being generous with their time and constructive
comments. In particular, I wish to thank Dennis for being an excellent resource and for
his enthusiasm. Helen Alexander, Bill MacDonald, Michael Milgroom, and Tobin
Peever also provided guidance and support for this work.

For making my experience as a graduate student at Michigan State University an
extremely positive experience, I express my appreciation to the faculty, staff, and
students of the Department of Botany and Plant Pathology and the Ecology, Evolutionary
Biology, and Behavior Program.

This study could not have been completed without the help of many people in the
field. For dealing with often unpleasant conditions with a minimum of grumbling I thank
Andy Jarosz, Jennifer Schaupp, Dennis Fulbright, Richard Leschen, Ngoc Kieu, Emily
Lyons, Jim Bier, Mario Mandujano, Amanda Posto, Katy Kampf, Lissa Leege, Tim
Tibbetts, and many undergraduates. I also wish to acknowledge the property owners,

particulary the Rau and Millard families, for allowing me to conduct my research on their

 

‘ '
,3fi1 ’

I
nth“. a .
. t
-i'\“ ,

 

H ~n

k...

L.
p

a. 06.

”a X".

~~ H.

3; I
:k: If“.

land. I also thank Ed Lawler for conducting the greenhouse experiment. For their
hospitality, I thank Jim and Norma Rice.

This work was supported by the National Science Foundation, the Association for
Women in Science, Sigma Xi, the College of Natural Sciences, the Department of Botany
and Plant Pathology, and the Ecology, Evolutionary Biology, and Behavior Program.

I am grateful to my parents for their love, patience, and support. I especially
thank my father for inspiring in me a love of reading and curiosity about the world around
me. I thank my cousins, Pericles Georges, Margaret Sinclair, and Helen Georges, for
their love and encouragement and always providing me with a good meal.

Finally, I could not have completed this work without my many wonderful
friends. My gratitude goes to Steve Buehne, Lissa Leege, Richard Leschen, Natalie
Panshin, Anne Plovanich-Jones, Amanda Posto, Susan Simmons, and Anna Wiese for
their support and encouragement. My deepest affection and appreciation go to Erika
Barthelmess, Chris Haufler, Andy Jarosz, Emily Lyons, Judy Mongold, Paco Moore,
Scott Nunes, Amanda Rinehart, Jennifer Schaupp, Denise Searles, Sue Stoltzfus, Carmen

Storm, and Sara Taliaferro for their unconditional love, encouragement, and support.

vi

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... ix
LIST OF FIGURES ................................................................................... xi
CHAPTER 1
INTRODUCTION ..................................................................................... 1
CHAPTER 2
EFFECTS OF BRANCH SIZE AND PATHOGEN VIRULENCE ON CANKER
DEVELOPMENT AND BRANCH MORTALITY ............................................. 10
INTRODUCTION ......................................................................... 10
METHODS ................................................................................. 13
Study Site ........................................................................... 13
Cryphonectria parasitica Isolates ............................................... 14
Inoculation Experiment ............................................................ 15
Natural Infections .................................................................. 15
Statistical Analysis ................................................................ 16
RESULTS ................................................................................... 18
Inoculation Experiment ........................................................... 18
Natural Infections .................................................................. 19
DISCUSSION .............................................................................. 21
CHAPTER 3
EFFECTS OF CRYPHONECTRIA PARASITICA AND DOUBLE-STRANDED RNA
INFECTIONS ON AMERICAN CHESTNUT DEMOGRAPHY ........................... 38
INTRODUCTION ......................................................................... 38
METHODS ................................................................................. 41
Study System ...................................................................... 41
Study Sites .......................................................................... 42
Population Projection Matrices ................................................. 42
Statistical Analyses ............................................................... 48
RESULTS .................................................................................. 49
Patterns in Transition Matrices ................................................. 49
Population Growth Rates ........................................................ 50
Size Distributions .................................................................. 52
Cross Sectional Area .............................................................. 53
DISCUSSION .............................................................................. 53

vii

 

 

 

~—
03.,
AL“

C. . v
ELLJL.

1'? «~—
Cn4. \
AIL“ ,

CHAPTER 4
DOUBLE-STRANDED RNA MEDIATED RECOVERY OF THE AMERICAN

CHESTNUT: WHAT CAN WE LEARN FROM HOST DEMOGRAPHY? .............. 80
INTRODUCTION ......................................................................... 8O
METHODS ................................................................................. 82

Study System and Study Sites ................................................... 82
Population Projection Matrices ................................................. 82
RESULTS ................................................................................... 84
Sensitivities ........................................................................ 84
Elasticities .......................................................................... 85
DISCUSSION .............................................................................. 88

CHAPTER 5

SEEDLING GROWTH AND SURVIVAL TN POPULATIONS OF AMERICAN

CHESTNUT THAT DIFF ER IN DISEASE STATUS ........................................ 122

INTRODUCTION ....................................................................... 122
METHODS ............................................................................... 124
Study System and Study Sites ................................................. 124
Natural Recruitment ............................................................ 124

Field Experiments ............................................................... 125
Greenhouse Experiment ......................................................... 125
Statistical Analyses .............................................................. 126
RESULTS ................................................................................. 126
Natural Recruitment ............................................................. 126
Greenhouse Experiment ......................................................... 128

Field Experiments ............................................................... 128
DISCUSSION ............................................................................. 129

CHAPTER 6

CONCLUSIONS ................................................................................... l 37

LITERATURE CITED ............................................................................. 142

viii

 

 

‘ c u q

.C .C .3 are x J .3. . Em . t .

.. . ...... u: ... .7. L L in. .: ...

w.» .5. ,J «L. 2. 2.. 2k .4 .C 3
v 1 T. T. T. v r TI >1 T1

 

LIST OF TABLES

Table 2-1: Profile analysis of canker area for the inoculation experiment ................... 26
Table 2-2:Profile analysis of canker width/branch circumference for the inoculation

experiment ............................................................................... 26
Table 2-3: Branch survival for the inoculation experiment ..................................... 27
Table 2-4: Canker morphology (non-callused or callused) for presence/absence of

Table 2-5:

Table 2-6:

Table 2-7:

Table 3-1:

Table 3-2:

Table 3-3:

Table 3-4:

Table 3-5:

Table 3-6:

Table 4-1:

Table 4-2:

Table 4-3:

dsRNA and branch size class for the inoculation experiment at County Line.28

Average branch diameter at the wound site in the natural infection study .....30
Survival, canker morphology, and average diameter of branch below canker
for natural infection study .............................................................. 31
Relationship between canker morphology, branch size, and survival for

natural infection study ................................................................. 32
Stage classes used to describe populations of American chestnut ............... 59
Transition matrices for six American chestnut populations ....................... 60
Finite rate of increase for American chestnut populations ........................ 67

Repeated measures analysis of finite rate of population increase for American
chestnut populations .................................................................... 68

Comparisons of observed population structures and calculated stable stage
distributions for six populations of American chestnut ............................. 69

Repeated measures analysis of cross sectional area at breast height for
American chestnut populations ....................................................... 7O

Sensitivity matrices for six American chestnut populations for two census
periods ................................................................................... 93

Elasticity matrices for six American chestnut populations for two census
periods .................................................................................. 106

Sums of elasticities within the G (growth), L (survival), and F (fecundity)
regions of transition matrices for six populations of American chestnut. . ....1 19

ix

 

‘ I

T

A“\.i\

.
u‘ .
3,5

.-4

Via

.LEI

Table 44: Sum of the elasticities for retrogressions for six American chestnut
populations for each census period ................................................. 120

Table 5-1: Proportion of seedlings or other small individuals less than 1 m in height,
which are clearly not seedlings but are similar in size due to herbivory or
infection by chestnut blight, from natural recruitment plots surviving in six
American chestnut populations ..................................................... 132

Table 5-2: Proportion of seedlings or other small individuals from natural recruitment
plots surviving in six American chestnut populations ........................... 133

Table 5-3: Mean final height of naturally recruited first year seedlings for six populations
of American chestnut .................................................................. 134

Table 5-4: Mean seed mass, proportion of seedlings emerging, and mean final height in a
greenhouse experiment for six populations of American chestnut ............. 135

Table 5-5: Proportion of seedlings emerging, within season survival, and mean final
height in field experiments for six populations of American chestnut 1 36

.-...'.
tin-5
»

 

b'V‘w;
J Aid-L
.

 

." "Ta
\mg

'b« y...,_

s‘.
~ 5

.,1

Figure 2-1:

Figure 2-2:

Figure 2-3:

Figure 2-4:

Figure 2-5:

Figure 3-1:

Figure 3-2:

Figure 3-3:

Figure 3-4:

Figure 3-5:

Figure 4-1:

LIST OF FIGURES

Mean canker area (cmz) i standard errors for branches inoculated with
isolates with and without dsRN A versus days since inoculation for the
inoculation experiment at County Line ............................................... 33

Mean canker area (cmz) i standard error for each branch size class versus
days since inoculation for the inoculation experiment at County Line... . ......34

Mean canker width/branch circumference i standard errors for each branch
size class versus days since inoculation for the inoculation experiment at
County Line ............................................................................. 35

Mean canker width/branch circumference i standard errors for branches
inoculated with isolates with and without dsRNA versus days since

inoculation for the inoculation experiment at County Line ...................... 36
Proportion of branches surviving during the inoculation experiment. . . . ....37
Map of American chestnut populations used in this study ........................ 71

Life-cycle for a recovering population of American chestnut and its
correspondence with the basic population projection matrix (A) ............... 72

Population structures observed in 1997 and 1998 and calculated stable

stage distributions for the census periods 1996-1997 and 1997-1998 for (a)

a healthy population, (b) a recovering population, and (c) a non-recovering
population ............................................................................... 75

Mean cross sectional area (cmz) _+. standard error of American chestnut
trees in each population type (healing, recovering, non-recovering) versus

census date ............................................................................. 79

Hypothetical time course of population growth rates for American
chestnut populations ................................................................... 80

Plot of G/L/F elasticities for six populations of American chestnut. . . .. .. . l 21

xi

’3‘; "’i
p. \'udLu(;!n‘ 0

 

Pris: 193,5.

I996 V:

«.711

CHAPTER 1

INTRODUCTION

Pathogens have the potential to influence plant community diversity and plant
species distributions through their effects on plant fitness and the outcome of intra- and
inter-specific competition (Burdon 1982; Burdon 1987). Both survivorship and
reproduction, the components of host fitness, can be reduced by pathogens. For
example, infection with a rust fungus decreased survival of groundsel both over winter
and during summer (Paul & Ayres 1986a; Paul & Ayres 1987). Other studies of natural
systems have demonstrated pathogen mediated decreases in host survival for both
herbaceous plants (Alexander & Burdon 1984; Parker 1986; Wennstrom & Ericson 1990;
Jarosz & Burdon 1992; Roy & Bierzychudek 1993; Thrall & Jarosz 1994) and trees
(Geils & Jacobi 1993; Burdon et a1. 1994). Disease has also been shown to decrease
reproduction, whether measured as flower number, fruit or cone production, seed
production, or seed biomass (Schaffer et a1. 1983; Alexander & Burdon 1984; Clay 1984;
Parker 1986; Paul & Ayres 1986b; Parker 1987; Paul & Ayres 1987; Wennstrom &
Ericson 1991; Jarosz & Burdon 1992; Roy & Bierzychudek 1993; Garcia-Guzman et a1.
1996; Marr 1997). The genetic composition of plant populations can also be altered by
pathogens through reduction of susceptible genotypes and increase of resistant genotypes
(Burdon et al. 1981; Carlsson et al. 1990; but see Parker 1991) and potentially, the
increased fitness of rare genotypes (Antonovics & Ellstrand 1984).

Many studies on the effects of disease on plant community structure have focused

on introduced pathogens in forest systems. One well known example is the change in

   
   

\
,_.u-

“WWW"
Hm!

t'9r“. W" .s
u.\\\.su Ur

 

(rag
. 5'".
c, \..A

3... ‘V‘O... A“
sick I." H

. ., -__ ,
QMC‘LK‘ 2' 1.1

A -

‘3’“). - -..
‘55:“‘3: &‘: \

5

‘3'- -'.p

composition of eastern deciduous forests in the US. after the introduction of chestnut
blight (Day & Monk 1974). Dramatic changes in species abundances, including the
devastation of dominant species, were also seen as the result of Phytophthora cinnamomi
infection of the eucalyptus forests of Australia (Weste 1980; Weste 1981; Dickman
1992). Community structure can also be altered by endemic pathogens. Elimination of
eucalyptus from forest areas infected with Armillaria root rot (Kile 1983) and the
increase in abundance of rare species in hemlock and Douglas fir forests in infection
centers caused by an unspecialized root rot fungus (Holah et a1. 1993) are examples of
indigenous pathogens changing plant community composition.

With the wide attention given to the effects of disease on individuals and plant
communities, it is surprising that little work exists addressing the effects of infection on
host population size and persistence. However, there is a small body of work which
suggests pathogens can have adverse effects on host populations. For example, there is a
negative relationship between population size in subsequent years and proportion of
diseased individuals and large population size in previous years in the Silenc-
Microbotryum interaction (Antonovics et a1. 1994). While these results suggest a
negative effect of infection on host population growth rates, the intrinsic rate of increase
for both disease-free and diseased populations was not explicitly measured. The potential
difference in growth rate for diseased and healthy populations has important implications
for the persistence of individual host populations in areas where disease is common.

The effects of a pathogen on host population and community structure is not
easily quantified because pathogens tend to affect some life history stages more than

others. For example, damping-off disease results in up to 74% mortality of seedlings in

.
556.1 2113’."

.
i. {I ,.
-I’IL‘WF. .{su‘

. g '
7"A“.a4 , .3"
uuhbssh as:—

 

a "Jv- -' fl.
3.x ‘

A-L..&»\‘

. .
"Iu‘d" 4 we .u,
“khan-\- H

“‘1

I
~et3. 31". ’3\:‘ ._

:I'QFMIAI. in r;
'2‘» n: - .
«in»: 01 a

'4‘“. v: _

. $4.3! 11::

an." ,

m3: , “I
315?“:

“UK
1.1;“ 1v
«‘5 R.
1.1411‘ _~‘
.“

three tree species while older individuals are virtually immune (Augspurger 1984). In
contrast, the anther-smut pathogen, Microbotryum violaceum, can infect both seedlings
and adults (Alexander & Antonovics 1988; Alexander 1989) but appears to only have
severe fitness consequences for adults by sterilizing infected plants. Further, size of an
individual rather than life history stage may influence the impact of disease. In the
Linum-Melampsora host-pathogen system, larger individuals are more likely to become
infected (Jarosz & Burdon 1992). However, small individuals of Podophyllum peltatum
experienced higher disease incidence and severity, and mortality than large plants when
infected with the pathogen, Puccinia podophylii (Parker 1988). How these differential
effects of disease on various life history stages alter host population growth rates has not
been investigated. High seedling mortality due to disease might have a small effect on
population growth rates if seedlings succumb to other factors, such as competition, in the
absence of disease. In contrast, a simple reduction in seed reproduction, not even
mortality, of a few large individuals could severely decrease population growth and
persistence.

Matrix projection models have been used to examine how perturbations to
particular life history stages might affect population size for endangered or rare plant and
animal species (e.g. Crouse et a1. 1987; Lande 1988; Menges 1990; Kephart & Paladino
1997). These methods should also be useful in examining the influence of disease on
host population growth rate and structure. Sensitivity analyses, which calculate the effect
of small perturbations in a projection matrix element on population growth, have been
particularly useful in guiding conservation efforts. Perhaps the best example of this

method's utility is seen in conservation efforts in the loggerhead sea turtle, Caretta

r‘

"V ' ’
.. .( .
turicn ‘ ‘

 

M$ۤ3 O. s:

v, “‘39,!“ 'F
“Emu—.55 ....

Liar“; ....

JOE—“BE.-. .-
-

. v ,
9m“), ~'- s
u.._.\. Lu‘ua\
.

Mgr-g- v —...
{ACNE “f...

‘I s «0 “~v"9
A \- ‘hh..\

‘

"u i
'r 1.1, 3-
C“ \\-~- i.

vat}-
' “‘1 RT’.)
rd: 3" D\' '
“~- p“-
; t-i-t
nu-

carettu (Crouse et a1. 1987). Early conservation efforts were targeted at increasing the
success of egg clutches that were laid on beaches. However, sensitivity analyses
indicated that this did little to increase sea turtle populations because the survival of
hatchlings after their release into the sea was exceedingly low. Sensitivity analyses
further indicated that slight increases in the survivorship of older juvenile and adult age
classes would have a much greater effect on increasing population size. This led to the
use of turtle exclusion fishing nets as a primary mode of conservation of this species
(Crowder et a1. 1994; Crowder et a1. 1995). Further, Silvertown et a1. (1993, 1996)
suggested using elasticity values, which measure each matrix element's proportional
contribution to a population's growth rate, to determine if a population is at risk. They
proposed that species have characteristic elasticity ratios with regard to growth (G),
survival (L) and fecundity (F). The G/L/F ratio is determined largely by life history, and
as a consequence annuals, herbaceous perennials and trees have characteristic G/L/F
ratios. Stressed populations will have G/I/F ratios that deviate significantly from the
ratios exhibited by species or populations having a similar life history (Silvertown et al.
1996)

I propose to extend the use of matrix projection analyses to evaluate the effects of
pathogens on host populations. At one level, matrix projection models can determine the
extent to which pathogens actually influence the growth and size of host populations. In
this context, it would be possible to answer the question of whether high pathogen
pressure (i.e., high levels of disease incidence and severity) influences the size or density
of plant populations. The matrix projection method will be particularly useful in

situations where pathogen pressure varies (either naturally or by manipulation) over space

  
  

viii“. '3‘.
L1 u“! .w~~

compass: ~

act‘s-3.33- "U
yA-nns ‘Lhu in“
l '

 

I
\
- A“.

van”, "up; .p
fi--.-I‘-4‘_ u
_

W’J‘Ia; 9
no. i‘.- .l\

29“: a":
‘ u".

":~ \u\st.

‘
--.
“,i\ ":~.
‘~“ u..._"‘_‘

-
,..,.

1
"E”

d. x p'Z'L

“In.“

"‘E'v 11“ .

y.‘LrC‘;

or time, allowing direct comparisons between populations. Among population
comparisons can be used to estimate the decrease in host population growth due to
disease, and it can also be used to identify growth stages that are most affected by
pathogen infection.

Matrix projections may also be used for the development of disease management
programs in a manner similar to their current use in developing species recovery
strategies for rare and endangered species. A number of introduced pathogens have had
large effects on some species in North America, Europe and other continents. Among the
most dramatic in their effects are introduced pathogens that infect tree species, which are
common within our native forests and urban landscapes. In North America, introduced
pathogens such as Cryphonectria parasitica on American chestnut (Roane et al. 1986),
Ophiostoma ulmi on elms (Brasier 1990), and Discula destructiva on native dogwoods
(Hibben 1990, Daughtrey & Hibben 1994) have had large effects on the structure of both
the host species population and the community. There is interest in controlling these
diseases using biological agents, the most likely being double-stranded RNA (hereafter,
dsRNA) elements that infect many fungal pathogens (Nuss & Koltin 1990). DsRN A
hyperparasites have been found in C. parasitica (Biraghi 1950a,b; Day et a1. 1977 ), 0.
ulmi (Brasier 1990) and D. destructiva (Yao & Tainter 1996, Yao et a1. 1997); Brasier
(1998) has proposed that they be introduced specifically for the purpose of controlling
these pathogens. However, the natural spread of these hyperparasites and subsequent
biocontrol have met with varying degrees of success.

Ophiostoma ulmi, the causal agent of Dutch elm disease, was introduced into

North America and Europe. Many elms survived the spread of this pathogen. A new

  
 

. l
m um

i h o
1 III "
1'? I“ '1
5",.‘Lurn ..
1 §

"fi‘ng “-9,; -
5‘1- L4.- 55

””3.

LP
‘1'

a.:':~~ «9
t Li

‘Y' A

i” 5 3:111;

more virulent form of the pathogenic fungus, 0. novo-ulmi, has evolved and appears to be
displacing the original less virulent form and having a much bigger impact on the
survival of elms (Brasier 1987). Reduced virulence in both species is reported to be the
result of infection of the pathogen by cytoplasmic intracellular hyperparasites (Brasier
1990). These hyperparasites (d-factors) are not widespread in either species and fungal
strains infected with d-factors are replaced by hyperparasite-free 0. nova-ulmi (Brasier
1990).

A second example of an introduced pathogen with the potential to be controlled
by a dsRNA hyperparasite is the focus of this dissertation. The pathogenic fungus
causing chestnut blight, Cryphonectria parasitica, was introduced accidentally into the
US. around 1904 in New York (Roane et al. 1986) and spread rapidly throughout the
range of the American chestnut, Castanea dentata, devastating populations. Infected
branches and trunks are girdled and killed by the fungus. A severe C. parasitica
epidemic occurred in Europe around 1938 (Roane et a1. 1986). Subsequently, dsRNA
hyperparasites spread throughout the pathogen populations in Europe (Biraghi 1950a,b).
Infection by the dsRNA hyperparasite reduced pathogen virulence (also termed
hypovirulence) (Jaynes & Elliston 1982), producing recovery of chestnut populations in
Europe (Heiniger & Rigling 1994). Hyperparasite infection debilitates C. parasitica and
reduces pathogen virulence on chestnut trees by reducing canker growth rate
(Anagnostakis & Waggoner 1981), asexual conidia production (Elliston 1985), and
sexual reproduction (Anagnostakis 1984, 1988). Reduced canker expansion rates will, in

some cases, allow the tree to produce enough wound callus tissue to wall of the fungus,

 

51.;.,‘ If

Kiwi; -“ '

fl J‘Wv-r’ v,:
53534.“: as y

 

 

 

v-‘w- .
.1911 ,2
‘1-8: k~s ‘ g

2r; -‘ ..
«m It: pm

‘93:. r. a. 1

7"! a 4' '
Elk-Li‘s“ n!” v-

0

es?
A.‘¥¥ g
L of‘
1 rs
\
'L‘ .

resulting in a superficial, non-lethal canker (healing canker). Trees producing non-lethal
cankers are said to be recovering.

Before the spread of the blight, chestnuts were a dominant overstory species in
eastern US forests (Roane et a1. 1986). Today, although few trees reach reproductive
status and populations are maintained by sprouts from the rootstock (Paillet 1984, 1988),
chestnuts are a dominant understory component of plant communities within their natural
range (Keever 1953, Russell 1987). Michigan is outside the natural range of the
American chestnut. However, chestnut populations have become established through
naturalization of planted trees (Brewer 1995). Trees began dying of blight in Michigan in
the latel9205 (Baxter & Strong 1931). In the late 19508, healing cankers began to appear
and the presence of dsRNA in isolates of the fungus were discovered in the late 19705
(Day et a1. 1977). Michigan appears to be the only area in the US. where dsRNA
induced hypovirulence has spread naturally (Fulbright et a1. 1983). For this reason,
Michigan provides an unique opportunity to study this interaction because of the presence
of recovering populations, non-recovering populations, and uninfected reproducing
populations of the American chestnut (Brewer 1995). Matrix projection models will
allow demographic parameters (population growth rate, sensitivity, and elasticity) of
these different population types to be compared.

Not only the influence of C. parasitica infection but also that of the potential
biocontrol of C. parasitica by dsRNA on American chestnut populations can be
evaluated. This matrix model approach may have general application to examining the
effects of decreases in pathogen virulence. Indeed, there is a strong theoretical basis for

this idea. The observation that pathogens with a large negative effect on plant fitness

l' ‘f “‘I‘.”

am .>

‘WL‘.\1.-u\»
u

aflf 3: "Q \
‘;..L ‘ u . A. .
. b

4.2. .,,.'...,.
“3 Aike‘b's“

"f‘! ' 1L“ 0
”\ak'tt} u \ l

.5-

 

v
‘1!" 7|” 9". ‘1'
p'itsni".1:5,, I

if} 13:; u g

”If“; ' ‘
“~55. Tn: ‘r

o ‘ . t
'8‘ o -' ."
than: k n.)-

1»:
A s,‘

. ra<.z‘ .
L. ‘~\“J‘e\:

i 33‘1’99. _
._ L-\‘

(high virulence) could destroy their own resource led researchers to postulate that
pathogens should evolve reduced virulence (Harper 1977). Implicit in this argument is
that reductions in virulence will lead to recovery of the plant population, which in turn
restores the resource base for the pathogen. There is empirical evidence for this
phenomenon from the interaction of Myxoma virus and rabbits. A virulent strain of
Myxoma was introduced into both Australia and Great Britain to control rabbit
populations (Fenner 1983). Rabbit populations were reduced initially but increased with
time. The loss of virus effectiveness in controlling rabbit populations was found to be the
result of the evolution of decreased virulence in the virus along with increased resistance
and immunity in the rabbits (Fenner & Ratcliffe 1965; Anderson & May 1982; May &
Anderson 1983).

However, there does not appear to be any evolutionary tendency for the evolution
of reduced virulence in natural plant-pathogen interactions (Jarosz & Davelos 1995).
Theoreticians have also pointed out that reduced virulence will not usually be favored by
individual selection, and may only be favored in situations where group selection is
important (May & Anderson 1990). Models incorporating ecological factors, such as
host density, predict evolutionary trajectories towards intermediate levels of pathogen
virulence (Lenski & May 1994). These models also found that reductions in virulence
need not result in appreciable increases in host population density.

In this dissertation, the impact of dsRNA mediated reductions in pathogen
virulence on the biology of American chestnuts will be investigated using matrix

projection analyses. Population growth rate, sensitivity, and elasticity (demographic

“27"“

Ng§xhr .

(:‘ZVA‘ions .

w-Q

 

' La;
9
1 11. -IC. :J'
§

“l
\* ‘ .zv- ~..
~1.'¥A .3 “T e

fi'yn- ~.‘, ~, .
t‘*r““'-1‘JI., I

' |
3“" ‘3,
Fir
'H. .'.
‘ >5A.

cf. ]_ _
“heifer

parameters derived from matrix projection models) will be used to evaluate the effects of
C. parasitica and dsRNA on chestnut populations.

The specific questions addressed are:

1. Does the pathogen have differential effects on individuals of different size?

2. How does disease affect chestnut demographics?

3. Do dsRNA epidemics result in full recovery of chestnut populations?
A fmther general objective is to determine the application of the G/L/F elasticity ratio of
Silvertown et a1. (1993, 1996) to the evaluation of disease processes. The interaction of
branch size and reductions in pathogen virulence, caused by dsRN A infection, on branch
survival will be examined in Chapter 2. The result of spread of dsRN A at the host
population level will be the focus of Chapter 3. Whether chestnut populations infected
with hyperparasitized forms of the pathogen have the same finite rate of increase as
disease-free populations will be examined. The implications of the demographic study
for management and conservation of the American chestnut is discussed in Chapter 4.
Greenhouse and field experiments examining seedling survival and growth are present in

Chapter 5. Chapter 6 contains conclusions and future directions.

[Til

q'm

“Rs“

3% -
\..‘

 

I"- .
I“ - .
'9. -;

\.‘\C
I, ,
71.11;»
'k‘

 

CHAPTER 2
EFFECTS OF BRANCH SIZE AND PATHOGEN VIRULENCE ON CANKER

DEVELOPMENT AND BRANCH MORTALITY

INTRODUCTION

Although most models of reductions in pathogen virulence, measured as a
negative effect on host fitness, focus on evolutionary changes in the parasite (e.g. Lenski
& May 1994), changes in virulence can also result from infection of a pathogen by a
hyperparasite. The relative importance of evolutionary versus ecological reductions in
virulence has not been explored empirically and they may result in very different
numerical and genetic dynamics of both host and pathogen populations. Hyperparasites
of fungal pathogens of plants are found commonly (Hollings 1982). Other groups of
organisms can also be adversely affected by intracellular elements. For example, a
plasmid which confers antibiotic resistance reduced growth of its bacterial host in the
absence of antibiotic (Lenski & Bouma 1987).

For cytoplasmic hyperparasites to be useful in biological control, we must not
only understand how they alter virulence but also how they spread within a pathogen
population. Many theoretical models have explored the relationship between pathogen
virulence and transmission between hosts (Anderson & May 1982, Ewald 1983, May &
Anderson 1983, Frank 1996) with increased virulence related to increased transmission;
hyperparasites may show similar dynamics. An empirical example with a parasite

(bacteriophage) and its bacterial host demonstrated that without horizontal spread (high

10

I ' '

._.' 3-..} -

. Is 3'- 6.....s \
b

5'- :JY‘ ‘J
ud»nb\. 1

an.” ~. «'3'0V‘
he .L)..\~bas-. ‘
a

I
-,,. _ ‘
1.234: A: K. :T. 3..
I ‘9 . . ‘ ' t
{rt-41.5% \:\;"\

-u. A“
‘1‘ "fl‘ ‘I Q
.’ ‘ Le t'

._ I
E-L‘ “a

x HA¢:.'} -
‘.

:I‘IAY '

A !

‘- ‘9‘ 9! (:1

DO: beet.

 

transmzssion) the bacteriophage evolved toward a mutualistic relationship with its host
(Bull et al.1991).

Double-stranded RNA (dsRNA) hyperparasites have been examined as potential
biological controls of plant pathogenic fimgi through their ability to reduce pathogen
virulence. Despite their presence in many groups of fungi, dsRNAs have not been found
consistently to reduce pathogen virulence (Nuss & Koltin 1990). This lack of effect may
result from dsRN A transmission being necessarily dependent upon pathogen transmission
because dsRN A is only transmitted over a distance via fungal spores. In contrast to the
conventional wisdom of increased virulence leading to increased transmission, for these
types of cytoplasmic hyperparasites increased virulence could lead to decreases in
pathogen, and therefore hyperparasite, transmission. A recent theoretical model on the
evolution of hyperparasites stresses that a balance between pathogen transmission and
hyperparasite debilitation (i.e., virulence of the hyperparasite on the pathogen) will need
to be maintained for successful biological control of a pathogen by a hyperparasite
(Taylor et a1. 1998). In the absence of other evolutionary pressures, the model also
predicts that hyperparasites should evolve toward mutualism (i.e., low virulence or
debilitation) with their pathogen host thereby having a detrimental effect on the plant host
population. In other words, biological control of a pathogen by a cytoplasmic element
may not be effective in the long term.

However, dsRNAs may indeed be useful biological control agents for the chestnut
blight fungus, Cryphonectria parasitica (Murrill) Barr. The accidental introduction of C.
parasitica, a pathogenic ascomycete, into the US. was first reported in 1904 in New

York (Merkel 1905) . The fungus spread rapidly throughout the range of the American

11

chestnut, Castanea dentata (Marsh) Borkh., devastating populations. Infected branches
and trunks are girdled and killed by the fungus. Chestnut blight was first documented in
Europe in 1938 and by 1967 infection had spread to most chestnut-growing areas
(Heiniger & Rigling 1994). Infection of C. parasitica by a dsRNA hyperparasite was
found to reduce pathogen virulence (also termed hypovirulence) (Jaynes & Elliston
1982), producing recovery of chestnut populations in Europe (Heiniger & Rigling 1994).
Hyperparasite infection debilitates C. parasitica and reduces pathogen virulence on
chestnut trees by reducing canker growth rate (Anagnostakis & Waggoner 1981), asexual
conidia production (Elliston 1985), and sexual reproduction (Anagnostakis 1984, 1988).
A superficial, non-lethal canker (healing canker) may result fi'om wound callus tissue,
which develops in response to pathogen infection in some cases, reducing canker
expansion rates and walling off the fungus. Trees producing non-lethal cankers are said to
be “recovering” because infected stems are not always killed by infection, allowing for
continued growth and seed production.

Introductions of dsRNA have not resulted in spread of the hyperparasite and
recovery of the American chestnut (MacDonald & Fulbright 1991). Michigan is the only
area in the US. where dsRNA induced hypovirulence has spread naturally (Fulbright et
a1. 1983). In Michigan populations, the proportion of dsRNA containing isolates and tree
recovery (presence of at least one callused canker on a tree) are positively correlated
(Davelos et al. 1997). However, dsRNA-containing strains are found in cankers below
surviving and dying branches with equal frequency (Davelos et a1. 1995). A preliminary
survey of two recovering Michigan populations found that infected branches which

survive have a larger diameter than those that die (Davelos, Fulbright, & Jarosz, unpub.).

12

    
  
  
 

’\'-

lass: 0b.».
151115 or t.
more 1:3:

This

“.Vr‘t'V" . ,\
.- Aunt-l

A '.,
a a. .
A»; :‘IQT'
l

"‘7. .h“
“Ar“.‘r‘tufl .-
- " ~ A.

“ «a:
9": .
.J‘I!

-S‘ t:
9:.

g.

These observations suggest that infections by both dsRNA-free and dsRNA—containing
isolates of C. parasitica kill small branches, while large branches are more likely to
recover from infections by dsRNA-containing isolates of C. parasitica.

This study focuses on the interaction between dsRNA—induced changes in C.
parasitica virulence and American chestnut branch size. Specifically, the following
questions were addressed: (1) How does the interaction between pathogen virulence and
branch size affect canker development and branch mortality; (2) Is the pattern of canker
development and subsequent branch mortality similar for cankers initiated from artificial
inoculations and from natural infections; (3) Is presence/absence of dsRNA predictive of
canker morphology in an inoculation study; and (4) Is branch size predictive of
morphology of naturally occurring cankers?

METHODS
S_tu_dy_S_it£

A number of naturalized populations of C. dentata have become established in
Michigan from trees planted by early settlers (Brewer 1995). Blight was first reported in
Michigan chestnut populations in the late 19205 (Baxter & Strong 1931). By the late
19505, healing cankers began to appear and the presence of dsRNA in isolates of the
fungus was discovered in the late 19705 (Elliston et a1. 1977; Day et a1. 1977). The site
that was selected for this study, County Line, has been infected with chestnut blight since
1958 and healing cankers began to appear in the early 19705 (Brewer 1995). Studies of
vegetative compatibility group diversity, RFLP variation, and DNA fingerprints of the C.
parasitica population at County Line revealed that a single clonal lineage predominates

(Milgroom 1995; Milgroom & Lipari 1995; Davelos et a1. 1996; Liu et a1 1996; Davelos

l3

et a1. 1997). The dsRNA population at this site has also been found to consist of a single
type (Paul & Fulbright 1988; Peever et a1. 1997). The dsRNA found at County Line
cross-hybridizes to Cryphonectria hypovirus 3-GH2 (Paul & Fulbright 1988; Peever et al.
1997) and is therefore a member of the virus family Hypoviridae (Hillman et a1. 1995).

A single spore isolate of the most common C. parasitica vegetative compatibility
genotype found at the County Line (CL) site, which also contained dsRNA, was selected
for the field inoculation experiment. This genotype was identified from a population
level survey (Davelos, Schaupp, Jarosz, & Fulbright, unpub.). The complimentary
dsRNA-free strain was generated from a single spore isolate of the hyperparasitized
strain; dsRNA is not present in all conidia of a culture (Fulbright 1984, Russin & Shain
1985, Enebak et a1. 1994). A vigorously growing isolate which was putatively dsRNA-
free was selected from among the conidia] offspring. This isolate was screened
essentially as described in Fulbright et a1. (1983) to confirm the absence of dsRNA.

Virulence of both the dsRNA-free and the dsRNA-containing isolates was
determined by comparing these strains with known controls (virulent dsRNA-free isolate
Ep155 and debilitated dsRNA-containing isolate GH2) using a bark inoculation test (Lee
et a1. 1992). The pair of isolates selected for the field experiment produced lesions on
excised bark that differed significantly in area from each other 5 days after inoculation
(dsRNA-free CL isolate: 3.8 cm2; dsRNA-containing CL isolate: 1.9 cm’; t = 4.43, df = 2,
P < 0.05). Lesions produced by the dsRNA-containing isolate did not differ in area from

those produced by the GH2 control (GH2: 2.1 cm2; t = 0.74, df = 2, NS); lesions

, . _
~r- JQ' 'Ak '
“g...“ ¥

7
--« .h in
C_eu:.. th»

. 1 ‘1
'3 aflgq. i
$555.51.; . .

:\< 1")0')? "F
“AI»S~‘ . H,

In.

T” ; o ..
fw'l' 1d: L

. ~.
a. - ,

" ‘1I‘ .‘
5.1.5. )gkuu

 

 

produced by the dsRNA-free isolate also did not differ in area from those produced by the
virulent control (Ep155: 5.3 cm2; t = 3.45, df= 2, NS).
Inoculation Experiment

The effect of the interaction between branch size and pathogen virulence on
branch mortality was explicitly tested by inoculating branches of varying size at the
County Line site. In July 1997, a total of 60 uninfected branches on separate trees were
selected: 20 small (5 2 cm in diameter; mean diameter: 1.6 cm), 20 medium (2-4 cm in
diameter; mean diameter: 3.0 cm) and 20 large (24 cm in diameter; mean diameter: 5.3
cm). Half of the branches in a size class were inoculated with the dsRNA-containing C.
parasitica isolate; the other half were inoculated with the complimentary dsRNA-free
isolate. Branches were assigned randomly to treatment groups. Mean branch size did not
differ significantly between treatments within a size class. Inoculations followed the
agar-disk cork-borer method of Griffin et a1. (1978). Canker size (length and width) and
morphology (non-callused or callused) were monitored bi-monthly through October 1997
and again in April, July, and September 1998. If a branch was girdled by a canker, i.e., if
the branch above the canker was dead, only canker width and morphology were recorded.

Natural Infections

To investigate how branch size affects branch mortality and canker morphology
of naturally infected branches, uninfected branches on 100 trees were selected randomly
at the County Line site. These branches were wounded with a sterilized aluminum nail to
promote natural infection because C. parasitica is thought to enter trees through wounds

in the bark (Anagnostakis 1982). Wounding was done in July 1997 because infection is

15

I" '
.‘N‘. .131.
m L. AAS'

mL

1 '
'1“" 531'
“.151. but“

9:15.430"- '
,, *I.J-h|-‘ I
|

“’7‘?" 3'1"."

tr. I. . am-\
«7v», \"
“mik‘i5 \

.I;-'.'i;'..39 1 -
N‘A.\~~i LI

 

9‘-
a
- (
is
‘5‘ ’IF
A "-
em 2
1L
...

_ .
“1‘, 121g:

most likely to occur at wound sites created in summer (Steven Jakobi, pers comm).
Branch diameter at the wound site was noted. These branches were monitored in
September 1997, April 1998, and September 1998 for the appearance of cankers. The
whole branch from the wound site distal was inspected for the appearance of new
cankers. When cankers were found, they were classified as non-callused or callused. The
diameter of the branch immediately below each new canker was measured.
Statistical Analyses

All analyses were performed with the SAS statistical package, Release 6.12 (SAS
Institute, Inc. 1997). To examine differences in canker expansion rates for the
inoculation experiment, two response variables were used: canker area (V2 width X '/2
length X 1:) and canker width / branch circumference. The former measure was selected
for comparison to previously published studies on canker expansion rates. The latter
measurement was chosen because of its biological relevance; a value of 1.0 indicates that
a branch has been completely girdled by a canker. These repeated measures data were
analyzed with profile analysis using multivariate analysis of variance (MANOVA) with
PROC GLM. Profile analysis examines the shape of the response curves
(T irne‘Size‘dsRNA), the levels of the main effects (Time*Size and Tirne*dsRNA), and
the flatness of the response curves (Time) (i.e., whether or not the slopes of the curves
differ from zero) (von Ende 1993). Area and width/circumference at each survey were
the response variables and branch size class, presence or absence of dsRNA in the
inoculation, and the interaction between them were the main effects included in the

models. Pillai’s trace is the test statistic reported because it is the most robust to

16

has: 315

553A 0: i
1.1mm:
11.5112; 11
5:111:51: pa:

‘--~3-. , .
lvk'-.~5U L'AL)

 

h .

W’A'argau .
y-~.\r:u. J,» .

 

 

 

violations of assumptions (Scheiner 1993). Contingency table analyses for the
relationship between branch survival to September 1998 and either presence/absence of
dsRNA or branch size class were performed using PROC FREQ. Comparisons among all
six branch survivorship curves were made using Peto and Peto’s logrank test (1972)
following the methods in Pyke and Thompson (1986 & 1987). Significant differences
between pairs of curves were examined using the same methods with Bonferroni
corrections applied to the significance levels for multiple comparisons. The relationship
between canker morphology (non-callused or callused) and branch size and
presence/absence of dsRNA was examined with logistic regression using PROC
CATMOD. Branch size and dsRN A were the main effects and canker morphology was
the dependent variable. This relationship was investigated both at the end of the 1997
survey (before any deaths had occurred) and at the end of the 1998 survey.

For analyses in the natural infection study, PROC GLM was used with branch
diameter as the independent variable and infection status, branch survival, or canker
morphology as dependent variables. Data were transformed (square root or log) when
needed to improve normality of residuals and homogeneity of variances. A posteriori
tests were performed using the Tukey-Kramer method. For analyses with survival or
canker morphology as the dependent variable, logistic regression (PROC CATMOD) was
used with branch size as a main effect for both models. Canker morphology and the
interaction between canker morphology and branch size were also included as main

effects in the branch survival analyses.

17

F9-

DE 11103;;

7"'."t '3 "‘.'
1.15.15 \.h

V

is

. .
‘ '1

'F'g ,«u
when; ..
m, ‘
1 .
:1 *‘T‘,‘ I.

‘swaab f...

For '.
75:33:23: 31--
L I.

N,“ -
P‘- be) “a -.

ifie.’ of re

fr

.-
*-

r-Q ' ‘
5‘.“ 5- .,‘

 

 

RESULTS
Inoculation Experiment

Four branches were eliminated from the study due to cankers developing below
the inoculation site or because the inoculation was unsuccessful. For the response
variable canker area, the level of the response for presence/absence of dsRNA was
marginally significantly different (Time*dsRNA) with cankers produced by dsRNA-free
isolates being larger and the slope of the curves was significantly different fi'om zero
(Time) (Table 2-1; Figure 2-1). However, branch size class (Time*Size) (Figure 2-2) and
the interaction between dsRNA and branch size class (Time*dsRNA*Size) did not differ
significantly for canker area.

For the response variable canker width/branch circumference, the level of the
response differed significantly for branch size class (Time*Size) and the slope of the
curves was significantly different from zero (Time) (Table 2-2; Figure 2-3). Although the
level of responses did not differ significantly for presence/absence of dsRNA, the ratio of
width/circumference was higher consistently for dsRNA-free isolates (Figure 2-4).

Survivorship of the branch distal to the inoculation site until the final survey
(September 1998) differed significantly among branch size class ( x2 = 23.14, df = 2, P <
0.001; Table 2-3) with large branches having the highest survival (95%), medium
branches being intermediate (68%), and small branches having the lowest survival (18%).
Survivorship to the final survey (September 1998) was not affected by presence/absence
of dsRNA in the isolate used for inoculation (x2 = 0.69 df = 1, NS; Table 2-3). When

the pattern of survival over time was examined, significant differences were found for

18

.
unfit -r‘ '-
SL null.)

~47: “ti—‘2‘
3.1..» sun:-

QI .1585.“

w“- 9
“‘5 «ACCT
:" 4' ."

 

survivorship curves among the six combinations of dsRN A presence/absence and branch
size categories (Log Rank = 105.09, df = 5, P < 0.001; Table 2-3; Figure 2-5). A general
trend was found for survivorship to decrease with decreasing branch size class. In
general, branches inoculated with dsRNA-containing isolates have higher survivorship
than those branches inoculated with dsRNA-free isolates. However, this latter trend was
reversed when considering small branches separately.

Canker morphology (non-callused or callused) may be influenced by the presence
or absence of dsRNA and branch size. This relationship was examined explicitly from
results of the inoculation experiment. At the end of the 1997 survey, canker morphology
was predicted by presence or absence of dsRNA in the inoculating isolate (x2 = 4.97, df =
1, P < 0.05; Table 2-4a). However, by the end of 1998, branch size category significantly
influenced canker morphology (x2 = 16.18, df = 2, P < 0.005; Table 2-4b) with larger
branches having the highest proportion of callused cankers. In contrast, dsRNA became
uninforrnative (x2 = 0.01 , df = 1, NS; Table 2-4a) with cankers being classified into
either category with equal probability. Further, there was no relationship between canker
ratings at the end of 1997 and at the end of 1998 (x2 = 3.50, df = 1, NS.) with about 37%
of cankers in each category in 1997 being classified into the other category in 1998.

Natural Infections
Whole branch-None of 100 uninfected branches wounded in July 1997 had detectable
infections at the wound site by September 1997. By April 1998, 90 branches remained in
the survey; the remainder were dropped from the study because they had died from

causes other than disease or a canker had developed below the wound site. Of the

19

  
 

"kf'l‘ ‘T’i "

1
,, Lubwnfi- .

"4’ 5‘»),
Ox*. 1‘:

Zinc-2 _.‘

.‘ ””1 ‘1
a: . “ ‘
.-
3‘.
~~ C‘-
\I
- Q .
\LILX ‘7 9.
~ A -
K-
s
._

\
-‘
.ah'
I‘lg
-‘\

branches remaining in the survey, 32% became infected at or distal to the wound site by
April 1998; an additional 50% of the remaining uninfected branches developed a canker
by September 1998 (Table 2-5). The total percentage of branches infected at some point
in time during the study was 66%. Of the 59 branches that became infected, 20 branches
developed more than one canker and 33 branches developed cankers at the wound site.
The size of the branch appeared to influence the likelihood of a canker developing
somewhere on the branch. For newly infected branches in both April and September
1998, the diameter of the branch at the wound site was significantly larger than those
branches that remained disease-free (Table 2-5).

Individual cankers.--The size of a branch where a canker develops could influence the
mortality of the branch distal to the canker. To examine this relationship between branch
size at the canker and branch mortality, individual cankers from the natural infection
study were analyzed. A total of 36 cankers were found on the 29 branches infected by
April 1998 with 53% of branches alive distal to these cankers. By September 1998, 50
new cankers were found with 86% of branches alive distal to these cankers. There were
no significant differences in branch size at the canker between branches that survived and
those that died early in 1998 (Table 2-6). However, by September there was a distinct
trend for surviving branches to be larger than those that died (Table 2-6). Branch size at
the canker was also related to canker morphology. Again there was no trend by April
1998 but by September, the diameter of branches with cankers which were classified as
callused were significantly larger than those with cankers in the non-callused category

(Table 2-6).

20

'

uh-

m

 

ms:

6

.611
su‘

54‘
35“-

.V
i«&

     
  

 

-. ~ 1
5.4:

b.
we“? '
L.h.5h

A

 

.\1

M
x »*\\u b

«6‘

s\ (.2'

I- 0.-
k...

h
5

. v
. ‘ .‘
‘Ab h—J

u..
A»
T:
..

With logistic regression one can investigate whether branch size is predictive of
canker morphology. In general, branch diameter immediately below a canker was
predictive of canker morphology (Table 2-7) with larger branches tending to have
callused cankers. Further, logistic regression can test whether canker morphology and
branch size are predictive of survival. Canker morphology was not related to
survivorship of a branch (results not presented) but branch diameter immediately below a
canker was marginally predictive of survival (Table 2-7); there was not an interaction
between canker morphology and branch size.

DISCUSSION

The importance of American chestnut branch size in determining the outcome of
infection of a branch by chestnut blight is supported by both the inoculation experiment
and the natural infection study. Branch size influences canker morphology and branch
survival although some of these effects develop over time. DsRNA reduces canker
growth rates and may also delay mortality of medium sized branches and potentially of
large branches.

The results showing that the presence of dsRN A reduces canker growth rates
regardless of branch size is consistent with the findings of other researchers. After 12
weeks significant differences among treatments were found when dsRNA-free and
dsRNA-containing isolates were paired in all combinations and inoculated onto chestnut
stems (Kuhlman 1978). Further, field inoculations of dsRNA-free and dsRNA-
containing strains from both Italy and the US showed significant differences in mean
canker area between strains with and without dsRNA after 6 months (Elliston 1985). The

measure canker width/branch circumference appears to be a good indicator of how

21

  
 

T121203 1

1': I ‘
u l M.)
111:” MAS

.
““!"“p,\
Gummsn
.

II? 55:55: _'

5'; ”a...-

‘ I .3

*‘u-kg..,l
.

 

1'7-‘->'--»,v-. a:
HAC\\\.\_\ .‘ _ ‘

‘-‘ ~

\ 1'3 '1 ”I

L ..’\§‘\ “\
c

U?!” «no. .

5.44.. A&_ ‘1

.11. -..,‘ I
1"“3‘3 L. \K

“-H-

1 .
. l b" .4
\~ 1..

F‘_
‘“?.
<‘\x
~-.
h.—
N..
h
‘3?
5‘
A‘.‘

infection is likely to influence branch survival. The closer the ratio is to 1, the more
likely the branch is to die, regardless of branch size or presence/absence of dsRNA.
Classification of cankers into callused or non-callused does not appear to
accurately predict the type of fungal isolate which initiated infection, especially after the
first season of infection. Branch size appears to be a more important influence on canker
development. This result is supported by both the inoculation experiment and the natural
infection study. In the inoculation experiment, a canker classified as non-callused is just
as likely to contain a dsRNA-containing isolate as a dsRNA-free one. Although isolates
from natural cankers were not screened for dsRNA, branch size was found to be
predictive of canker morphology especially as time since infection increased. A study by
McManus et a1. (1989) found no correlation between canker morphology and
presence/absence of dsRNA in naturally occurring cankers. This observation came with
the caveat that the cankers were several years old and therefore might contain a mixture
of pathogen isolates both with and without dsRN A which could confound the
interpretation of the relationship. Although it is possible that the inoculations in this
study subsequently became colonized with other strains of the pathogen, the result of no
relationship between canker morphology and presence/absence of dsRNA is consistent
with the findings of McManus et a1. (1989). This outcome regarding the lack of
importance of dsRNA and the role of branch size in canker morphology has implications
for field surveys of chestnut populations where the degree of recovery is rated by the
presence of callused cankers. If a population is dominated by small trees, many non-
callused cankers might be found and the population would be considered non-recovering.

However, dsRNA could be present in these isolates, which could have implications for

22

 

 

I}: L2'L\ -H a h;

”‘0, 3."
W‘\.J'LZ‘.

- -.
4‘; fl ..

‘~
\ a...

a." -,.
5.1.; .
r .. .5“

'-¢
.
o .-

“35‘
‘ ‘5 u -.

.l
V ‘ >‘V'
““A. \

£350”

‘1

11“

W's-'15: L

N

the long-tenn prognosis of the population. Conversely, the presence of callused cankers
within populations containing large trees may not indicate that the chestnut population is
recovering due to the spread of dsRNAs.

The effect of branch size at time of infection on survivorship was most striking in
the inoculation study, where it was found that small branches were more likely to die
regardless of the type of inoculum used. The finding that dsRNA alone is not related to
branch survivorship emphasizes that the presence of dsRNA in a pathogen population
may be necessary but not sufficient for biological control of disease. Although the
presence of dsRNA may enable large branches to survive, small branches still succumb to
infection by both dsRNA-containing and dsRNA-free isolates. This trend for disease to
have differential impacts on small versus large individuals has also been found in another
empirical study. In the Podophyllum peltatum-Puccinia podophylli plant-pathogen
interaction, small ramets had higher disease incidence and severities than large ramets,
resulting in higher mortality for small individuals (Parker 1988).

The length of time a branch has been infected appears to play an important role in
branch survivorship. No branch deaths were observed in either study until the second
season. Further, because many branches in the large and medium size classes were still
alive at the end of the study, the ultimate influence of dsRNA on branch survivorship
cannot be determined. However, there is a trend for the presence of dsRNA to delay
branch death for medium sized branches. Because there was little mortality of branches
in the large size class, no effect of dsRN A was observed; however, the presence of

dsRNA is expected to delay death for branches in this size category as well.

23

 

 

iv" (“'

1“ ‘
{.mlbu - II

 

 

<~- ..‘
““r “55;
.
n .
I

C
. -

“‘
_.

.

‘\ l

Overall, branch size at the time of infection appears to have a large influence on
branch survivorship, at least in the short term. Indeed, higher survival and more
superficial (i.e., callused) cankers were observed on chestnuts in managed clearcuts than
those in forest clearcut sites (Griffin et a1. 1991). Although the effects of tree size and the
presence of dsRN A were not explicitly examined in the study, trees in the managed
clearcut were probably larger due to lack of competition from other hardwoods. This
larger size could have contributed to the higher survival and presence of callused cankers.
A similar trend was found for European chestnuts; sprouts with smaller diameter at breast
height were more likely to succumb to infection by chestnut blight or competition
(Bissegger et a1. 1997).

These findings on the influence of branch size on survivorship have important
implications not only for individual trees but also for American chestnut populations as a
whole if one assumes that effects on trees are similar to effects observed for branches.
Significant reductions in fitness of smaller trees and saplings due to infection by either
dsRNA-free or dsRNA-containing strains of chestnut blight could affect population
growth and persistence. The presence of dsRNA appears to at least delay death of large
trees. In the meantime, these trees may reproduce adding new recruits to the population.
However, if these small individuals never attain reproductive size due to the negative
effects of infection, then the population will not persist through time although individual
large trees are growing and reproducing. Therefore, the presence of a sufficiently
debilitating dsRNA in the pathogen population may not be the only requirement for

recovery of American chestnut populations. The size of an individual tree at the time of

24

manta,

may»... -

 

”Jr ‘3'
ASIA. 5511.

 

infection could also be an important factor in determining the long term survival of

American chestnut populations.

25

 

iv

I
WI I nD-r

riru‘lfi
[.r\\s

 

,Img' 4

null. U

Thug

,.
4 ln‘\

Ir- .
1*.1' ~ .
1.1.1» I.‘

 

N
e“;\

Table 2-1. Profile analysis of canker area for the inoculation experiment. F-values are

for Pillai’s trace.

 

Effect F Numerator df Denominator df Prob > F
Time*dsRNA*Size 0.69 14 88 0.7787
Time*Size 1.33 14 88 0.2076
Time*dsRNA 2.22 7 43 0.0507
Time 20.95 7 43 0.0001

Table 2-2. Profile analysis of canker width/branch circumference for the inoculation

experiment. F-values are for Pillai’s trace.

 

Effect F Numerator df Denominator df Prob > F
Time*dsRNA*Size 0.92 16 86 0.5479
Tirne*Size 3.79 16 86 0.0001
Time‘dsRNA 0.58 8 42 0.7896
Time 87.75 8 42 0.0001

26

Table 2-3. Branch survival for the inoculation experiment. Number of branches alive or
dead in September 1998 for each category of branch size at the inoculation site and
presence/absence of dsRNA for inoculations performed at County Line in July 1997.
Different letters indicate significant differences in survivorship curves (Pyke &

Thompson 1986,1987 with Bonferroni corrections for multiple tests).

 

 

 

 

STATUS
DSRNA BRANCH SIZE ALIVE DEAD SURVIVORSHIP
CURVES
ABSENT SMALL (S 2 cm) 3 6 c
MEDIUM (2 - 4 cm) 4 5 be
LARGE (2 4 cm) 9 1 ab
Total 16 11
PRESENT SMALL (g 2 cm) 0 8 d
MEDIUM (2 - 4 cm) 9 1 ab
LARGE (2 4 cm) 10 0 a
Total 19 9

27

. _::::. C :.. 25:55:);

. 7...... 1.... A Ira: .7. .2. re...— :.u,:::» )1: .2 1:2: LL : ~ . .1
. 1.... a... 2:11.... ..\.7. £3.21. 7.... (.211. T. 9.1.9.12: a... . s s 2 ‘ : s \ < .. y s . .e\.\.,.\
.21....1 . .

E 3 2 o_ PZMmmME

 

 

E 3 I o. HZMmm/x
DumDAq/xu DmmDAA<U-ZOZ QmmDAA<U Qmmeq<U-ZOZ <deQ
Goa mo 95v A32 mo 6:3
>OOAOIAEOE mmMZ<U >OOAOILMOE Mm¥Z<U

473% .«o oocomnmbocomoa van 322.988 5&ch .m
.252 05 E cows—2: 8m 3:285 280 as: bcsoU «a EuEton—xo

coca—sooE of 5m mmflo ufim £285 was <de1 90 35388533 .5.“ Atoms—Bo .o vows—Eoéocv $205908 3&ch .v-m 2an

28

 

71.2.3 .¢\.7_ 2.12:»-...2.>31..T:T:.:: .3422. v Q

.26

00520005 05 3030 $8150 00 000800—050 05 00 000 :3 0:0 32 0002503 300% 05 80¢ .030th 203 mug—0:05 060m ”802

 

M: m 3 0 A80 v NV mom/3

w I o_ S A80 v - NV 35592

N 2 0 2 05 N WV 35%
00535 050020-202 850,20 003155-202 0N5 muzém

 

032 0c 23 A32 .8 .080

>OOAOE§OE MMMZ<U >GOAOIQ¢OE Mm¥Z<U

.330 BE £0005 000 30—2900.— .3000 .0

29

 

1V;-:~)',3o
bun-‘5: '

“.12)-”,
u‘n....s\ \»

rivgjv
b..«., H.
(”.3
5‘.»

Table 2-5. Average branch diameter at the wound site in the natural infection study.

Date of survey, number of new branches found with at least one canker, the average
diameter of a branch at the wound site (i.e., general branch size) for newly infected versus
uninfected branches for the natural infection study at County Line. Means : standard

errors are presented.

Average diameter of

branch at wound site

 

 

Survey date Newly infected Infected Uninfected
branches branches branches
September 1997 O - -
April 1998 29 2.6 i 0.19 1.7 i 0.08 FL87 = 26.31, P < 0.0001
September 1998 30 1.9 _+_ 0.12 1.5 i 0.10 FL57 = 7.15, P < 0.01

30

n

.hlvuod————.-.v .~—_-H .lnh...i-.l-III-1 .II Ila-II I (q .

 

l .
III—D} 1:.).I.:.7 r: 5.44:3: 33...}. £9.22». t.
. up: - .7 can . a - .
— .. .. ax . 4 4...A— >73... Cal—9.4.... .2232: .7: 5.41....09 3:13». 5.1.2:. ‘3 ......::...:. 912...}: :1: CAR. 4. 2‘12...‘ ~ 0
, . 12.2.1. .x...o. ¢~: .
\ ,. I. -\ r.‘ U\u 0
. P a \s. \

 

 

 

58.0 v 0 Rod v 0 002 00%
.00: u E0 3.0 H 2 8.0 H 3 .30 u 2.0 8.0 H 2 20 H 2 80 $2 .000 200B 2:002
:50 v 0 80 v 0 $2 .000 2&3
. .I I . .“wv; .I. ...l. . .Q .Q
$0730 20+: 20:0 5: 0 02+: :33 So 32 0 cm 0032 . <03?
0.2 .mz
.30 u .20 2.0 H 3 $0 0 00 .0: n 50 2.0 H Z 20 H 00 $0 32 .a< 080 .a< 2800
80:28 00 00
0002—00002 0002—00 0009 wEZEm 0080005 _0>_>.Sm 0000008 0000
000800 30—00 00005 000—000 30—00 00005

.00 00.0800 0w000><

00 0000800 0w000><

000000000 000 $0.00 0000005 H 00008 30% .005

50000 00 0.00 00000.08 _00800 05 8 0000—000 032208 00.0 $200808 000—000 0000:0000: 00m00> 0002—00 00.0 000 0000005 0000

030000, w8>m>80 00.0 000—000 30—00 00005 .«0 0000800 000 @8380 0000005 mo 000000000 .300 003200 000—000 00 .0000 00005 0000

.0008 000—000 0003 3080 00000.08 _00800 00.“ 000—000 30—00 00005 .00 0000800 0w000>0 000 .30—00808 000—000 ._0>_>00m .0-N 030k

31

 

 

I
.117 .722..— 5...4..:... 0:2- 43:17. 2.1.9.0.... .5232: 2.. _...:):.7 .12.. gr}... £2.2E: i... .1-.-

83 v a ._ ”(a wow n ”X :55 v a ._ may 52: u N96 M32 .aum M32 .aom 2&3 2522

:2 .23 2&3

 

o; v m ._ it e: n “X mood v a ._ u% .2.” u. ”a 32 .25 a woe .a< .32

o; v m ._ ":12: n ma .m.z ._ ”a. «2 n ”X woe .a< M32 .a< Baum
awe—0:92: 8

333% can 88:85 “8.58 can 88:35 a: 12,326 380%: San

6:5 fiasco S 32% :ouoomfi .832 05 5 £823 3:239: 8m 3333 can 36 5an 5253
2:28:28 95 3205908 “8.58 28 5x58 323 58:36 555 5253 @3223": .55 3333 “8.58 8 Emma. 328.5

8% .958 8x53 8mm .avam cozoomfi .83? Ba Rita was .83 £285 .323908 .828 5253 afimcozfiom .n-~ 2an

32

2...? 2.256) 2.: < 534:... w

.35 fiasco «a Evacuees coca—=02: 05 how noun—:85

85m mag 32?» <mew 5223 can 55, 830$ .33 Baa—:85 8:255 Bu $8.6 235% H 983 33 how—:3 582 .Tm oBmE

coca—=3:— ..o=a aha:

o—n 3N EN 09— e: we @—

b
I

II
*-

p
O

I
‘

.12.

 

+<Z¢nu+
u<zmnvIOI

 

 

8
(mm) my

 

ram

flank. zap-2' 60h< hue—END

33

 

85:. 2.3.3.: 2.... f. 93...... v

 

6:5 fiasco E Eon—toga

coca—=85 2.: 8m cots—=85 85m 98v 333 mmflo ofim 505:9 some 8m Btu Eng—Sm H $83 8.6 8x50 :32 .N-~ oSmE

5:232: been 9?:

3n can an 90— e: 8 @—

 

 

v

853+ m
Siam—all! \m,
352ml?- 0.

 

 

 

25,—. 2.9.: 3.2 kin-.0

34

6:5 .0550 E 868598 gas—:85

65 :8 832305 665m 98: mam:6> $26 6N6 5:8: :86 Sm 68:6 23:53 H 8:269:586 5:652:23 :8—56 :32 .m-~ Rama

522:3:— ._8._a 9:5

c; can 03 can 3N 8—

p r
I q

I.

 

 

 

 

 

 

 

 

m
w
:92:qu M
2252+ m
4255* m.
w m
n t ”.5 m

n I n

. . u.—

 

oEC. 6:29» «vacuum—83:53.23

35

 

.a:-m.-. r-ZaLEi .-.u-§i.uthu-uo\1~. \\ 4‘1":-

0:5 3580 E £08833»: gum—=85 on: :8 Evan—=85 8:6 via: 96:?»

<me: 5933 v5: 53> anus—Ag 5:5 :22505 35:8: :8 88:0 835% H 8:288:83 55:52:23 88:8 :32 .v-~ 0.5me

5.53::— ..8._a 9:5

 

 

 

 

 

 

 

3.. won in can in co—

m
m.
w m
+<Z¢€IOI n...
n I . 0
-<z¢%l0l n c m
‘ - fl.
._ r w... u
u
a

. a 5.:

. . a a...

. J . :5

.r _

25,—. 25:9, 35:883.:Uiu—83

36

 

<Z¢me+ max—<1— +
Slam—T HU¢<J+

<Z¢mv+ ED—DHEIII
tide—T ED—GQEIII
<dee+ Jagmlol

<21»? dd<~2ml§l

 

 

 

 

458530 Sam—:85 0:: wfiSc w:_>m>:=m 3:055 8o 8:535 .m-~ 0:53:

5:232:— 32: was:

c—v can c—m can 3" we—

 

 

:—

 

95:. .9, 32?:5 5:95

we

@—

... N6

:2.

:9:

live:

 

SngAguns saqauug 30 uomodmd

37

NUS

5

‘v\“‘.

--u

..

Fh'

Ln”):

..“

.u.\

a)

CHAPTER 3
EFFECTS OF CRYPHONECTRIA PARASI T ICA AND DOUBLE-

STRANDED RNA INFECTIONS ON AMERICAN CHESTNUT DEMOGRAPHY

INTRODUCTION

Pathogens can influence plant community diversity and species distributions
through their effects on fitness and the outcome of intra- and inter-specific competition
(Burdon 1982; Burdon 1987). Both host survivorship and reproduction can be reduced
by pathogens (Alexander & Burdon 1984; Clay 1984; Parker 1986; Paul & Ayres
1986a&b; Parker 1987; Paul & Ayres 1987; Wennstrom & Ericson 1990; Wennstrom &
Ericson 1991; Jarosz & Burdon 1992; Roy & Bierzychudek 1993; Thrall & Jarosz 1994;
Garcia-Guzman et a1. 1996; Garcia-Guzman & Burdon 1997; Marr 1997). However, the
focus of most empirical studies has been on the individual rather than on population-level
effects.

Matrix projection models can be used to determine population growth rates and
Stable stage distributions for age or size structured populations given information on
survival and reproductive rates. The finite rate of population increase (1), a measure of
p0pulation growth rate, indicates whether a population is growing or declining in size
(Caswell 1989; Silvertown & Lovett Doust 1993). The finite rate of population increase
has been shown to vary with habitat and over space and time for a variety of plant species
(e-g., Menges 1990; Kalisz & McPeek 1992; Horvitz & Schemske 1995; Kephart &

Paladino 1997; Leege 1997). Matrix models have rarely been used to evaluate the impact

38

of 6;

Hill

 

-, 193‘
l-yJ-I‘tl

1 ‘vgfi
. ‘ #\
In. x»...

VHF. "r
K
“”5 '. .
.

‘(Hf-v
5.“...

I.“ i
'n.
r "f:
.
‘1.
“Q
.-
h‘\
*s.
s-

of disease on 7» (Emery 1998). However, estimates of changes in population size over
time for populations of Silene Iatifolia (= S. alba) infected with Microbotryum violaceum
(= Ustilago violacea) show that populations with a high proportion of diseased individuals
increase in size at a slower rate (Antonovics et a1. 1994). Further, the population
expansion rate of Silene dioica populations was reduced by infection with M. violaceum
(Carlsson & Elmqvist 1992). This potential difference in population grth rate has
important implications for the persistence of host populations in areas where disease is
common. Stable stage distributions, the fixed ratio of the number of individuals between
stages, can be compared to the observed population structure to predict whether or not
the size structure of a population will change over time (Silvertown & Lovett Doust
1993). For example, if disease causes greater mortality for small versus large individuals,
the size distribution of infected populations could shift towards larger individuals. Matrix
population models have been used to examine how disease may alter the stage
distributions of infected plant populations (Antonovics & Alexander 1989; Frantzen
1994)

The effects of pathogens on population level processes of their hosts can be
investigated using matrix projection models. By comparing uninfected host populations
to populations infected with pathogens of different levels of virulence, defined here as the
negative effects of infection, the influence of pathogen virulence on population grth
rates and stable stage distributions can be directly assessed. To examine how infection
may affect host demographic processes, I focused on the interaction between the

American chestnut, Castanea dentata, and its pathogen, Cryphonectria parasitica, the

39

.cv, ‘
4-
bun...

I.‘

.‘.

causal agent of chestnut blight. This pathogenic fungus itself can be chronically infected
with a cytoplasmic hyperparasite, a double-stranded (ds) RNA, which debilitates the
pathogen by reducing growth and sporulation, effectively reducing its virulence. In host
populations infected with hyperparasitized fungi, trees may respond to infection by
producing wound callus tissue and walling off the fungus leading to recovery of the
infected individual. However, no work has been conducted to determine if tree recovery
is associated with recovery at a population level, i.e. successful recruitment of seedlings
into the population. It is possible that seedlings could become established in populations,
yet never reach reproductive status because they are killed by blight. Indeed, in a
population of American chestnuts maintained by root sprouts, stems are small in size and
had high mortality (64%) over a six year period (Parker et al. 1993). Even if the
pathogen’s virulence level is reduced, the small stems of young trees could be girdled and
killed by the fungus. Further, infection could have a more dramatic effect on the growth
of small versus large trees. This could have important implications for the long-term
persistence of chestnut populations if infection prevents small trees from reaching
reproductive size even if large trees are relatively unaffected.

The objective of this study was to compare demography of healthy, non-
recovering, and recovering American chestnut populations. The presence of large
infected, yet surviving, trees has been thought of as chestnut population recovery
(Fulbright et al. 1983); however, this may not represent recovery in a demographic sense.
I investigated the extent of recovery by comparing the finite rate of increase (2») in healthy

and recovering chestnut populations. The effects that virulent forms of C. parasitica

40

A,“
:- ..

0553

TS:

‘A'
f

.Y’

L”.
1“

”T1

(i.e., not infected with dsRNA) have on host demography also were determined using x.
The finite rate of increase was calculated from transition matrices constructed from
measurements of growth of adult trees, seed production, and seedling establishment and
growth at each population site. Further, differences among size distributions and the
observed and stable stage distributions for each population were examined.

Another result of this study is the examination of the role of changes in pathogen
virulence on host population level processes. A trend towards evolution of reduced
virulence should be found because pathogens with high virulence could destroy their
resource base (Harper 1977). However, in a review of empirical studies of plant-
pathogen interactions trends favoring reduced virulence were not found commonly
(Jarosz & Davelos 1995). Further, a model by Lenski and May (1994) indicates
reductions in pathogen virulence need not result in an increase in host population size, i.e.
recovery. The study presented here represents a situation in which reductions in
virulence are due to the ecological process of infection of the pathogen by an intra-
cellular hyperparasite rather than an evolutionary change in the pathogen population.
Demographic methods may be an effective tool for evaluating the use of hyperparasites as
potential biological controls.

METHODS
Study System

For a description of the introduction of C. parasitica into the US. and the effects

of dsRNA on the pathogen, see Chapter 2. Before the spread of chestnut blight

throughout the eastern US, chestnuts were a dominant overstory species (Day & Monk

41

.g)‘
Sub.

"-
.wu'

H.
p:

.-§

7.‘

|

1974; Karban 1978; Russell 1987). Today, although few trees reach reproductive status
and populations are maintained by sprouts fi'om the rootstock (Paillet 1984, 1988, 1993),
chestnuts are a dominant understory component of plant communities within their natural
range (Keever 1953, Russell 1987; Stephenson et al. 1991).
M
For the history of American chestnuts and chestnut blight in Michigan, see
Chapter 2. Michigan appears to be the only area in the US where dsRNA induced
hypovirulence has spread naturally (Fulbright et al. 1983). For this reason, Michigan
provides an unique opportunity to study this interaction because of the presence of
healthy, non-recovering, and recovering populations of the American chestnut. l have
defined recovering populations as those dominated by trees with healing cankers; non-
recovering populations have less than 30 percent of trees with healing cankers. Six sites
in Michigan were selected for study: two healthy populations (Leelanau and Missaukee
Healthy), two non-recovering populations (Stivers and Missaukee Diseased) and two
recovering populations (County Line and Frankfort) (Figure 3-1).
Population Projection Matrices
Stage classes.--Individuals were classified into size-based categories to construct
transition matrices. Lefltovitch (1965) stage-based matrices were used instead of Leslie
(1945) age-based matrices because individuals of the same age could be very different
sizes due to disease or other environmental factors. Therefore, size class is likely to be
more predictive of an individual’s fate than age, at least in the short term (Caswell 1989).
Lefkovitch matrices have non-zero elements in the main diagonal, which represent

survival within a stage from year to year, and in the top row of the matrix, which

42

ll'

YETTCS:
l

I
0' ng‘
c. 5 6i\

.
Elam;
"“15,

represent production of offspring. The subdiagonal represents growth from one stage to
the next, not an increase in chronological age (Silvertown & Lovett Doust 1993).
Elements above the main diagonal represent reductions in size, also termed
retrogressions. As in other demographic studies of woody plants, I based stage classes on
plant size measured as height and stem diameter at breast height (dbh) (e.g., Huenneke &
Marks 1987; Enright & Watson 1991; Pascarella & Horvitz 1998).

Eight classes were constructed for each population (Table 3-1; Figure 3-2). Stage
1 includes only first year seedlings; often the seed remained attached to the new seedling
at the start of the season so they were identified easily. Beyond this stage, some small
plants were derived from these first year seedlings; others were the result of infection of
larger plants or size suppression by herbivore browsing. Infection by chestnut blight can
drastically reduce plant size by girdling and killing chestnut stems leaving only small
sprouts from the root collar surviving. Significant differences in survival probability
between true seedlings and other small individuals were found (see Chapter 5, Table 5-1).
Therefore, separate classes were used for small plants derived from seedlings (Stage 2),
and for those that result from disease or herbivory (Stage 3). Transitions for plants larger
than 50 cm in height were based solely on size, not on their history.
Population sarnpling.-- Transition probabilities between size classes of some juvenile and
all potentially reproducing trees were obtained from yearly measurement of diameter at
breast height (dbh) for the largest stem of all individuals. Size reductions (retrogressions)
from these larger size classes occur when the largest stem dies and the surviving sprouts

are smaller in size.

43

 

 

l

Estimates of natural recruitment and transition probabilities for small individuals
were obtained by setting up permanent plots in each of the six study populations in May
1996. Plots were 9 m X 9 m and were placed haphazardly in the population. This type of
plot placement allowed different microsites to be sampled. In a site in Wisconsin with
few diseased trees, Paillet and Rutter (1989) found 1-5 seedlings/yr/hectare became
established in chestnut-dominated woodlands; higher rates of establishment were found
in other microsites, such as field edges. The number of plots per population ranged from
8 - 16; enough plots were used to cover approximately half the area of a given population.
Plots were divided into quarters and the comers were marked permanently with flags. All
seedlings within a survey plot were marked. In each quadrat, up to two resprouts,
individuals sprouting from a root collar with the largest living stem shorter than 100 cm
in height, and two damaged individuals, shorter than 100 cm in height but clearly not first
year seedlings, were marked (maximum of eight each per plot). If an individual within
these categories had multiple stems, one stem was selected randomly to be followed.
Surveys were conducted in early (May/June), middle (July/August), and late (September/
October) season. New seedlings were marked and followed as they appeared. Height
was measured for each plant. If a plant was dead for two successive surveys it was
removed from the census.

Reproduction by an individual tree was determined by counting the number of
branches with btu'rs for each tree. Three branches were selected haphazardly and the
number of burrs on these branches was counted. This method allowed the total number

of burrs produced on a given tree to be estimated. Since nearly all burrs contain 3 seeds

44

 

UN '3

ub NI

.
My ‘0‘
u‘““i

.
’3: ‘7-
.g“‘

4.
"“M

V a
1“,

:1"

'I

,\
.A‘~

(Jayne s 1978; Paillet & Rutter 1989), number of burrs was used as the measure of
reproduction for an individual.
Projection matrices.--To calculate demographic parameters for each population, the
following matrix model was used:

n(t+1) = An(t)
where n(t) is a vector of the number of individuals in each size class at time t, n(t+1) is
the vector for the population at the next time interval, and A is a matrix that shows how
individuals in each stage class at one time may become or contribute to each stage class
one time unit later, in which columns refer to the stage at time t and rows refer to the
stage at time t +1 (Figure 3-2). To calculate demographic parameters, the elements of A,
fecundities and transition probabilities, are assumed to be constant parameters (Caswell
1989; Silvertown & Lovett Doust 1993). The vector n(t+1) is calculated iteratively by
multiplying the transition matrix, A, by the vector n(t) until a stable stage distribution is
reached at which point A, the finite rate of population increase, can be estimated. For this
study, A is an 8 X 8 matrix whose elements a” represent the transitions or contributions of
individuals in the jth class at time t to the ith class at time t + 1 (Figure 3-2). These are
given by the survivorship, growth, and fecundity of individuals between time t and t +1
(Caswell 1989). In this study, the time interval is one year. The finite rate of population
increase (I) and stable stage distributions were calculated for each population using
RAMAS/stage (Ferson 1994). Further, the minimum 7. for each population was
calculated. This value is equivalent to the average survivorship for all individuals in the

population (Silvertown et al. 1996).

45

Because there appears to be no seed bank in these American chestnut populations
(Davelos & Jarosz, unpublished data), including a seed to seedling stage in the transition
matrix would introduce a time-lag into the population projection (Caswell 1989; Enright
& Watson 1991). Therefore, the average number of seedlings produced per tree in each
size class was calculated to fill in the seedling row (stage 1) of the matrix (see Table 3-2).
Approximately half the area of a population was covered with natural recruitment plots;
hence, the number of seedlings found in the plots was doubled to obtain an estimate of
total seedling production for a population. The average number of seedlings per

individual in a stage was determined using the following formula:

(no. of burrs in stage/no. of burrs for mpulation) X no. of seedlings for mpulation

no. of trees in stage

Since a maximum of 8 each of resprouts or damaged plants were followed in the
plot census, the total number of small non-seedling individuals is underestimated.
Therefore, to determine the number of small non-seedling individuals within each
population, the number of resprouts and damaged individuals within each survey plot was
counted in 1998. This number was doubled to estimate the total number of small non-
seedlings for a population. The observed ratio from the census of individuals in stages 3
and 4 for 1996 and 1997 was multiplied by this estimate to give the total number of
individuals in these stages for 1996 and 1997. Further, at the start of the census in 1996,
second year seedlings could not be identified. To determine the number of individuals in
stage 2 in 1996, the observed ratio of stage 2 to stage 3 from 1997 was used to estimate

what proportion of 1996 stage 3 individuals should be assigned to stage 2.

46

Some biologically important transitions were not observed in all populations. For
example, in some populations, no individuals were observed to grow from stage 2 to
stage 4. Further, for both non-recovering populations, the forward transition for stage 7
individuals to stage 8 was not observed in either year, i.e. no growth of trees in the 10-20
cm dbh size class to the greater than 20 cm dbh size class was observed. In these cases,
estimates of transition probabilities were made by calculating the average probability of
that transition across all populations. If the transition did not occur in any other
population (e.g., death in the largest size class), the total number of individuals across all
populations in a stage was determined. A transition probability was estimated by
assuming there was one more individual in that class and that individual was assumed to
make the transition; the probability of that individual transitioning (either growing to the
next size class or dying) was used as an estimate for that matrix element. For the 1996-
1997 transition matrices, stage 2 probabilities from 1997-1998 were used since these
individuals (second year seedlings) could not be identified with certainty in 1996 when
the census began.

To examine how these estimated transition probabilities affected population
growth rates, simulations were performed in which the transition probabilities were
altered and the effects on I. were investigated. For the transition from stage 2 to stage 4,
the transition probability was increased by 10, 100, and 1000 percent. For the transition
of stage 7 to stage 8 in non-recovering populations and for death in stage 8, the effect of

reductions in these probabilities by 1/10, 1/50 and 1/1000 was investigated.

47

Statistical Analyses

ngu_lati_on_grg_“_rt_h.--Finite rates of population increase were determined for each
population for two census periods. These data were analyzed with profile analysis using
multivariate analysis of variance (MAN OVA) with PROC GLM. Profile analysis
examines the levels of the main effects (Time*Disease), and the flatness of the response
curves (Time) (i.e., whether or not the slopes of the curves differ from zero) (von Ende
1993). The disease status of a population--healthy, non-recovering, or recovering--was
the main effect included in the model. Roy’s greatest root is the test statistic reported
because it has the greatest power of the tests for significant differences among groups
(Scheiner 1993).

Size distributions.--Comparisons of the observed proportion of individuals in each size
class among years, size distributions at each site to their projected stable stage
distributions, and stable stage distributions in different years were made with the log
likelihood ratio, G, using PROC FREQ in SAS. This test is more robust than 12 for
evaluating goodness of fit (Sokal & Rohlf 1981). The stable stage distributions were
generated using RAMAS/stage (F erson 1994). These distributions are given in relative
frequencies; therefore, to make comparisons to the observed data, the relative frequency
distributions were converted to expected number of individuals in each stage.
Cross-sectionm.--To examine how disease might affect the average size of
individuals within a population, the mean cross sectional area at breast height for trees
greater than 0.2 cm dbh was calculated for each population. These data were analyzed

using repeated measures analysis as described above.

48

RESULTS
Battems in Transition Matrices
Transition matrices for each of the six study populations for 1996-1997 and 1997-
1998 are presented in Table 3-23-1. Survival within stage 1 (first year seedlings) was
generally lower in healthy populations as compared to infected populations. For stage 2
individuals (seedlings at least 2 years of age), survivorship was high. However, growth
from stage 2 to stage 4 was observed in only one population for one census period
(Missaukee Diseased, 1997-1998). Survival within stage 3 was generally intermediate
between that observed for stage 1 and 2; the transition from stage 3 to 4 was relatively
high in non-recovering populations.

Retrogressions (size reductions) were observed in all populations for trees in
stages 4 to 6, and the pattern of retrogressions (all transitions down to stage 3 were
possible) for these stages was similar among populations. In general, the magnitude of
growth transitions (growth to stage 5) was greater than the magnitude of retrogressions
(size reduction to stage 3) for individuals in stage 4. For stage 5 individuals, in half the
matrices, growth transitions had a greater magnitude than retrogressions; in the other half,
the reverse pattern was observed with retrogressions being of larger magnitude than
growth transitions. In contrast, in 9 out of 12 cases, the magnitude of retrogressions in
stage 6 was greater than the magnitude of the grth transitions. Further, at the Frankfort
site, for all transitions in these three stages across two years, the magnitude of
retrogressions was greater than that of growth transitions.

Populations differed in the pattern of size reductions for stages 7 and 8. No

retrogressions were found for these size classes in either healthy population across two

49

years, while all populations infected with C. parasitica had some large trees (> 10 cm
dbh) which were reduced in size. The severity of size reductions was greatest at the two
non-recovering sites (Stivers and Missaukee Diseased). In particular, the proportion of
stage 7 and 8 trees moving to smaller size classes at the Stivers site was extremely high in
1996-1997.

The pattern of reproduction also changed when disease was present. At the two
healthy sites, only stage 8 trees reproduced. In contrast, trees in stage 7 produced seed at
all four diseased populations across two years and stage 6 trees produced seed at three of
these sites in at least one year.

Bopflgfiion Growth Rafi

Finite rates of population increase (7t) ranged from 0.978 to 1.000 based on 1996-
1997 matrices and from 0.992 to 1.021 based on 1997-1998 transition probabilities
(Table 3-3). There were no significant differences in It among population types, and A
did not change over time within a population (i.e., slopes were not significantly different
from zero) (Table 3-4). However, the relative ranking of the populations based on the
1996-1997 transition matrices suggests that disease is having an effect on the grth of
chestnut populations. Indeed, for estimates of population growth based on 1996-1997
transition matrices, there was a significant correlation between population type and 2. (rs =
0.956, P < 0.0014). Healthy populations have the highest population growth rates and the
two populations classified as non-recovering have the lowest values, with the recovering
populations being intermediate. The results from 1997-1998 indicate there is year to year

variability in estimates of the finite rate of population increase both within and among

50

populations (Table 3-3). The highest value of A was found for a recovering population
but again the lowest population growth rate was observed for a non-recovering
population. There was no correlation between population type and 7t based on 1997-1998
transition matrices (rs = 0.667, NS).

Minimum values of A ranged from 0.823 to 0.971 for 1996-1997 and from 0.855
to 0.967 for 1997-1998 (Table 3-3). The observed values of population grth rates were
closer to the minimum value for non-recovering populations than for both healthy and
recovering populations. Indeed, observed population growth rates for non-recovering
populations were generally less than 5 percent above the minimum value while the
observed growth rates for the other population types ranged from 5.1 to 21.5 percent
higher than the minimum value of 7t (Table 3-3).

Reducing the transition probability of stage 7 to stage 8 to near zero (0.00005)
for the two non-recovering populations (Stivers and Missaukee Diseased) did not
qualitatively alter results based on 1996-1997 transition matrices (Table 3-3). However,
this change did affect the relative ranking of population growth rate for the Stivers
population in 1997-1998. When this low transition probability was used, both non-
recovering populations had relatively lower population growth rates than the other
population types. However, this change did not alter the results of the repeated measures
analysis (results not presented). Because the death of an individual in the largest size
class (dbh > 20.0 cm) was not observed, the probability of an individual of this size dying
was estimated at 0.003 as described in the methods. Decreasing the probability of death

of trees in this largest size class (stage 8) did not qualitatively alter estimates of the finite

51

rate of population increase (the greatest observed increase in it was 0.3 percent) and did
not affect the repeated measures analysis (results not presented). Further, increasing the
transition probability for stage 2 to stage 4 did not change the relative rankings of
populations (a maximum increase in A of 2 percent was observed) and again did not affect
the repeated measures analysis (results not presented).
Size Distributions
In five out of six sites, observed population structures differed within a site

between 1996 and 1997 (Table 3-5; Figure 3-3). In contrast, four out of the six sites did
not differ for observed population structures between 1997 and 1998. For the healthy
Leelanau site, the observed population structure was not significantly different among
years. At four of the six sites in 1996-1998, stage 6 individuals were found in the highest
frequency. At one of the other two sites, Missaukee Healthy, stage 3 individuals were
found in the greatest proportion across the three years. At the County Line site, a
different stage class had the highest frequency in each year (stage 6 in 1996, stage 3 in
1997, and stage 2 in 1998). In three out of the four populations where C. parasitica was
present, the proportion of individuals in stage 7 was greater than that in stage 8. In
contrast, for healthy populations, there was a higher frequency of individuals in the
largest size class (stage 8) than in the next largest class (stage 7). This pattern was also
seen at the Missaukee Diseased site (Figure 3-3).

Observed population structures differed from calculated stable stage distributions
for all sites in all years, and calculated stable stage distributions differed significantly

between years (Table 3-5; Figure 3-3). These results indicate that the study populations

52

are not at equilibrium. In four out of six populations, the stage with the highest frequency
was the same for both calculated stable distributions. In all cases, the stable stage with
the highest frequency was either stage 2, 6, or 8. In five out of eight year-by-population
combinations, stage 6 or 8 represents the greatest proportion of the population at a stable
stage distribution in populations with disease. At the recovering County Line site, the
calculated stable distribution for both censuses predicts stage 2 individuals will dominate
the population. In contrast, for the recovering Frankfort population, the largest size class
(stage 8) will dominate the population at a stable distribution. At the non-recovering
Missaukee Disease site, stage 6 trees will constitute the highest proportion of individuals
within the population at a stable stage distribution. This pattern was also the result based
on one census period at the other non-recovering site, Stivers .
Cross Sectional Area
Cross sectional area at breast height differed both with time and among
population types (Table 3-6). Both healthy and recovering populations increased in cross
sectional area with time while the non-recovering populations decreased slightly in size
over time (Figure 3-4). Individuals from healthy populations were larger in size and trees
in non-recovering populations had the smallest cross sectional area.
DISCUSSION
While the significant impact of Cryphonectria parasitica on American chestnut

populations is clear from observations of Eastern hardwood forests, this study suggests
that the negative effects of C. parasitica are not uniform across all chestnut size classes.
Instead there are specific changes in the dynamics of the largest trees within diseased

populations. No retrogressions of trees in size classes 7 or 8 were ever observed in the

53

two healthy populations (Table 3-2a-d), but they were observed in both of the non-
recovering populations (Table 3-2i-l).

However, the size reductions of large trees did not translate into a major decline in
the growth rate of non-recovering chestnut populations (Table 3-3). Two explanations
might account for this lack of effect. First, infections decrease plant size by girdling
branches and main trunks, but only rarely kill trees since the pathogen does not enter and
kill the root system. Once the main trunk is killed, the stump sprouts from the root collar
commonly grow for a few years before succumbing to C. parasitica infections.
Therefore, mortality rates are not appreciably increased in diseased populations and
average survivorship is still very high. The average survivorship sets a minimum value
for population growth estimates (Silvertown et al. 1996). Thus, although growth rates
appear to be only slightly depressed in populations infected with C. parasitica, they are
actually quite close to the minimum possible value set by average survivorship (Table 3-
3).

The other explanation for the lack of depressed population grth rates in non-
recovering populations is that smaller trees (stages 6 and 7) produce burrs, a phenomenon
not observed in healthy populations. This pattern could be the result of selection for
reproduction rather than grth of infected individuals, physiological stress from disease,
or a greater light availability in diseased populations due to increased canopy openings
from the die-back of larger trees. Since reproduction appears to be related to light
availability(Paillet & Rutter 1989), the latter scenario seems most likely.

Recovering populations did not share all the characteristics of healthy

populations, but instead retained some features of diseased populations. For example,

54

retrogressions of stage 7 and 8 trees and reproduction by stage 6 and 7 trees were found in
both recovering populations. The retrogressions of large trees were offset by increased
probabilities of growth. As a consequence, the number of stage 7 and 8 trees increased
steadily over the two years (Frankfort +10; County Line +8), while the number of stage 7
and 8 trees decreased in the two non-recovering populations (Missaukee Diseased -8;
Stivers -6) and increased only slightly in the healthy populations (Missaukee Healthy +2;
Leelanau +3).

Population growth rate estimates fell into a rather narrow range and they did not
differ statistically. However, the ranking of populations within each year was consistent
with the hypotheses that: 1) disease was reducing chestnut population grth rates and 2)
dsRNA was allowing chestnut populations to recover to some degree. The only possible
exception was the ranking for the Stivers non-recovering population in 1997-1998 where
the growth rate estimate was highly dependent on the transition probability from stage 7
to 8. No tree was observed to grow from stage 7 to 8 in either non-recovering population
in either year. The actual probability for this transition is likely to be near zero. Since the
total number of stage 7 trees is low in both populations, this transition had to be
estimated. Initially, I estimated this transition by using the average growth rate across all
six populations (i.e., making the assumption that this transition probability did not differ
across populations). Growth from 1997-1998 in the Stivers population was the second
highest using this estimate. If the true value of this transition is actually close to zero
(i.e., 0.00005), then the grth rate estimate for Stivers was reduced to 0.992, a value

similar to the other non-recovering population for both census periods (Table 3-3).

55

The observed structure of all populations differed significantly from the predicted
stable stage distributions. This result indicates that the environment has not been stable
for a long enough period to attain equilibrium. The expectation of equilibrium has been
challenged recently by a number of researchers (Ehrlén 1995; Pascarella & Horvitz 1998;
Valverde & Silvertown 1998). Temporal changes in the environment may be relatively
predictable for species that exploit temporary environments such as light gaps (Pascarella
& Horvitz 1998; Valverde & Silvertown 1998) or more episodic for environmental
disturbances such as hurricanes (Pascarella & Horvitz 1998) or herbivory (Ehrlén 1995).

There are many explanations for the lack of equilibrium in Michigan chestnut
populations. All populations were established in the last century after the introduction of
chestnuts into Michigan by settlers. When farms were abandoned in the mid to late
1800’s, chestnuts began invading feral lands and have continued to establish naturalized
populations that are just now beginning to contain a significant number of large trees
(Brewer 1995). The introduction of C. parasitica into these chestnut populations altered
the dynamics; the predicted stable stage distribution of non-recovering populations would
be expected to contain relatively few large individuals compared to the distributions
predicted for healthy populations (Figure 3-3). When dsRNAs invade the C. parasitica
pathogen population, they further alter chestnut population dynamics so that the expected
stable stage distribution will contain a larger percentage of stage 7 and 8 individuals. The
net effect is that most chestnut populations have not experienced a stable environment
through time, and the observed population structure deviates quite significantly from the

expected stable stage distribution based on the current environment.

56

It has been suggested that pathogens can affect many aspects of the ecological and
evolutionary dynamics of plant populations, including alterations in p0pulation size
(Burdon 1987). However, very few studies have documented that disease is actually
reducing population growth rates (Carlsson & Elmqvist 1992; Antonovics et al. 1994;
Emery 1998). The results of this study indicate that C. parasitica epidemics do have the
potential to reduce growth rates within infected chestnut populations. This reduction
leads to a gradual decrease in chestnut population size. In addition, infections result in
significant size reductions of surviving individuals (Figure 3-4). For chestnuts, this has
resulted in a change in the species’ place in the community; it is now restricted to the
forest understory (Keever 1953; Russell 1987; Stephenson et al. 1991) while trees in
healthy populations are a major component of the forest overstory.

DsRN A hyperparaSites have been suggested as potential biocontrol agents for
chestnut blight by reducing pathogen virulence (Van Alfen et al. 1975; MacDonald &
Fulbright 1991; Nuss 1992). When dsRNAs do spread within the pathogen population,
extant trees recover in the sense that they begin to attain larger size. The hypothesis that
recovery of extant trees translates into increased population growth rates was supported
by the results of this study. Indeed, recovering populations exhibit increased growth rates
that approach the rates found in healthy populations. The following scenario is proposed
for C. dentata populations in Michigan: Healthy populations have growth rates that are at
or slightly above one, i.e., they are near equilibrium (Figure 3-5). When C. parasitica
epidemics begin, the grth rate of chestnut populations gradually declines to near the

minimum value as the larger trees retrogress but survive and seed production declines to

57

near zero. Grth rates remain near the minimum A until the dsRNA hyperparasite
successfully spreads through the pathogen population. DsRNA spread allows extant trees
to increase in size until reproduction again becomes significant. It is not certain that
recovering populations attain population growth rates that are similar to rates found in
healthy populations over a range of temporal and spatial conditions. It is also unclear
whether dsRNAs will provide long-term biological control of the C. parasitica pathogen
population, since selection may favor less debilitating forms of dsRNA (Taylor et a1.
1998). The effects of these different forms of dsRNA on American chestnut population

processes is unknown.

58

 

038583: 33:58: cm A 2: A w
3:26:50: 33:58: cm W :5.“ o_ A so. A N.
03:03:83: $3358: 3 W :5 _ A 2: A o
8:53: _ .v. 2: A m
8.225.. :95 8. w Ea on A a.
vowmfimc 235:0: :o 8:85 omlv. m
$568.: :02: :o :8» “6::on omlw m
38:38 :3.» EE omW ~
bomofio A83 :8: A88 5303 owflm

 

«Em

.3585 585:3. mo gown—smog 08:08: 8 tom: momma—o owmfi ._.m 038.

59

Table 3-2. Transition matrices for six American chestnut populations. Stages represent
the following size classes: Stage 1, first year seedlings; Stage 2, Second year or older
seedlings, 0-50 cm tall; Stage 3, Others, 0-50 cm tall; Stage 4, 50.1-100 cm tall; Stage 5,
greater than 1.0 m in height and dbh less than 1 cm; Stage 6, l.l-10.0 cm dbh; Stage 7,
10.1-20.0 cm dbh; and Stage 8, > 20.0 cm dbh. N is the number of individuals in a stage
in at the beginning of a census period. Probabilities in italics represent survival or
remaining in a given stage from year to year. Probabilities below the diagonal represent
growth and probabilities above the diagonal represent reductions in size. Probabilities
within a column do not always sum to one due to mortality of individuals within a stage.

Transition probabilities followed by an * are estimates.

60

TABLE 3-2a. Transition matrix for Missaukee Healthy 1996-1997.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

STAGE STAGE IN 1996
IN 1997 l 2 3 4 5 6 7 8

l 2.25

2 0.47 0.87*

3 0. 71 0.17* 0.003* 0.0007*

4 0.01* 0.06* 0.62 0.097 0.0693

5 0.21* 0.87 0.02

6 0.03 0.90

7 0.01* 0.95

8 0.05* 0. 997*

N 106 46 315 19 144 101 10 16

TABLE 3-2b. Transition matrix for Missaukee Healthy 1997-1998.
STAGE STAGE IN 1997
IN 1998 1 2 3 4 5 6 7 8

1 0.25
2 0.39 0. 79
3 0. 79 0.21 0.06 0.05
4 0.01* 0.02 0.46 0.06
5 0.29 0.90 0.02
6 0.01 0.85
7 0.01 0.80
8 0.20 0. 997*
N 36 50 355 45 125 95 10 16

 

6l

 

 

TABLE 3-2c. Transition matrix Leelanau 1996-1997.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

STAGE STAGE IN 1996
IN 1997 1 2 3 4 5 6 7 8
1 0.52
2 0.38 0.8 7*
3 0. 75 0.13* 0.005* 0.0002*
4 0.01* 0.06 0.47 0.155 0.0198
5 0.20 0.57
6 0.24 0. 95
7 0.02 0. 94
8 0.06 0. 997*
N 16 6 64 22 37 132 17 27
TABLE 3-2d. Transition matrix for Leelanau 1997-1998.
STAGE STAGE IN 1997
IN 1998 1 2 3 4 5 6 7 8

1 0.93

2 0.57 0.99

3 0.69 0.33 0.02

4 0.01* 0.06* 0.50 0.04 0.01

5 0.17 0.91 0.04

6 0.04 0. 92

7 0.01 0.94

8 0.06 0. 997*

N 14 6 65 3O 26 133 18 28

 

62

 

 

TABl

_mfl C KFCL:

TAB

fisD:Lh\\\\\\\

TABLE 3-2e. Transition matrix for Frankfort 1996-1997.

 

 

 

 

 

 

 

 

 

 

 

STAGE STAGE IN 1996
IN 1997 1 2 3 4 5 6 7 8
1 0.004 0.13 0.17
2 0.62 0.87*
3 0. 71 0.12 0.0057* 0.0005*
4 0.01* 0.04 0.62 0.1843 0.0495
5 0.04 0. 69 0.06 0.03
6 0.08 0.86 0.07
7 0.004 0.03 0.87
8 0.03 0. 997*
N 58 39 209 58 263 388 72 17

 

 

 

 

 

 

 

 

 

 

 

TABLE 3-2f. Transition matrix for Frankfort 1997-1998.

 

 

 

 

 

 

 

 

 

 

 

STAGE STAGE IN 1997

IN 1998 1 2 3 4 5 6 7 8
l 0.27 0.06
2 0.86 0.60
3 0. 71 0.22 0.05 0.02
4 0.01* 0.06 0.41 0.06 0.02
5 0.14 0. 78 0.05 0.01
6 0.02 0.09 0.87 0.03
7 0.02 0.91
8 0.05* 0.997*
N 14 36 241 125 206 359 76 20

 

 

 

 

 

 

 

 

 

 

 

63

IA

TA.

hhhhhh

TABLE 3-2g. Transition matrix for County Line 1996-1997.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

STAGE STAGE IN 1996
IN 1997 1 2 3 4 5 6 7 8
1 0.001 0.04 1.08
2 0.88 0.87*
3 0.83 0.17* 0.0075* 0.0001*
4 0.01* 0.06 0. 62 0.2425 0.0099
5 0.21* 0. 73 0.04
6 0.02 0. 91 0.04
7 0.04 0.89 0.12
8 0.07 0.87 7*
N 166 87 138 27 64 181 102 69
TABLE 3-2h. Transition matrix for County Line 1997-1998.
STAGE STAGE IN 1997
IN 1998 1 2 3 4 5 6 7 8
1 0.001 0.15 1.13
2 0.74 0. 94
3 0. 91 0.06
4 0.01* 0.07 0.67 0.02
5 0.27 0.89 0.01
6 0.09 0.97
7 0.02 0.92 0.01
8 0.08 0. 987*
N 78 146 208 62 53 169 107 68

 

 

 

 

 

 

 

 

 

 

 

TABLE 3-2i. Transition matrix for Stivers 1996-1997.

 

 

 

 

 

 

 

 

 

 

 

STAGE STAGE IN 1996

IN 1997 1 2 3 4 5 6 7 8
1 0.007 0.16 1.13
2 0.71 0.8 7*
3 0. 71 0.15* 0.0015* 0.0001*
4 0.01* 0.21 0.53 0.0485 0.0099
5 0.26 0. 73 0.09
6 0.20 0.88 0.55 0.40
7 0.002 0.40
8 0.003 0.05* 0.597*
N 14 10 125 61 440 637 11 5

 

 

 

 

 

 

 

 

 

 

 

TABLE 3-2j. Transition matrix for Stivers 1997-1998.

 

 

 

 

 

 

 

 

 

 

 

STAGE STAGE IN 1997

IN 1998 1 2 3 4 5 6 7 8
1 0.001 0.09 0.58
2 0.83 0.99
3 0. 73 0.02 0.02 0.002
4 0.01* 0.17 0.45 0.05 0.01
5 0.33 0.65 0.08
6 0.02 0.25 0.90 0.17
7 0.01* 0.78
8 0.05* 0.997*
N 12 10 103 103 394 664 6 5

 

 

 

 

 

 

 

 

 

 

 

65

TABLE 3-2k. Transition matrix for Missaukee Diseased 1996-1997.

 

 

 

 

 

 

 

 

 

 

 

STAGE STAGE IN 1996

IN 1997 1 2 3 4 5 6 7 8
1 0.004 0.36
2 0.65 0. 8 7*
3 0.82 0.18* 0.0027* 0.0001 *
4 0.01* 0.11 0. 64 0.0873 0.0099 0.09
5 0.12 0. 79 0.01 0.14
6 0.11 0. 94 0.08
7 0.03 0. 72
8 0.05* 0. 91 7*
N 68 30 59 54 1 1 1 1 3 5 22 39

 

 

 

 

 

 

 

 

 

 

 

TABLE 3-21. Transition matrix for Missaukee Diseased 1997-1998.

 

STAGE STAGE IN 1997

 

IN 1998 l 2 3 4 5 6 7 8

 

0.02 0.15

 

0.57 0. 90

 

0. 83 0.07

 

0.10 0.09 0.53 0.03 0.01

 

0.30 0. 88 0.06 0.05 0.03

 

0.09 0. 89 0.19 0.03

 

 

 

Zooqmmhww...

0.01 0. 71
0.01 0.05* 0. 93 7*
14 42 78 71 99 141 21 36

 

 

 

 

 

 

 

 

 

 

 

66

 

5m 266 new mad
w.m 036 N36 m.m omod wad
ed wood : mwod
_.v wood 3c; 2 :65 mag
w.m mood _NA: fin $06 35.0
ed 206 hood 5o ooad wood
Wm ~36 woo; wd cad god
0.3 mmwd Sod m. _N mmwd coo.—
EEEEE Esp—EMS
o>onm o>ona

owficoeom 83552 “62,330

”@8580.“ 83552 Bingo

 

 

M32418—

moo _ .mai

@2590 ”dogma—db wnhv
God E0329: w-D 3335 ooxsmmmmE
Amooood “:2sz wfiv
God ”comzmcfiz ”.3 @535

 

OZEm>OUmMTZOZ

on: .5550
tot—5H...—

 

O§m>00mm

3:284
.3283 0838mm:

 

>EHA<mI

.mcocflsmom 353:0 gotofi< 8m 3.38.: we 88 BEE .m-m 2an

67

Time

11mg

Table 3-4. Repeated measures analysis of finite rate of population increase for American

chestnut populations. F-values are for Roy's greatest root.

 

F Num df Den df P
Time 5.1579 1 3 0.1078
Time*Disease 0.9989 2 3 0.4651

68

Sod v m .12.. .36 V m ...

1.1.3.5 1.1.3.3. 31.003 3.5500 nod 11.8.3. Q2
1.33de 11.3.03— ...Immdn *Imodm 1.0—.3 1.3.: km
1.36.02 11.8.3: ***mm.mvw_ isvoém wow ***\.v.vo mm
*Iboém ***No._mm iswodom 2.18.3 Vmfi 3.1.8.5. 40
11.3.8— ...Imwdmm 32¢.ch 3.8 ova omd 1—

simwfiom ***mm.w_m_ 3.1.3.02. :..._N.om_ 1.56.8 3.59% 22

 

$3.82 038m $3-30— oESm moo—-002 033m wag 32030 wag 32030 503 32030

.m> .m> .m> .m> .w> .m>

32.33 283m woo. 33830 83 32630 33 33030 33 @3530 82 votomno

.3585
58:33. («o macaw—smog xmm com gown—33% owfim 038m cog—=28 98 $5836 cons—23m wotomno mo mcoflfinEoU .m-m 2an

69

1201:

Arne:

Table 3-6. Repeated measures analysis of cross sectional area at breast height for

American chestnut populations. F-values are for Roy’s greatest root.

 

F Num df Den df P
Time 254.0145 2 2 0.0039
Time*Disease 493.5203 2 3 0.0001

70

 

tHea
o Rec
V No

'A' Healthy population
+ Recovering population
V Nonrecovering population

LEELANAU

FRANKFORT 1-

COUNTY LINE + *7 MISSAUKEE
' V

STIVERS

Figure 3-1. Map of American chestnut populations used in this study. Recovering
populations are infected predominantly with dsRNA-containing chestnut blight (reduced

virulence) and non-recovering populations are infected predominantly with dsRNA-free

pathogens (virulent).

7l

Figure 3-2. Life-cycle for a recovering population of American chestnut and its

correspondence with the basic population projection matrix (A).

72

.on xEmE w:€:oq$t8 m: 98 ~836wa some

5953 5:858 2.: 38me 886— 28 .mowflm 5253 83:3“: 07:33 05 26% $588 .mommflo omSm 882%: 3.85 .m

 

73

.Amv .5333 98 ADV 538m A”: .3658» 68032 3582.... x332 .n

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

am go o o c o o o w
am am .6 o o o o o a
o gm 3m .6 o o o o c
o o am am .6 o o c m
C 0 3m 3m 3m 90 New C V
o o gm am mm mm o o m
o o o o o o am .6 m
2a 2a 2a o o o o o _
w a e n v m N _ N + a as: 3
~ 2:: an owfim owfim

 

 

 

74

 

Figure 3-.
distfibutéc

POPUIaIiO

Figure 3-3. Population structures observed in 1997 and 1998 and calculated stable stage
distributions for the census periods 1996-1997 and 1997-1998 for (a) a healthy

population, (b) a recovering population, and (c) a non-recovering population.

75

a...» 3» 1b flb . «w. 1M 0.5 a... —

swam—ova A3

3.3.. E...“— o>..¢_o¢ .3533...“— 9583—

1
.
Q
1"
.3;
lb—

 

_ 93%

u a .95
n ouaam

 

v gnaw
m 095
a 33m
5 ouaam

 

4 a 38m

 

839...... 33:533.. 9. 39?... 5.2.5.533
net-=32— uuu: 03a: 5.3-=52. on... ~38.
.853: 33 33-52 10:93.. 32 33.33

76

chaflm «En—ou— huaoacouh 03...:—
31 cal J. N.» 9 3. am a.» a.» L m .3 e... 1.. «.m. a «H g i. a...
_

‘ ‘ ‘ — ‘

m

1 ‘

       

 

 

9:52....” cecaebae 232...: 55.5339
5:32.2— ousu “.33... Eta—.32— 09.» 933».
.8203: :2 03:52 .8293: 28— 53.6%—

8: 9:8 as

l—

. ousm
_ N 35m
m .35
v .35
m 99.5
e .95
h «93
a 09.5

77

mucus—3....— 9523—

»oausuoum 25:3—

»... m6 ‘9

NP

vow—“85 ooxaammME A8

 

 

 

 

 

 

w a... Q g F a a... El m.» S.
9:525». 5322.»...
coca—2.2. can: 033...

Err—«3o waa— 33-3:

9.52....»
saga—=9:—

coZouao baa—

 

 

 

"... a. Q a... _
~ ouaum
N ouauw
n .95
v .95
m .25
J a ouaum
a. .95
a .95
5.35.5...
on!» 938»
$3.93—

78

IN"!

bus—u v=r=hb v—cvhh.’ =95: .an-C‘ubfin O'CLD :30:

.83 95:8 mama?» Amctgooofico:
$5852: $5185 093 sown—anon some 5 moo: 853:0 sucrose «0 echo Havana H 983 was 38308 $20 :82 .Ym oSwE

 

 

 

 

 

 

..¢o>
«as. bag. was.
u .1 . c
H m an
2:
W
a
qum>cum¢iozlll .—. .— m
ozEm>cum¢Iol .— 3. m
>=S<m=lol H m
u.
2: m
H
u
:3"
:-
.Is...
:3».

 

Qua—u mama—00 mzmh?’ 89—6 —fl:0_a00m macho EGG:

79

.muouflaoa 353:0 snow—0:2 8% mouse 532w scam—30a .«o 08:8 2:: Recon—8&2 .m-m 83me

«EC. & 83552

 

 

e334
acts—sac.—
oufleofiaoua— Howe—.22—
nuouao <Z-mc
3:33.:
:5— auguuuum .U
j

\

ma...
mnaé

cc.—

mouam 539.0 acts—=93— ...Emosu .3 3.250 0.5%

80

 

DOl

6:010;
Ina-"lag;

ethos
histor}
bCCD Ll)
should :
Species

 

 

CHAPTER 4
DOUBLE-STRANDED RNA MEDIATED RECOVERY OF THE AMERICAN

CHESTNUT: WHAT CAN WE LEARN FROM HOST DEMOGRAPHY?

INTRODUCTION

Information on demographic parameters not only provide insights into the
ecological and evolutionary dynamics of populations but also allows for effective
management of rare and endangered species. Matrix population models may be the best
method for making predictions about population dynamics and identifying critical life
history stages of threatened species (Schemske et a1. 1994). Indeed, these models have
been used to evaluate population viability and to determine which life history stage
should be the focus of management for a variety of endangered and rare plant and animal
species (e. g. Crouse et al. 1987; Lande 1988; Menges 1990; Byers & Meagher 1997;
Kephart & Paladino 1997).

Although the finite rate of population increase (A) is an important demographic
parameter which indicates whether a population is increasing or decreasing in size
through time, other demographic parameters that indicate which particular stage or age
class contributes most to population growth rate are more useful for management
decisions. Matrix models are constructed by assigning individuals to age or stage classes
and then determining the probability of an individual making a transition to another class
(Caswell 1989). Sensitivity values measure the relative effect of small changes in

transition probabilities on A. and elasticities represent the proportional contribution to A.

80

for each transition probability (Caswell 1989; Silvertown & Lovett Doust 1993). These
parameters may be used to make recommendations about which life history stages of
endangered or rare species to protect or manage (e. g. Crouse et al. 1987; Lande 198 8;
Kephart & Paladino 1997). Further, Silvertown et al. (1996) propose using elasticities to
examine particular regions (growth, survival, and fecundity) of the transition matrix and
their impact on population growth rates as a means of identifying populations that differ
from the pattern normally found in a species.

Sensitivity and elasticity values could also be used to develop disease
management programs. This approach may be particularly useful in evaluating the
effectiveness of potential biological control agents of pathogens. By comparing these
demographic parameters among healthy populations, diseased populations, and
populations where biological control has been attempted, the success of treatment can be
evaluated. Further, which particular life history stage of the plant should be treated to
have the greatest impact on population growth rates can be determined.

Based on estimates of the finite rate of population increase, the introduction of
double-stranded RNA (hereafter, dsRNA) into populations of Cryphonectria parasitica,
the pathogenic fimgus causing chestnut blight, may promote the recovery of American
chestnut populations (see Chapter 3). Examination of other demographic parameters,
specifically elasticities and sensitivities, may provide an alternate means of evaluating the
effectiveness of dsRNA as a treatment for chestnut blight and provide insight into how
best to deploy this management strategy.

The specific goals of this study were to (1) determine which size class would have

the biggest impact on population growth as measured by sensitivity analysis thereby

81

providing an indication of where management efforts should be concentrated; and (2)
compare elasticities among American chestnut populations to evaluate the effectiveness
of dsRNA mediated recovery; Do recovering populations have the same patterns of
growth, survival, and fecundity as healthy populations or as non-recovering populations?
METHODS
Study Svstem andS_tud_y_S_it_e§
For a description of the interaction between the American chestnut (Castanea
dentata), chestnut blight, and dsRNA, see Chapters 2 and 3. Study sites utilized in this
project are detailed in Chapter 3. Estimates of population growth rates revealed that
recovering populations typically exhibit rates of growth intermediate between healthy and
non-recovering populations (see Table 3-3).
Population Projection Matrices
Stage classes and projection matrices-Construction of size classes and a description of
the matrix model are presented in Chapter 3.
Data analyses-Sensitivity and elasticity values were calculated for each population using
RAMAS/stage (Ferson 1994). Sensitivity values (3,} = 51/809) measure the effect of a
small change in a transition probability on A relative to the same magnitude of change in
other probabilities (Caswell 1989; Silvertown & Lovett Doust 1993; Valverde &
Silvertown 1998). Elasticity values (eij = aij/k X 6N8aij) represent the proportional
contribution to l for each transition probability (Caswell 1989; Silvertown & Lovett

Doust 1993). Within a given transition matrix, elasticities sum to one which allows

82

comparisons among populations to be made (de Kroon et al. 1986, Caswell 1989;
Silvertown et al. 1996).

Silvertown et al. (1993) divide elasticity matrices into three regions and sum
elasticities within each region. The G region (elements below the main diagonal of the
transition matrix) represents the effect of growth on A, the L region (non-fecundity
elements above and on the main diagonal) measures the impact of survival within a stage
and retrogression to a smaller stage, and the F region represents the effect of reproduction
(top row of matrix). The G/L/F ratio for each population can be plotted on demographic
triangles to investigate their relative positions and potential trajectories.

Some biologically important transitions were not observed in all populations and
therefore were estimated as described in Chapter 3. To examine how these estimated
transition probabilities affected sensitivities and elasticities, simulations were performed
in which the transition probabilities were altered and the effects on sensitivities and
elasticities were investigated. For the transition from stage 2 to stage 4, the transition
probability was increased by 1000 percent to match the highest transition probability
observed within a population (see Chapter 3). For the transition of stage 7 to stage 8 in
non-recovering populations and for death in stage 8, the effect of reductions in these
probabilities by mom was investigated. This transition in non-recovering populations
and death in stage 8 were not observed and were estimated (see Chapter 3). Since these
values were essentially observed to be zero, the effects of reducing the estimated

probabilities to near zero was investigated.

83

RESULTS
Sen—shiv—itigé

Sensitivity analysis describes how small perturbations in transition probabilities
would impact the finite rate of population increase (7»); a large sensitivity for a transition
indicates a small change in that transition would have a relatively large effect on k
(Silvertown & Lovett Doust 1993). The highest sensitivity values within a stage were
often the transition representing growth to the next stage (Table 4-1a-l). Reproduction
generally had the lowest sensitivity values (< 0.005). The only exceptions to this pattern
were in the non-recovering populations where reproduction was relatively more important
for stage 6 trees at Stivers for both the 1996-1997 and 1997-1998 censuses and for the
stage 8 trees in the Missaukee Diseased site for 1997-1998. For diseased populations,
the highest sensitivities were for a stage 6 transition, either growth to a larger stage (7 or
8) or survival in stage 6. In the recovering County Line population, sensitivity for grth
(from stage 2 to 4 or from stage 7 to 8) was most important. However, for the other
recovering site, Frankfort, and both healthy sites, the highest sensitivity overall was for
survival in the largest size class (stage 8). For the Frankfort site in both census periods
and for 1997-1998 at the Missuakee Healthy site, this transition overwhelmingly had the
highest potential impact on population growth rate (sensitivity > 0.999).

Sensitivity values in the non-recovering populations indicated that stage 6 trees
were very important, since this stage had the top five sensitivity values found in the

matrix. The only exception was Stivers in 1997-1998, where the highest value was for

84

growth from stage 6 to 7 but the growth transition from stage 2 to 4 was second highest
and survival of stage 8 trees was third highest.

Decreasing the death rate by a factor of 1/1000 in stage 8 did not alter the general
pattern of sensitivities for any population in any census period and had only a minor
impact of the magnitude of sensitivities (data not shown). Changing these transitions did
not materially change the pattern or magnitude of sensitivities for most populations. In
those populations where changes occurred, the sensitivity values for some of the growth
and fecundity transitions increased in value; however, there was no consistent pattern in
which specific transitions changed in magnitude.

m

Elasticity analysis describes the proportional impact of a transition on the finite
rate of population increase (Silvertown & Lovett Doust 1993). Across all populations,
survival within a stage (diagonal elements of the matrices) had the greatest contribution
to A (Table 4-2a-l). However, which particular stage was most important differed among
populations and among census periods. For the 1996-1997 census period, one of the
larger size classes (stages 6 to 8) had the greatest elasticity value within each population.
For both healthy populations and overwhelmingly for the recovering Frankfort site,
survival in the largest size class (stage 8) had the largest proportional impact on
population growth rate. At the remaining recovering site, County Line, survival in stage
7 was most important. In contrast, for the non-recovering populations, survival in stage 6
had the highest elasticity. For the 1997-1998 census period, the elasticity for survival in

the largest size class was greatest for five of the six populations. The one exception was

85

the non-recovering Missaukee Diseased site where again survival in stage 6 had the
greatest proportional contribution to A.

The relative contribution to A of survivorship (L), growth (G), and fecundity (F)
can be assessed by summing elasticities across all stages (Silvertown et al. 1993).
Survivorship made the greatest contribution to it in all populations across both years,
while fecundity never contributed more than 0.01 for any population-by-year
combination (Table 4-3). Growth was relatively more important in the chestnut
populations where C. parasitica was present. The only exception to this pattern was at
Frankfort where grth elasticity values, based on observed transitions, were extremely
low. Indeed, growth values for Frankfort were one to two orders of magnitude lower than
growth values for any other population.

A possible explanation for the extremely low values for growth at Frankfort may
be associated with the lack of observed retrogressions of stage 8 trees (see Table 3-2e-f).
Stage 8 retrogressions were observed in all other chestnut populations where C.
parasitica was present. The absence of retrogressions at Frankfort is most likely due to
the small number of stage 8 trees (17 in 1996 and 20 in 1997), which meant that
retrogressions might not have been observed by chance alone. If the stage 8 retrogression
values are estimated from the average stage 8 retrogressions observed in the remaining 5
populations, then grth elasticity values increase by more than 150 fold and the G/L/F
ratio becomes very similar to those observed in the two healthy populations (Table 4-3).

Silvertown et al. (1993) included retrogressions in the L region. Because

retrogression appears to be a characteristic of diseased populations (see Chapter 3), I

86

examined the sum of the elasticities for retrogressions alone. In populations with disease,
there is an order of magnitude difference in the corresponding elasticities for populations
with disease versus healthy populations (Table 44). Again, the recovering Frankfort
population is an exception to this pattern.

All populations matrices contained one or more transition values that were not
observed over the two years of the study, but were estimated because they must occur for
biological reality. For example, no stage 8 tree died in any population over the two
census periods, yet it is reasonable to assmne that these trees do occasionally die. The
mortality rate for stage 8 trees was estimated by assuming that the next sampled stage 8
tree would have died during the study. This “death” was added to the existing stage 8
trees and a grand mortality rate (0.003) was calculated across all populations and used in
the matrix for each individual population. It is possible that the mortality rate of stage 8
trees is significantly lower than this value. Reanalyzing the matrices with a lower stage 8
mortality (0.000003) resulted in only slight increases in the contribution of survivorship
(L) to 3. (data not shown). Growth from stage 2 to stage 4 was observed only at the
Missaukee Diseased site for the 1997-1998 census. In the other matrices, this transition
was estimated as the grand average across all population by year combinations (0.01).
Increasing the stage 2 to 4 growth rate to the value observed at the Missaukee Diseased
site from 1997-1998 resulted in a slight increase in the importance of growth (G) to A
(data not shown).

Finally, growth from stage 7 to 8 was not observed in either of the non-recovering

populations. As a conservative estimate of this transition, the average stage 7 to 8

87

transition (0.05) observed across the other four populations was used. However, the
presence of C. parasitica in these non-recovering populations may actually reduce this
transition to near zero. Reducing this transition in non-recovering populations resulted in
only slight increases in the importance of survivorship (L) to it in these populations (data
not shown).
DISCUSSION

Elasticity values and the sensitivity analyses indicate that survival of the largest
trees has the greatest impact on population growth in both healthy populations. This
result is not surprising given that chestnuts are normally long-lived trees. Indeed, other
studies of woody plants have revealed similar patterns of population growth being most
sensitive to survivorship of large individuals (e.g. Pifiero et al. 1984; Huenneke & Marks
1987). The blight pathogen, C. parasitica, affects demographic patterns within infected
chestnut populations. The most noticeable changes are seen in the largest stages of these
diseased, non-recovering populations. Transitions that were not observed in healthy
populations had a significant impact within diseased populations. Stage 7 and 8 trees
retrogressed to smaller stages, and stage 6 and 7 trees were observed to set seed (Chapter
3). In addition, the growth transition from stage 7 to 8 was not observed in either of the
diseased population. These changes resulted in a slight decrease in population growth
rate so that X values were near the minimum possible for a population, given its survival
schedule (see Table 3-3). Sensitivity and elasticity values also indicated a major change
in population demographics. The most striking change was seen for the stage 8 trees at

Missaukee Diseased and at Stivers in 1996-97 where stage 8 trees were no longer having

88

the largest effect on population growth (Tables 4-1i,k & l and 4-2i,k & 1). At these
diseased sites, stage 6 trees were most important. These results are in general accord with
known dynamics of C. parasitica epidemics, in which infections rarely kill trees, but
instead cause a reduction in size. This trend has also been found in diseased chestnut
populations within the natural range where large trees are extremely rare, but chestnuts
are still a major component of the forest understory (Stephenson et al. 1991; Parker et al.
1993)

Silvertown et al. (1996) proposed that a graphical presentation of elasticity values
for growth (G), survivorship (L) and fecundity (F) can be used to identify populations at
risk. This concept was proposed as an aid for plant population conservation; populations
with altered G/L/F ratios normally have reduced population growth rates (Silvertown et
al. 1996). When this ratio is combined with sensitivity analyses decisions can be made to
determine which growth stages should have highest priority with regard to conservation
efforts. Demographic data have been used to identify unique features of rare species
(Byers & Meagher 1997) and aid conservation efforts (Crouse et al. 1987).

One major objective of this study was to determine whether G/L/F ratios could be
used in a more general context to highlight altered population demographics. There was
no clear separation between healthy, non-recovering and recovering populations when
they were graphically plotted for G/L/F ratios and population grth (3.) (Figure 4-1).
The lack of any clear separation can be explained by the fact that infections do not
materially increase tree mortality within infected populations and survivorship (L)

remains unchanged because of the way it has been traditionally calculated. The L

89

component of the G/L/F ratios includes retrogressions (Silvertown et al. 1993).
Retrogressions are uncommon for woody plants, and thus including retrogressions into L
does not normally influence the G/L/F ratios. As discussed above, the increase in
retrogressions of large trees within diseased chestnut population is one of the major
effects of C. parasitica epidemics. Thus, one problem of this graphical analysis is the
fact that potentially important alterations in demographics can be hidden within the L
variable.

It is doubtful that diseased populations would appear to be aberrant even if
retrogressions were discounted from the calculation of L. Retrogressions contribute only
between 1 and 8% to population grth (Table 4-4). Further, sensitivity values are
usually low for retrogressions, except for those involving stage 6 trees (Table 4-1i-l). The
inability of this graphical method to detect altered demographics is due to the long-lived
nature of chestnuts. The G/L/F ratios for all long-lived plants are always overwhelmingly
dominated by L. A survey of 21 woody plant species found that L contributed greater
than 78% to the population growth rate (Silvertown et al. 1993). As a result, long-lived
perennials are constrained to a very restricted corner of the G/L/F space unless the factors
altering demographics cause large increases in mortality. This study demonstrates that
significant changes in the demographics of long-lived species can occur without
increasing mortality. The net effect is that the G/L/F graphical method should be used
with caution. While the method may be useful for a generalized evaluation of some
species, it will not always detect significant changes in population demographics.

This study also sought to evaluate the extent of recovery in chestnut populations

where the dsRNA hyperparasite has invaded the C. parasitica pathogen population.

90

Recovery, as indicated by the presence of non-lethal cankers, was first noted in the 19705
at the County Line and Frankfort sites (Brewer 1995). A recent survey of C. parasitica
from County Line and Frankfort found that dsRNA had successfully spread to the
majority of the pathogen population with 89% and 94%, respectively, of the sampled
cankers containing dsRNA (Davelos et al. 1997). Population growth rates of the
recovering populations tend to be similar to those of healthy populations (Chapter 3,
Table 3-3). Recovering populations also retain a number of transitions that are found in
diseased populations. For example, retrogressions of stage 7 and 8 trees are found in
non-recovering and recovering populations, but not healthy populations. In addition,
stage 6 and 7 trees were observed to reproduce at recovering and non-recovering sites, but
not at healthy sites. Recovery seems to be due to the reduced magnitude of retrogressions
and increased growth of large trees at recovering sites.

The G/L/F ratios were again inconclusive for the recovering populations. Ratios
for County Line were very similar to those ratios found at the non-recovering sites, and
the Frankfort site had ratios that were unique but closest to the healthy sites (Table 3-3).
If estimates for stage 8 retrogressions were used at Frankfort, then Frankfort's G/L/F
ratios became very similar to ratios found at the healthy sites. A detailed examination of
the elasticity and sensitivity matrices indicated that the recovering sites had largely
recaptured much of the important dynamics found at healthy sites. Similar to healthy
sites, the highest elasticity and sensitivity values were found for survivorship of stage 8
trees. These data indicate that the dsRNA hyperparasite can help to mediate recovery of

chestnut populations.

91

The pattern of sensitivity and elasticity values at the non-recovering sites can also
be used to aid efforts to introduce dsRNAs for the purpose of saving chestnut populations
(MacDonald & Fulbright 1991). The high sensitivity and elasticity values for stage 6
trees argue that these trees will have the most immediate effect on population growth, and
initial efforts to introduce dsRNAs should concentrate on this class of trees. Knowledge
about the contribution of life history stages to population growth and stability is crucial
for species conservation efforts where limited resources need to be utilized effectively
(Crouse et al. 1987). These results emphasize the need for a careful examination and

interpretation of all demographic parameters when developing management strategies.

92

Table 4-1. Sensitivity matrices for six American chestnut populations for two census
periods. A large value indicates a small change in that transition would have a
relatively large effect on 3. (Silvertown & Lovett Doust 1993). Values for survival

within a stage are italicized.

93

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

36.3 s 33$ 6 w

VNVms s wanna... b

.33... s mwcm — ... c

93...... mmwg ... v..mm..... m

we. 2.... 252.... 24%.. 3...... papa... v

.32.... was... 92...... neg. S m

MQ§N§S 33.... N

3.2.... .
w h o w v m N . 33 Z.
coo. Z. mO<Hm mO<Hm

 

 

 

Sam—-93. 3:3: 03.5.3.2 .8. «fizzzmuom .3-.. ”mu—mt...

94

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.9.36 .......... w

3§§S 82...... n

V~§§$ N.N....... o

N........... MN§§§S h........... m

............ V§§§s w........... 83.... v

............. ............ ............. 3§§§6 m

3§§6 m........... N

w........... .
w h c m v m N . waa. Z.
3a. Z. mO<Hm m0<,..m

 

 

 

£23.53 ~3:3... 23.3.3.2 .5. no.._>...m:om 5.-.. HAQ<P

95

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MRS. s «mag... w

§QN§$ 93%.... ..

23%.. 22...... c

has... NN....... m

«3...... ...N....... 3:35 3&8... N. .5... v

3.3.... .m......... 32...... nm¢§§s m

”Nags meg... N

aumec... .
w n o w v m N . 53. Z.
e3. 2. m0<...m mO<...m

 

 

 

SSTEK. aqua—00... .3. noggin—om 6.-.. HA..<...

96

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5&3... $32.... w

3&36 2N3... 5

N23... wean... o

mama... NNVN... Emma... m

VNm...... 3&3... 3N3 S 53...... 3va... v

.mN....... 3.8.... «N 3%... m

«SANS 3.25... N

3.23.... .
m N. o w v m N . .3. Z.
3a. Z. m0<...m m©<...m

 

 

 

£23.58. 323.001. .8. no...>_..m:om 6.-.. HAM—(C.

97

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3...... 2...... w

A ~..... N5... 5

4...... 2.... bN..... .

....... m...... ....... m..... m

....... ....... ....... N...... was... 4

....... ....... ....... ....... m

‘3... m...... N

....... ....... ....... .
w .. o m 4 m N . 3.. Z.
e... Z. mU<...m m0<...m

 

 

 

.b...-.... tax-.2... .3. magisnuom 6.4 mag...

98

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.23.... .m..... a

N3... .4..... ..

....... .N..... .m..... mm..... o

m...... m...... 4...... ....... m

4...... m...... ....... m...... .mN... 4

....... ....... ....... ~...... m

....... m...... N

M4..... ....... .
w h c m 4 m N . w... Z.
3.. Z. mO<Hm m0<...m

 

 

£2.75... 28.8.52”. 8. no...>...m:om ...4 mama...

99

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5...... «.2»... w

Rm... .43.... N.....N.. .

444MN.. V3.4. .. Nwmm... .

m..m... $5.... m.mm... m

.m.N... .N.N... ..NN... ..4N... mm... . .. 4

.45.... 3...... 5...... 4...... m

.u M.. .. 4...... N

m..... .NN.... m4..... .
w n m. m 4 m N . 5... Z.
o... 7.. m0... m©<..m

 

 

 

.b...-.... on... .3580 .8. no...>...n:um .w.4 9:33..

100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

...m.... .N4m... .

hm..N.. REN... .Nm.4.. ..

.M.NN.. 43.... w

.m..... R N.. .. .Nm4... m

.m..... .Nnm... ...N... ...... 4

.43... .6... .. m

8.2.. 4.... N

.N..... mm»... m...... .
. h e m 4 m N . .... Z.
n... Z. m©<.... mO<....

 

 

.....-..... on... 3580 .8. 333.328. A. .4 Nam...

10]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

aw3§$ hangs w

3:36 N35... 5

3:56 2:3... §§V¢$ mmvaé c

m 3am... an? N s cwamcé m

933.: gm — N6 a ~35 S arena... mmemcé v

2.5mm... 23.3.: mvmmeé megS m

wNM§§S $25.: N

2:25.: $25.: 933.: _
w h o m v m N ~ 3&— 7:
33 Z— mO<Hm mO<Hm

 

 

.baa—éaa— 9.035 he motmzzaom .24. HAM—«AH

102

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

33‘s 532.... w

nmwgs vaceé N.

323.: an: s 5:36 3.53.: o

SMN—é MRES was... m

3.39.: N— 28.: 2‘36 @386 $9.33 v

«as... mwagé 2~N§§$ m

3325.6 33:... N

933.: «38.: mvwmcé _
w h o n v m N ~ 33 Z—
32 Z mD<Hm mO<Hm

 

 

 

.waa—-baa— 9.935 .2: angina—Sm .34. HAm—(h

103

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.936 aeN§~$ w

aédSs mNeNmé N.

35%.... NEEeS MNch—é o

ewmmcé aN—mvé 5mm: 6 «N535 m

mm—mcé 3.3.”: 255:... amass 933.: mchmcé v

Nth—.9 abché vw—vcé V3666 m

th§§$ ”was... N

325.: N38... _
w R. c w v m N _ 32 7:
32 E mO<Hm mO<Hm

 

 

 

Saa—éaam 6023an 03—53mm: 58 “Sarina—Sm 4:4. HAM—(uh

104

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

cXNNS «Kags mun—w... w

N336 3:.de h

593.: veN—eé 263.6 bNNmmé o

«3.3.: NmN—cé uNmmmd mgVMS m—cmcé m

unemNé MEN... 3.35 s mam—c... mm 3:... v

353.: Nmn§§s m

3%.is wawccé N

ammmcé vmmecé _
m N. c m V m N _ wag Z—
32 7: mO<Hm “LOAF—km

 

 

 

.maa—ubaa common:— ooxaammm—Z he mogzzmuum ATV magi;

105

Table 4-2. Elasticity matrices for six American chestnut populations for two census
periods. Within a given population for each census period, elasticities sum to 1. Values

for survival within a stage are italicized.

106

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Magus 52.6.6 m

=4is MbN....... n

3366 Nomi... c

«3...... 3.366 mm......... m

32...... 3.3.... §NMSS 52.2.... MbN....... v

............. N........... 32...... aa3§$ m

Rag... m.N....... N

Mchc... .
w h o w v m N . 3o. 2..
coo. Z. m©<hm mO<Hm

 

 

 

$37.53 .3..on 03.53.»..2 .8.. 33.3.35 .aNuv HAM—4x...

107

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

34:36 N§§§§s w

aSSSs N253... h

N3§§ s N923... o

258.: §N§§s 825.: m

253.: N§§§§s 325.: N625... v

25cc... 252.... 253.: N§§§§S m

§§§§ s N259: N

N925... _
m N w m e m N w :3 7:
32 7: mafia mO<Hm

 

 

 

.waa—Afia— >513: 3:33am: .8.. 353.22% fiNé mdmfih

108

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

225.: K 55.: w

3&5: :52: 5

:Nn 9.5.: N:—::.: :

23:2: mNN::.: m

:N:::.: .255: NNN::.: «.55.: :52: v

:::::.: 55:: v—:::.: 35:2: m

3.2:: :52: N

:52: _
w b c w v m N _ 5:2 2—
:aa 7: mO<Hm mOfl—b

 

 

 

.haaué— anus—vou— 58 359635 .oNuv HAM—flu.

109

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

nann‘: :vv::.: w

N525: :ve::.: 5

$55.: m:m::.: :

535.: V: 5.: 5m::.: m

m5::.: 25:: 3.55.: 5:55 345.: v

m::::.: 5:::.: 3.52: m

:MNNNS 3.5.: N

3.95.: _
w b c w v m N _ M5:— 7:
52 7: m0<Hm mO<Hm

 

 

 

£57523 swan—cow— .8: 85333—2 .eNé Gaga.

110

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.25.: N::::.: w

w. 5:: .: N5::.: N

:55: :~:::.: N::::.: :

:::::.: :::::.: w::::.: 55.: m

:::::.: :::::.: m::::.: :::::.: N::::.: v

55:: 5:5 :::::.: :::::.: m

M~::: .: N::::.: N

N::::.: :::::.: :::::.: ~
w h w m v m N _ 53 Z—
022 Z— mO<Hm mO<Hm

 

 

 

.h::~-:::— tau—Er...— 58 3593mm:— .oN-v mun—mgr

ll]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:25: .: 5:::.: m

:N:::.: 5:::.: N.

:::::.: VN:::.: 5:::.: 5:::.: :

:::::.: 5:::.: ::::.: N::::.: m

::::2: :::::.: 5:::.: 5:::.: M52: v

::::2: ::::2: ::5:.: 5:::.: m

w::::.: n::::.: N

5:::.: ::5:.: _
w h c w v m N _ 52 7:
5:— 2: mpg/am mO<Hm

 

 

 

.waauuhaa tau—5...: .3.— 359335: .uNuv Hum—<8

112

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.25.: N55: w

::5:.: w::::.: m: — 5.: 5

95:2: n~:m~.: :52: c

552: :55: 65:2: m

:N:::.: ::5:.: :NVN:.: ::5:.: N555 v

::::2: ::::2: 9.52: mmn::.: m

u:NN:.: N25: N

552: ::::2: 5:::.: :
m N. o w v m N _ 52 7:
0:2 7: mO<Hm mO<Hm

 

 

 

#23::3 95n— bnacU .3: 85933.: .wNuv HAM—4G.

113

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Nfi 3:: ::N5: :

::N::.: .36.: .: 5:5: N.

:N 5.: :35: o

:52: 3.25.: :35: m

::5:.: :VVN:.: :25: 5:::.: v

::5:.: RN22: m

NMNNN: 5:::.: N

:55: Nm:::: 5:::.: _
: h e w v m N _ ::3 7:
5:— 7: 935: Noll:

 

 

 

.:::Tb::— 2.: 5550 .8: 35032:: A54 HAM—flu.

114

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

55:: ::5:.: :

:N:::.: :5: h

::5: :5: 5:3: :55: 0

3:2: ::V:N.: 5:5: :

::5:.: ::5: 5:::.: 2:5: 3:5: v

:5: :N5: 3:5: 255.: m

::5:.: 5:5: N

::5: :5: 55.: _
: N. c n v m N : 5:: Z:
5:— 7: 559: mO<Hm

 

 

 

.h5—éaa E94“: .3: 35032:: .mNé HAM—(F

115

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

36:2: Nu V::.: :

2:5: ::5:.: K.

5:5: NN:V~.: 5:5: 9:52: o

5:::.: :::::.: 555.: m

:5: :52: 5:::.: :::5: 55:: v

:55: 555.: 5.5:: m

N:~:N.: 59.5.: N

Nbv5: :—5: :55: _
: n c w v m N _ :5: Z:
5:— 7: 555: ::5:.:

 

 

 

$83.3. 525 .5. 8:325 .3... 339

116

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

AN:N:.: mam::.: :

V:VV:.: 23:: h

hwvgé 33%.: 33:: o

::b::.: «35:: :N::~ .: cum—:6 m

::N::.: :N::.: :be::.: «#3:: ::b::.: :m:::.: v

N::::.: m—:::.: :58: «636.: m

::V::.: :m:::.: N

:m:::.: —::::.: _
m N. c m v m N ~ 82 7:
0:2 2: maxim m0<Hm

 

 

 

.237:::— 10333— moss-«mam: .3: 35932:”.— .u—Né HAQ<H

ll7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

M365: 3.3:: :5:: w

32:: —:v::.: 5

:MN::: NVN::.: nu: NM: 32:: :

:NN::: m::::: :m _ 8.: 3:3: 33:: m

NmN::.: :33: :::~:: :m~::.: 33:: v

:23: MR::: m

::m: .: :~m::.: N

m:m::: — —:::.: _
m N. o w v m N _ ma: _ Z:
3: _ Z— mO<Hm mO<Hm

 

 

 

.:::—-5::— tomaomma 03—53%: .3.— momzoflma—H ._N-v HAM—«Q

118

Table 4-3. Sums of elasticities within the G (growth), L (survival), and F (fecundity)

regions of transition matrices for six populations of American chestnut.

a. 1996—1997 census

 

 

 

 

Population Growth Survival Fecundity
(G) (L) (F )
Healthy
Missaukee Healthy 0.02296 0.97429 0.00273
Leelanau 0.01 115 0.98713 0.00171
Recovering
Frankfort (Observed) 0.00012 0.99985 0.00002
(With stage 8 retrogressions) 0.02323 0.97664 0.00010
County Line 0.06251 0.93577 0.00172
Non-recoverirg
Stivers 0.08983 0.90978 0.00037
Missaukee Diseased 0.06270 0.93673 0.00059
b. 1997-1998 census
P0pulation Growth Survival F ecundity
(G) (L) (F)
Healthy
Missaukee Healthy 0.00012 0. 99985 0.00002
Leelanau 0.03012 0. 96551 0.00440
Recovering
Frankfort (Observed) 0.00018 0. 99980 0.00003
(With stage 8 retrogressions) 0.02961 0. 97026 0.00014
County Line 0.06550 0. 92478 0.00971
Non-recovering
Stivers 0.04377 0. 95131 0.00491
Missaukee Diseased 0.07307 0.92175 0.00516

ll9

Table 4-4. Sum of the elasticities for retrogressions for six American chestnut

populations for each census period.

 

1996-1997 1997-1998
Missaukee Healthy 0.00575 0.00000
Leelanau 0.00067 0.00349
Frankfort 0.00000 0.00001
County Line 0.05201 0.00778
Stivers 0.08253 0.01390
Missaukee Diseased 0.03786 0.04038

120

 

1.0
0.0

 

0.0

 

1.0 ¢

 

 

 

 

 

Figure 4-1. Plot of G/L/F elasticities for six populations of American chestnut. a. 1996-

1997 census. b. 1997-1998 census.

121

CHAPTER 5
SEEDLING GROWTH AND SURVIVAL‘IN POPULATIONS OF AMERICAN

CHESTNUT THAT DIFFER IN DISEASE STATUS

INTRODUCTION

Pathogen infection has been shown to decrease plant reproduction, whether
measured as flower number, fruit production, seed production, or seed biomass
(Alexander & Burdon 1984; Clay 1984; Parker 1986; Paul & Ayres 1986a; Paul & Ayres
1986b; Parker 1987; Paul & Ayres 1987; Wennstrom & Ericson 1990; Wennstrom &
Ericson 1991; Jarosz & Burdon 1992; Roy & Bierzychudek 1993; Thrall & Jarosz 1994;
Garcia-Guzman et al. 1996; Marr 1997). Clearly reductions in reproduction by seed have
negative fitness consequences for an individual plant. Moreover, this negative effect on
individuals may translate into effects at a population level by decreasing recmitment and
population expansion rates. One extreme example is anther smut of Silene alba.
Infections completely sterilize the host, and no seed is produced (Alexander et al. 1996).
As a consequence, population growth rates are depressed when the smut fungus,
Microbotryum violaceum, is found within a host population (Antonovics et al. 1994). A
similar trend of reduction in the rate of population expansion was observed for
populations of Silene dioica infected by M. violaceum (Carlsson & Elmqvist 1992).

However, pathogens may also cause more subtle changes in reproductive effort
through effects on seed quality, which also affect grth and competitive ability in the
next generation. For example, seed from wheat infected with Erysiphe graminis (a fungal

pathogen that causes a foliar disease and does not infect the seed produced by a plant)

122

The chestnut blight pathogen, Cryphonectria parasitica, clearly has a negative
impact on reproduction of diseased American chestnuts because the main trunk is often
killed as a result of infection. However, when trees are infected with a strain of the
pathogen hyperparasitized by a double-stranded RNA and therefore reduced in virulence
(hypovirulent), infected stems may survive to reproductive size. Whether pathogen
virulence has any prolonged effects on the next generation was examined by studying
populations infected with either virulent or hypovirulent forms of C. parasitica. Survival
and size of naturally recruited seedlings were compared to individuals which were similar
in size to seedlings due to infection or herbivory from healthy populations and infected
chestnut populations. Greenhouse and field experiments investigated whether infection
resulted in any adverse effects in seed size and subsequent seedling development of
American chestnuts (i.e. seed quality). Further, the effects of herbivores on seedling
establishment and growth were investigated with a caging experiment.

METHODS
Study System and Study Sites

For a description of the interaction between the American chestnut (Castanea
dentata), chestnut blight, and dsRNA, see Chapters 2 and 3. The six sites (two healthy,
two non-recovering, and two recovering) utilized in this project are the same as those
used in the demographic study (see Chapter 3 for details).

Natural Recruitment

A series of 9 m X 9 m plots were placed haphazardly with the number of plots

varying from site to site so that approximately one-half of a population’s area was

monitored. Seedlings and other small chestnut plants less than 1 m tall, resprouts and

124

herbivore damaged individuals, were marked and censused repeatedly for survivorship
and growth from June 1996 to October 1998 (see Chapter 3 for details).
Field Exmriments

Bulk seed collections were made at the six study sites in fall 1996. Seeds were
stored overwinter at 4 °C. In April 1997, two treatments were initiated at all sites. One
treatment consisted of planting 45 seeds in four unprotected (open) plots placed
haphazardly at each site. The other treatment was a cage experiment in which four plots
of 45 seeds each were protected from mammalian herbivores and seed predators by wood
and wire mesh cages. Only 30 seeds per plot and three plots per treatment were used at
the Frankfort site. Seeds were planted back into their site of origin. Height, basal
diameter, number of leaves, and length and width of the first, third, and fifth leaves were
recorded monthly from June to September. Leaves were marked with paint for
identification. Only proportion emerging, final height, and survivorship to the end of the
season will be discussed here.

Greenhouse Experiment

One hundred seeds from each site (50 from Frankfort) were reserved for a
greenhouse study. The mass of each seed was recorded prior to planting in a four-inch
plot filled with a standard potting mix. Pots were placed randomly on greenhouses
benches and were re-randomized every third week. Plant measures as described above

were made weekly for 11 weeks.

125

Statistical Analyses

Differences in emergence and survival among disease status types were examined
using PROC GLM with arcsine-transformed proportions as the dependent variable and
population as the unit of replication. Plant type (seedling or other) was also included in
models of survivorship for the natural recruitment study. Type III sums of squares were
interpreted throughout. To examine size differences, the effect of disease status was
analyzed using a nested AN OVA with populations nested within a disease status type.
Since disease status did not have a significant effect on seed mass and final height, one-
way AN OVA's were used to examine differences among sites. All analyses were
performed with the SAS statistical package (SAS Institute, Inc. 1997) except for
correlation analyses which were performed with Minitab (Minitab, Inc. 1994).

RESULTS
Natural Recruitment

Preliminary analyses revealed no significant differences in survival or size of
resprouts or herbivore damaged individuals (results not presented). Therefore, these two
classes of plants were grouped together into an “other” class for comparison to true
seedlings. There was a trend for survival of American chestnut seedlings to be lower than
that of other small individuals, both within a season and overwinter (Table 5-1).
Survivorship was generally lower within the 1997 season and overwinter 1997-1998 than
for the other census periods. There was no trend for greater survivorship either within a
season or overwinter. Overwinter survival was generally lower in 1996-1997 than within

the 1996 season. This pattern was reversed for 1997 with overwinter survival greater

126

than within season survivorship. Overall, the Missaukee Healthy site had low survival
for both seedlings and other small individuals.

Survival did not vary with disease status, either within a season or overwinter
(Table 5-2). However, there was a trend for populations infected with chestnut blight to
have higher survivorship than healthy populations. The only exception to this trend was
within the 1998 season where healthy populations showed higher survival than
recovering ones. There was a distinct pattern of seedlings having lower survival than
other small plants (Table 5-2). Indeed, for two out of three within season periods (1997
and 1998), seedlings had significantly lower survival than resprouts and damaged
individuals.

Final seedling height did not vary with disease status for any census period
(results not presented). End of season seedling height varied significantly among sites in
1996 and 1998 (F5.220 = 9.49, P < 0.0001 and F,‘78 = 6.22, P < 0.0001, respectively; Table
5-3). There was no significant difference in end of season seedling height in 1997 (F5,78 =
3.84, N.S., Table 5-3). In all years, seedlings from County Line, a recovering site, were
smallest. For two out of three years, seedlings from the non-recovering Stivers
population were largest.

No consistent correlation was found between final seedling height and survival.
In 1996, there was a non-significant negative trend (rs = -0.368, df = 4, NS.) while non-
significant positive trends were evident in 1997 and 1998 (r5 = 0.45, df = 4, NS. and r5 =

0.633, df = 4, N.S., respectively).

127

Greenhou_se Experiment

There was no effect of disease status on mean seed mass (F2.3 = 0.28, P = 0.771),
but there was a significant site effect (F 5.5,, = 60.51, P < 0.0001; Table 5-4) with the
highest (Stivers) and lowest (Missaukee Diseased) average mass being found in non-
recovering sites. Mean seed mass and emergence were positively but not significantly
correlated (r, = 0.66, df = 4, NS). Disease status did not affect the proportion of
seedlings emerging (F 2.3 = 0.11, N.S.; Table 5-4) nor the relationship between mean seed
mass and emergence (r = -0.50, df = 1, NS).

Individuals from only four sites survived until the end of the experiment in
sufficient numbers to be included in analyses of final height. Therefore, only among site
differences could be examined. Significant differences among these four sites in final
seedling height were found (F3.I03 = 7.45, P < 0.0001; Table 5-4) with Frankfort seedlings
being largest and seedlings from the Missaukee Healthy site being smallest. There was
no relationship between final seedling height and either mean seed mass (rs = 0.00, df = 2,
NS.) or emergence (rs = 0.00, df = 2, NS).

Field Exgriments

The proportion of seedlings emerging did not differ for disease type (F 2 3 = 0.23,
NS.) or for sites nested within disease type (F3.6 = 3.26, NS). However, there was a
significant difference in proportion of seedlings emerging between open and caged plot
treatments (F63,4 = 16.74, P < 0.0001) with caged plots usually displaying higher
frequencies of emergence (Table 5-5). When emergence data from open and caged plots
were combined to give an overall emergence rate for a site, greenhouse and field seedling

emergence proportions were significantly positively correlated (r, = 0.93, df = 4, P <

128

0.01). For within season survival, there were no significant differences in emergence for
disease type, site, or plot treatment (F 2‘, = 5.54, N.S.; F3, = 3.57, N.S.; and F63, = 1.43,
N.S.; respectively).

Significant differences in final height of seedlings for sites nested with disease
type were found (F2,5 = 10.09, P < 0.0176; Table 5-5). However, there were no
differences in final seedling height for either treatment (caged versus open) or population
disease type (F5.23 = 1.07, NS. and F2.2 = 0.82, N.S., respectively). Final size in the field
and greenhouse studies were positively but not significantly correlated (rs = 0.72, df = 2,
NS).

The Missaukee Diseased population was not included in the following statistical
analyses because no seed germinated in either the open or caged plots at this site.
Survival of chestnut seedlings in open plots was not correlated with survival of first year
naturally recruited seedlings (rs = 0.235, df = 3, NS). Naturally recruited seedlings at the
non-recovering Stivers location had 100% survival (highest among naturally recruited
seedlings) while seedlings in the open plot experiment only had a 41% survival
probability (lowest for the open plot experiment). However, final height of open plot
seedlings and naturally recruited seedlings was positively and significantly correlated ( rs
= 0.915, df= 3, P < 0.05).

DISCUSSION

The large among site differences in seed mass and emergence were not correlated
with the disease status of the Castanea dentata populations. Thus, effects of
Cryphonectria parasitica infection on C. dentata adults may not carry-over into their

offspring, contrasting with the "Ghost of infection past" effects found in other studies

l29

(Jarosz et al. 1989). From a practical standpoint, this means that the effects of infection
can be evaluated exclusively from the performance of infected adult trees. Infected trees
are known to display significant reductions in growth, survivorship and number of seeds
produced (Griffin et al. 1983). There is a trend for populations infected with the virulent
form of the pathogen to have relatively lower finite rates of increase, while populations
infected with hyperparasitized forms of C. parasitica have intermediate growth rates
between healthy and non-recovering American chestnut populations (see Chapter 3).

Seedling survival of naturally recruited chestnut seedlings varied among sites and
among years. True seedlings generally had lower survival than other older individuals of
the same size. This finding may result from these older individuals having a larger more
established root system (Paillet 1984, 1993). Surprisingly, post-emergent survival may
have been affected by the disease status of a population. For example, the relatively low
survival observed at the Missaukee Healthy site could be due the vigor of the adult trees
at the site. The closed canopy may effectively shade the seedlings and there may be
increased competition from the high density of seedlings and other small individuals
(pers. obs.).

There were some enhancements to seedling performance in chestnut populations
where C. parasitica was present. Most notable was a trend for increased survivorship
within recovering populations. There are several possibilities to explain why this pattern
might be seen. This class of population may be closer to optimal conditions for seedling
growth. The trees are clearly recovering from the infection and growing to the point
where they are producing seed, which may be provisioned adequately for future growth,

as indicated by relatively large seed mass (Table 5-4). At the same time, average tree size

130

of adults is smaller than trees found at either healthy population studied here (see Figure
3-3). Smaller adult tree size due to disease may indirectly increase seedling performance
because of increased light levels to the forest floor (i.e., disease causes a more open
canopy in recovering populations). Alternatively, adult-seedling competition may be
reduced because of the smaller adult size.

The lack of correlation between survival of naturally recruited seedlings and
seedlings in the open plot treatment of the field experiment could be due to the
differences in sample size. Up to 5.5 times the number of seedlings were included in the
experiment as were found in the natural recruitment plots. This discrepancy could be the
result of not accurately censusing all naturally recruited seedlings at the start of the
season. Naturally recruited seedlings suffering from severe herbivore damage are
particularly difficult to identify and may die before being included in the census.
However, the strong correlation between the final size of naturally recruited and open plot
seedlings indicates that the open plot seedlings are representative of what is occurring
naturally within the populations. Therefore, the comparison of the open and caged
experiments is valid. The increased size and survivorship of caged individuals indicates
that protection of chestnut seedlings from mammalian herbivores might be an effective
management strategy to increase the transition of stage 2 to stage 4 individuals, an
important transition in some diseased populations as indicated by sensitivity and

elasticity analyses (see Chapter 4).

I31

2: no: 3.: ::4 mo: mammbo

 

 

 

 

3o 86 who :3 £5 moESm—mm
amm<mm5 mgammmz
2: 23 3o 2: 35 $550
3o 2: 3o 23 $5 85455
$525
ozam>oum¥zoz
So So 23 3o 8.. mamas
3o 33 a; £5 2: 82:83
58223::
35 85 3o 93 2: 955.0
cod 35 3o 85 3d mcéqommm
m2: 5.238
ozam>oumm n
2: 85 3o 93 93 955.0 1
was 93 one 2: 2: 8.548%
325%:
So 23 :3 33 3o mumEo
43 86 Go go :5 moéqnmmm
>m5<mm $2322
ESSE
$2.82 $2.82
Mae 25:3 mags/mus 32 25:3 mmpéamgo 82 7553 5.5

.mcozfizmo: 358:0
2.85:2 xi 5 mfittzm $2: €683.82 .553: SE: .szn 558:0 3 538:5 co b03382 9 26 86 E 8:86

6.8 :5 me—woom Ho: .2826 2m 2053 Emma: E E _ :9: $2 $3232: 550 no @5698 :o 5an8: ._.m mqufl.

 

 

 

 

 

m:.: V m .9: .m.Z Jam: 3.: v a: .:v.: .m.Z ._:._ .m.Z .m:.m $.— nmcv mm>H

m:.: N:: :w: 0:: w:.: 95:50

::.: cm: :5: 3.: mm: mOZSDmm—m
w:::-h::_ 326::—

w::_ 7:353 MmHZEMm>O 8:: 722.53 MmHEBMm>O 0:2 2:33» mm>H H7313

.m.Z :NN .m.Z .5: .m.Z do: .m.Z a:0: .m.Z .wmd Am: ":3 mm<mmHQ

8.: m:.: 0:: N:.: m:.: OZEm>OUmMTZOZ

3.: N:.: am: 0:: w:.: OZEm>Oomm

v:.: 3.: an: 3.: cm: >mhq<mm
w::_-h::_ 326::—

w::_ 2:535 awhémm>0 3:: 75.5.; mmHZHBMm>O :::_ 75th? mDH<Hm mm<mm_D

:08: :08 :0: 8:82 08 829 r: 0:030:30: 358:0

808:3 8 E @3155 32: “8:58:00: 3:30: So: 03:23:: :85 8:8 :0 $5380 :0 comtoqoi .N-m mqmflw

133

 

 

 

an and H 2.2 a w:._ H 3.2 pm and H 2.3 Dmm<mmRH mm¥3<mm=z
m N»: H 3.: a mod H :92 m Du.— H S.: mmm>r~m
OZEm>OUmMTZOZ

pm :cd H 3:2 5 om._ H 3.2 a and H 3.2 HMOMMZém
n wmd H N5: a :6 H mod— ; Nvd H c:.: m2: >HZDOU
O§m>oomd

pm 3; H om:— m _:.o H ems— m hm; H :6— D<Z<Amm4
an mmd H 8.2 m NV: H mnN— an cm: H N92 >$Hq<mm mm¥3<mm_2
>th<mm

w::_ 5::_ ©::_ mtm

£82: 5 80:22.26 Emu—mama 28me 228— 288me .8382: 2a 228 23.8%

H. 2:82 .5588 508:5 mo macaw—anon xmm 8.2 me—uoom 89» 22: “82:82 312:8: mo Emma: Ram 582 .m-m mama;

134

 

 

 

 

8: o S: H Rd nmm<mm5 mmmimaz

B o; H 3.8 o; a 2: H Ma: 9525
czam>oom:-zoz

a 8: H :3 $3 3 Ed H on: 28:25:

8.0 a 8: H 3: m2: Z758

022385:

an SH H 3% S: on 85 m 85 D<z<qmmq

u on: H 8.: and B 85 + 8.0 ESE: mmmafimuz
>518:

505: .275 252 ozammzm 22295:: as): 8% 72:: Ba

458530 2: mo ::o 2: 8 @0233 _m:E>€:_ u:o b:o 3:82. 3682: 8: mm 0:5 3:500 .8.: onm 3:5 :52:

E mageflbc ::5:.::me 2385 £052 “:obba .3338: 2: 32.5 935% H mama—2 .3585 538:2 .«o gown—sac:

xi :8 ::oEtoqxo 3:04:08» m 5 Emma: 35.: :38 ::m .meBEo $5608 :0 53.83:: .32: v08 :32 .v-m mum»;

135

 

dorm—ac: £5 80¢ vofifigom 303 02 u .Q.Z

 

.Q.Z cod cod Qmm<mmfiu mm¥D<mm__2
Nod H mad _ Nwd mmd Qm0<0
co; H 3.2 36 36 meO

as Ed H 8.2 3.0 end myfiS—Hm
mvd H ema— Nod 36 QmO<U
Ed H 2.: 2: mod ZmEO

a: 9.6 H 3M _ Nod #00 HMOmv—Zém
owd H med 2: cod QmO<U
:._ H 36 and wed meO

o :6 H o:. mod nod m2: >HZDOU
New H 3.2 mud Nod DmO<U
wad H 50.2 006 86 meO

: mmN H 8.2 Sd mod D<Z<Ammq
cod H om.N_ cmd m _ .o QmO<U
mo; H wed mod cod meO

a #06 H 2.: who :6 >IHA<MI mm¥D<mm=2

A<>H>m3m OEOMmZm
HID—mm A<ZE Z<m2 ZOm<mm 25.2.3 ZOEKMOAOME mtm

233:: 05 88.: 3388 mm? Soho

:RE mafia 83$: ::aou_:wmm-:o: 05 Sam 2:2: 86 039:8 8 38:88: 203 $3: totmudm V ::obbmu base—MEwmm

8: 2: 8:3 2:8 05 3 832—8 332 .3888: 8: 29:0 23:8...“ H 232 .3535 538:3 .«o 30:23::

xfi 8.: 355:5on 22.: E Ems: 3:: :38 ::m .3233 533 553» .wEwSEo me—voom mo :oEomoi .m-m mqufi—l

136

CHAPTER 6

CONCLUSIONS

The presence of double-stranded RNA in populations of the chestnut blight
pathogen, Cryphonectria parasitica, allows recovery of American chestnuts, both at an
individual and population level. While dsRNA appears necessary for recovery to occur,
the mere presence of dsRNA in isolates of the pathogen is not sufficient for recovery of
individuals or populations. The effects of dsRNA may be altered by the size of the
individual infected and spatial and temporal variation among populations.

The importance of American chestnut branch size in determining the outcome of
infection by chestnut blight was supported by both an inoculation experiment and a
natural infection study. Branch size influenced canker morphology and branch survival
although some of these effects only developed over time. DsRNA reduced canker
growth rates and appeared to delay mortality for medium and large size branches.
Classification of cankers into callused or non-callused did not accurately predict the
type of fungal isolate which initiated infection, especially after the first season of
infection when branch size was a more important influence on canker development.
The finding that dsRNA alone is not related to branch survivorship emphasizes that the
presence of dsRN A in a pathogen population may be necessary but not sufficient for
biological control of disease. Although the presence of dsRNA may enable large
branches to survive, small branches still succumb to infection by both dsRNA-

containing and dsRNA-free isolates. Overall, branch size at the time of infection

137

appears to have a large influence on branch survivorship, at least in the short term.
Therefore, the presence of a sufficiently debilitating dsRNA in the pathogen population
may not be the only requirement for recovery of American chestnut p0pulations. The
size of an individual tree at the time of infection could also be an important factor in
determining the long term survival of American chestnut populations.

Investigations on the influence of the interaction of pathogen virulence and
branch size on mortality and canker development should be expanded to include
dsRNAs with a range of debilitating effects. In addition, efforts should be made to
examine natural canker establishment, colonization and subsequent spread of dsRN A
within the cankers, and ultimate branch fate.

Matrix projection models of population growth rate and size class distributions
revealed that chestnut blight is having an impact on the structure and growth of infected
American chestnut populations. In the presence of disease, large trees may be reduced
in size, a trend which was not observed in healthy populations during either census
period. Indeed, retrogressions (reductions in size) were not found in any of the 21
woody plant species examined in a recent review (Silvertown et al. 1993). Further,
reproduction of individuals in smaller size classes was found in infected populations.
Population growth rates (1) varied among sites and among years. However, the relative
rankings, with non-recovering populations having the lowest growth rates over two
census periods, indicate that disease is indeed having a negative impact on chestnut

population growth rates. The intermediate rankings of recovering populations indicates

I38

that the presence of dsRNA can promote ecological recovery of American chestnut
populations. Observed population structures differed significantly from the predicted
stable stage distributions. The pattern of diseased populations being dominated by
intermediate-sized individuals at stable stage, while healthy populations are dominated
by small individuals, was further indication that disease is influencing American
chestnut populations.

There are several avenues of investigation that might be pursued fitrther. First,
demographic studies of healthy, recovering, and non-recovering American chestnut
populations should be continued to examine temporal variation in estimates of the finite
rate of population increase (7t). When transition matrices are available that represent a
range of conditions (i.e. good versus had years), more accurate long-term predictions of
the fate of the populations can be made. Models can be constructed in which transition
matrices representing different conditions are selected at random and the trajectories of
the populations can be predicted. Further, following populations for a longer period of
time will allow the trajectory and pattern of population growth rates after the
introduction of C. parasitica and then dsRN A to be determined (see Figure 3-4).

Matrix projection models can also produce sensitivity and elasticity analyses to
examine the contribution of particular stages and transitions to populations growth rates.
The sensitivity analyses reveal that, in general, for healthy and recovering populations
of American chestnut, changes in the transition probability for survival in the largest

stage would have the greatest impact on population growth. An exception to this finding

139

is that for non-recovering populations the highest sensitivities were generally found for
growth of stage 6 (1 - 10 cm dbh) individuals. This result may have important
management implications. If one wanted to treat chestnut trees and introduce dsRNA
into populations, these findings suggest that trees in this size class should be treated.
Elasticity analyses revealed that survival within a stage has the largest proportional
contribution to the finite rate of population increase. Silvertown et al. (1993) divide
elasticity matrices into three regions (G, L, and F) and sum elasticities within each
region. For species such as the American chestnut, which appear to occupy a very
narrow range of G/L/F space, evaluating successful management strategies may be
problematic with this method. These results emphasize the need for examination and
interpretation of a combination of demographic parameters to evaluate management
strategies for successful treatment and conservation of American chestnut populations.

While the use of G/UF ratios in conservation management strategies is
promising (Silvertown et al. 1996), the interpretation of the resulting patterns among
populations within a species should be refined to reflect the narrow range some species
such as the American chestnut occupy in G/L/F space. Further, the influence of
environmental factors such as herbivory and disease as opposed to successional stage
should be examined to determine the potential trajectories of populations under these
more subtle yet still important disturbances.

Cryphonectria parasitica infections of adult chestnut trees do not appear to have

any prolonged adverse effects on the performance of offspring. Therefore, the effects of

I40

infection can be evaluated exclusively from the performance of infected adult trees.
Survival of naturally recruited chestnut seedlings varied among sites and among years.
True seedlings generally had lower survival than other older individuals of the same
size. However, there was a trend for increased seed mass and enhanced germination and
survivorship of seedlings within recovering populations. Disease status did not affect
final seedling size whereas final seedling size did vary with population. The increased
size and survivorship of individuals in a cage experiment versus those seedlings in open
plots indicate that protection of chestnut seedlings from mammalian herbivores might
be an effective management strategy to increase the transition probabilities of small
individuals, an important transition in some diseased populations as indicated by
sensitivity and elasticity analyses.

Studies on the effects of infection on reproduction could be refined to account
for the relative health of the seed parent and the subsequent effects on seed quality and
seedling performance. Further, genetic versus environmental components of seedling
performance could be teased apart by using maternal half-sib families in greenhouse and

field reciprocal transplant studies.

I41

LITERATURE CITED

142

LITERATURE CITED

Alexander HM. 1989. An experimental field study of anther-smut disease of Silene alba
caused by Ustilago violacea: Genotypic variation and disease incidence. Evolution
43:835-847.

Alexander HM & Antonovics J. 1988. Disease spread and population dynamics of anther-
smut infection of Silene alba caused by the fungus Ustilago violacea. Journal of
Ecology 76:91-104.

Alexander HM & Burdon JJ. 1984. The effect of disease induced by Albugo candida
(white rust) and Peronospora parasitica (downy mildew) on the survival and
reproduction of Capsella bursa-pastoris (shepherd's purse). Oecologia 64:314-318.

Alexander HM, Thrall PH, Antonovics J, Jarosz AM &Oudemans PV. 1996. Population
dynamics and genetics of plant disease: a case study of anther-smut disease of Silene
alba caused by the fungus Ustilago violacea. Ecology 77: 990-996.

Anagnostakis SL. 1982. Biological control of chestnut blight. Science 215:466-471.

Anagnostakis SL. 1984. The mycelial biology of Endothia parasitica. 1. Nuclear and
cytoplasmic genes that determine morphology and virulence. IN: The Ecology and
Physiology of the Fungal Mycelium (Ed. by DH Jennings & ADM Rayner), pp. 353-
366. Cambridge University Press, Cambridge.

Anagnostakis SL. 1988. Cryphonectria parasitica, cause of chestnut blight. IN:
Advances in Plant Pathology, Vol. 6 (Ed. by GS Sidhu, DS Ingram & PH Williams),
pp. 123-136. Academic Press, London.

Anagnostakis SL & Waggoner PE. 1981. Hypovirulence, vegetative incompatibility, and
the growth of cankers of chestnut blight. Phytopathology 71 :1 198-1202.

Anderson RM & May RM. 1982. Coevolution of hosts and parasites. Parasitology
85:411-426.

Antonovics J & Alexander HM. 1989. The concept of fitness in plant-fungal pathogen

systems. In: Plant Disease Epidemiology: Genetics, Resistance, and Managment, Vol.2
(Ed. by KJ Leonard & WE Fry), pp. 185-214. McGraw-Hill, Inc., New York.

Antonovics J & Ellstrand NC. 1984. Experimental studies of the evolutionary

significance of sexual reproduction. I. A test of the frequency-dependent selection
hypothesis. Evolution 38: 103-1 15.

143

Antonovics J, Thrall P, Jarosz A & Stratton D. 1994. Ecological genetics of
metapopulations: the Silene-Ustilago plant-pathogen system. In: Ecological Genetics
(Ed. by LA Real), pp 146-170. Princeton University Press, Princeton.

 

Augspurger CK. 1984. Seedling survival of tropical tree species: interactions of dispersal
distance, light gaps, and pathogens. Ecology 65:1705-1712.

Baxter DV & Strong FC. 1931. Chestnut blight in Michigan. Michigan State College
Agricultural Experiment Station Circular Bulletin 135: 1-18.

Biraghi A. 1950a. Caratteri di resistenze in Castanea sativa nei confronti di Endothia
parasitica. Bollettino della Srazione di parologia vegetale di Roma 7:161-171.

Biraghi A. 1950b. La distribuzione del cancro del castagno in Italia. Italia forestale e
montana 5:18-21.

Bissegger M, Rigling D & Heiniger U. 1997. Population structure and disease
development of Cryphonectria parasitica in European chestnut forest in the presence of
natural hypovirulence. Phytopathology 87:50-59.

Brasier CM. 1987. Recent genetic changes in the Ophiostoma ulmi population: the threat
to the future of the elm. In: Populations of Plant Pathogens: Their Dynamics and
Genetics (Ed. by MS Wolfe & CE Caten), pp.213-226. Blackwell Scientific
Publications, Oxford.

Brasier CM. 1990. The unexpected element: mycovirus involvement in the outcome of
two recent pandemics, Dutch elm disease and chestnut blight. In: Pests, Pathogens and
Plant Communities (Ed. by JJ Burdon & SR Leather), pp. 289-307. Blackwell
Scientific Publications, Oxford.

Brasier, CM. 1998. Virus-mediated biological control of plant pathogens. In: Proceedings

of the Brighton Crop Protestion Conference - Pests and Diseases 1998, Vol. _2_. pp. 425-
432. British Crop Protection Council, F arnham, UK.

 

Brewer LG. 1995. Ecology of survival and recovery from blight in American chestnut
trees (Castanea dentata (Marsh) Borkh.) in Michigan. Bulletin of the Torrey Botanical
Club 122:40-57.

Bull JJ, Molineux IJ & Rice WR. 1991. Selection of benevolence in a host-parasite
system. Evolution 45:875-882.

Burdon JJ. 1982. The effect of fungal pathogens on plant communities. In: The Plant
Communig as a Working Mechanism ( Ed. by El Newman), pp 99-112. Blackwell
Scientific Publications, Oxford.

Burdon JJ. 1987. Diseases and Plant Population Biology. Cambridge University Press,
Cambridge.

Burdon JJ, Groves RH, Kaye PE & Cullen JM. 1981. The impact of biological control on
the distribution and abundance of Chondrillajuncea in south-eastem Australia. Journal
of Applied Ecology 18:957-966.

Burdon JJ, Wennstrom A, Muller WJ & Ericson L. 1994. Spatial patterning in young
stands of Pinus sylvestris in relation to mortality caused by the snow blight pathogen
Phacidium infestans. Oikos 71 : 130-136.

Byers DL & Meagher TR. 1997. A comparison of demographic characteristics in a rare
and a common species of Eupatorium. Ecological Applications 7:519-530.

Carlsson U, & Elmqvist T. 1992. Epidemiology of anther-smut disease (Microbotryum

violaceum) and numeric regulation of populations of Silene dioica. Oecologia 90:509-
517.

Carlsson U, Elmqvist T, Wennstrom A & Ericson L. 1990. Infection by pathogens and
population age of host plants. Journal of Ecology 78:1094-1105.

Caswell H. 1989. Matrix population models. Sinauer Associates, Inc., Sunderland MA. 3

Clay K. 1984. The effect of the fungus Atkinsonella hypoxylon (Clavicipitaceae) on the

reproductive system and demography of the grass Danthom'a spicata. New Phytologist
98: 165-175.

Crowder LB, Crouse DT, Heppell SS & Martin TH. 1994. Predicting the impact of turtle
excluder devices on loggerhead sea-turtle populations. Ecological Applications 4:43 7-
445.

Crowder LB, Hopkins Murphy SR & Royle JA. 1995. Effects of turtle excluder devices
(TEDs) on loggerhead sea turtle strandings with implications for conservation. Copeia
4:773-779.

Crouse DT, Crowder LB & Caswell H. 1987. A stage-based population model for
loggerhead sea turtles and implications for conservation. Ecology 68: 1412-1423.

Daughtrey, ML & Hibben CR. 1994. Dogwood anthracnose - A new disease threatens
two native Camus species. Annual Review of Phytopathology 32:61-73.

Davelos AL, Schaupp JK, Fulbright DW & Jarosz AM. 1995. Double-stranded RNAs in
three populations of Cryphonectria parasitica and their relationship with branch
recovery. Phytopathology 8521176. (Abstract)

145

Davelos AL, Schaupp JK, Huber DH & Fulbright DW. 1996. Variation in diversity of
vegetative compatibility groups of Cryphonectria parasitica among populations in
Michigan. Phytopathology 86:811. (Abstract)

Davelos AL, Schaupp JK, Jarosz AM & Fulbright DW. 1997. Relationship between
vegetative compatibility group diversity and spread of double-stranded RNA in
Michigan populations of Cryphonectria parasitica. Phytopathology 87:823. (Abstract)

Day FP & Monk CD. 1974. Vegetation patterns on a southern Appalachian watershed.
Ecology 55:1064-1074.

Day PR, Dodds JA, Elliston JE, Jaynes RA & Anagnostakis SL. 1977. Double-stranded
RNA in Endothia parasitica. Phytopathology 67:1393-1396.

de Kroon HJ, Plaiser A, van Groenendail J & Caswell H. 1986. Elasticity: the relative
contribution of demographic parameters to population growth rate. Ecology 67:1427-
1431.

Dickman A. 1992. Plant pathogens and long-term ecosystem changes. In: fig Fungal
Community: I_t_s_ Organization an_d_ Role Q m; Ecosystem (Ed. by GC Carroll & DT
Wicklow), pp 499-520. Marcel Dekker, Inc., New York.

Ehrlén J. 1995. Demography of the perennial herb Lathyrus vernus. II. Herbivory and
population dynamics. Journal of Ecology 83:297-308.

Elliston J. 1985. Characteristics of dsRNA-free and dsRNA-containing strains of
Endothia parasitica. Phytopathology 75:1 51-158.

Elliston JE, Jaynes RA, Day PR & Anagnostakis SL. 1977. A native American
hypovirulent strain of Endothia parasitica. Proceedings of the American
Phytopathological Society 4:83. (Abstract)

Emery KM. 1998. Population dynamics of birdsfoot trefoil in relation to disease and
microclimate. Ph.D. thesis, University of Missouri, USA.

Enebak SA, MacDonald WL & Hillman BI. 1994. Effect of dsRNA associated with
isolates of Cryphonectria parasitica from the central Appalachians and their relatedness
to other dsRNAs from North America and Europe. Phytopathology 84:528-534.

Enright NJ & Watson AD. 1991. A matrix population model analysis for the tropical tree,
Araucaria cunninghamii. Australian Journal of Ecology 16:507-520.

Ewald PW. 1983. Host-parasite relations, vectors, and the evolution of disease severity.
Annual Review of Ecology and Systematics 14:465-485.

146

Fenner F. 1983. Biological control, as exemplified by smallpox eradication and
myxomatosis. Proceedings of the Royal Society of London B 218:259-285.

Fenner F & Ratcliffe FN. 1965. Mflomatosis. Cambridge University Press, Cambridge.

Ferson S. 1994. RAMAS/stage: Generalized Stage-based Modeling for Population
Dynamics. Applied Biomathematics, Setauket, NY.

Frank S. 1996. Models of parasite virulence. Quarterly Review of Biology 71:37-78.

Frantzen J. 1994. The role of clonal growth in the pathosystem C irsium arvense -
Puccinia punctiformis. Canadian Journal of Botany 72:832-836.

Fulbright DW. 1984. Effect of eliminating dsRNA in hypovirulent Endothia parasitica.
Phytopathology 74:722-724.

Fulbright DW, Weidlich WH, Haufler KZ, Thomas CS & Paul CP. 1983. Chestnut blight
and recovering American chestnut trees in Michigan. Canadian Journal of Botany
61 :3164-3171.

Garcia-Guzman G & Burdon JJ. 1997. Impact of the flower smut Ustilago cynodontis
(Ustilaginaceae) on the performance of the clonal grass Cynodon dactylon (Gramineae).
American Journal of Botany 84:1565-1571.

Garcia-Guzman G, Burdon JJ, Ash JE & Cunningham RB. 1996. Regional and local
patterns in the spatial distribution of the flower-infecting smut fungus Sporisorium

amphilophis in natural populations of its host Bothriochloa macra. New Phytologist
132:459-469.

Geils BW & Jacobi WR. 1993. Effects of comandra blister rust on growth and survival of
lodgepole pine. Phytopathology 83:63 8-644.

Griffin GJ, Elkins JR, Tomimatsu G & Hebard F V. 1978. Virulence of Endothia
parasitica isolated from surviving American chestnut trees. In: Proceedings of the
American Chestnut Symmsium (Ed. by WL MacDonald, FC Cech, J Luchok & C
Smith), pp 55-60. West Virginia University Press, Morgantown.

Griffin GJ, Hebard F V, Wendt RW &Elkins JR. 1983. Survival of American chestnut
trees: evaluation of blight resistance and virulence in Endothia parasitica.
Phytopathology 73:1084-1092

Griffin GJ, Smith HC, Dietz A & Elkins JR. 1991. Importance of hardwood competition

to American chestnut survival, growth, and blight development in forest clearcuts.
Canadian Journal of Botany 69:1804-1809.

147

Harper JL. 1977. P_opulation Biology pf Plants. Academic Press, New York.

Heiniger U & Rigling D. 1994. Biological control of chestnut blight in Europe. Annual
Review of Phytopathology 32:581-599.

Hibben, CR. 1990. Anthracnose threatens the flowering dogwood. Amoldia 50:16-20.

Hillman BI, Fulbright DW, Nuss DL & van Alfen NK. 1995. Hypoviridae. In: _Si_xth
Report of the International Committee for the Taxonomy of Viruses (Ed. by FA
Murphy, CM F auquet, DHL Bishop, SA Ghabrial, AW Jarvis, GP Martelli, MP Mayo
& MD Summers), pp. 261-264. Springer Verlag, New York.

Holah JC, Wilson MV & Hansen EM. 1993. Effects of a native forest pathogen,
Phellinus weirii, on Douglas-fir forest composition in western Oregon. Canadian
Journal of Forest Research 23:2473-2480.

Hollings M. 1982. Mycoviruses and plant pathology. Plant Disease 66:1106-1112.

Horvitz CC & Schemske DW. 1995. Spatiotemporal variation in demographic transitions
of a tropical understory herb: projection matrix analysis. Ecological Monographs
65: 1 55-192.

Huenneke LF & Marks PL. 1987. Stem dynamics of the shrub Alnus incana ssp. rugosa:
Transition matrix models. Ecology 68: 1234-1242.

Jarosz AM & Burdon JJ. 1992. Host-pathogen interactions in natural populations of
Linum marginale and Melampsora Iini. III. Influence of pathogen epidemics on host
survivorship and flower production. Oecologia 89:53-61.

Jarosz AM, Burdon JJ &Mr’lller WJ. 1989. Long-term effects of disease epidemics.
Journal of Applied Ecology 26:725-733.

Jarosz AM & Davelos AL. 1995. Effects of disease in wild plant populations and the
evolution of pathogen aggressiveness. New Phytologist 129:371-3 87.

Jaynes RA. 1978. Selecting and breeding blight resistant chestnut trees. In: Proceedings
o_f mg American Chestnut Symmsium (Ed. by WL MacDonald, FC Cech, J Luchok &
C Smith), pp 4-6. West Virginia University Press, Morgantown.

Jaynes RA & Elliston JE. 1982. Hypovirulent isolates of Endothia parasitica associated
with large American chestnut trees. Plant Disease 66:769-772.

Kalisz S & McPeek MA. 1992. Demography of an age-structured annual: resampled
projection matrices, elasticity analyses, and seed bank effects. Ecology 73:1082-1093.

148

Karban R. 1978. Changes in an oak-chestnut forest since the chestnut blight. Castanea
43:221-228.

Keever C. 1953. Present composition of some stands of the former oak-chestnut forest is
the southern Blue Ridge Mountains. Ecology 34:44-54.

Kephart SR & Paladino C. 1997. Demographic change and microhabitat variability in a
grassland endemic, Silene douglasii var. oraria (Caryophyllaceae). American Journal of
Botany 84:179-189.

Kile GA. 1983. Armillaria root rot in eucalypt forests: aggravated endemic disease.
Pacific Science 37: 459-464.

Kuhlman EG. 1978. Interactions of virulent and hypovirulent strains of Endothia
parasitica on American chestnuts in North Carolina. In: Proceedings of the American
gigstnut Symposium (Ed. by WL MacDonald, F C Cech, J Luchok & C Smith), pp
1 15-117. West Virginia University Press, Morgantown.

Lande R. 1988. Demographic models of the northern spotted owl (Strix occidentalis
caurina). Oecologia 75:601-607.

Lee JK, Tattar TA, Berrnan PM & Mount MS. 1992. A rapid method for testing the
virulence of Cryphonectria parasitica using excised bark and wood of American
chestnut. Phytopathology 82: 1454-1456.

Leege LM. 1997. The ecological impact of Austrian pine (Pinus nigra) on the sand dunes
of Lake Michigan: an introduced species becomes an invader. Ph.D. thesis, Michigan
State University, USA.

Lefltovitch LP. 1965. The study of population growth in organisms grouped by stage.
Biometrics 21 :1-18.

Lenski RE & Bouma J. 1987. Effects of segregation and selection on instability of
pACYC184 in Escherichia coli B. Journal of Bacteriology 169:5314-5316.

Lenski RE & May RM. 1994. The evolution of virulence in parasites and pathogens:
reconciliation between two competing hypotheses. Journal of Theoretical Biology
169:253-265.

Leslie PH. 1945. On the use of matrices in certain population mathematics. Biometrika
33:183-212.

Liu YC, Cortesi P, Double ML, MacDonald WL & Milgroom MG. 1996. Diversity and

multilocus genetic structure in populations of Cryphonectria parasitica.
PhytOpathology 86:1344-1351 .

149

MacDonald WL & Fulbright DW. 1991. Biological control of chestnut blight: Use and
limitations of transmissible hypovirulence. Plant Disease 75:656-661.

Marr DL. 1997. Impact of a pollinator-transmitted disease on reproduction in healthy
Silene acaulis. Ecology 78:1471-1480.

May RM & Anderson RM. 1983. Epidemiology and genetics in the coevolution of
parasites and hosts. Proceedings of the Royal Society 8219:281-313.

May RM & Anderson RM. 1990. Parasite-host coevolution. Parasitology 100:889-8100.

McManus PS, Ewers FW & Fulbright DW. 1989. Characterization of the chestnut blight
canker and the localization and isolation of the pathogen Cryphonectria parasitica.
Canadian Journal of Botany 67:3600-3607.

Menges ES. 1990. Population viability analysis for an endangered plant. Conservation
Biology 4:52-62.

Merkel HW. 1905. A deadly fungus on the American chestnut. 10‘” Annual Report of the
New York Zoological Society, Bronx. pp. 97-103.

Milgroom MG. 1995. Population biology of the chestnut blight fungus, Cryphonectria
parasitica. Canadian Journal of Botany (Supplement 1) 73:8311-8319.

Milgroom MG & Lipari SE. 1995. Population differentiation in the chestnut blight
fungus, Cryphonectria parasitica, in eastern North America. Phytopathology 85: 155-
160.

Minitab, Inc. 1994. Minitab Reference Manual, Release 10. Minitab, Inc., State College,
PA.

Nuss DL. 1992. Biological control of chestnut blight: an example of virus-mediated
attenuation of fungal pathogenesis. Microbiological Reviews 56:561-576.

Nuss DL & Koltin Y. 1990. Significance of dsRNA genetic elements in plant pathogenic
fungi. Annual Review of Phytopathology 28:37-58.

Paillet FL. 1984. Growth-form and ecology of American chestnut sprout clones in
northeastern Massachusetts. Bulletin of the Torrey Botanical Club 111:316-328.

Paillet FL. 1988. Character and distribution of American chestnut sprouts in southern
New England woodlands. Bulletin of the Torrey Botanical Club 115233-44.

150

Paillet FL. 1993. Growth form and life histories of American chestnut and Allegheny and
Ozark chinquapin at various North American sites. Bulletin of the Torrey Botanical
Club 120:257-268.

Paillet FL & Rutter PA. 1989. Replacement of native oak and hickory tree species by the

introduced American chestnut (Castanea dentata) in southwestern Wisconsin.
Canadian Journal of Botany 67:3457-3469.

Parker GG, Hill SM & Kuehnel LA. 1993. Decline of understory American chestnut
(Castanea dentata) in a southern Appalachian forest. Canadian Journal of Forestry
Research 23:259-265.

Parker MA. 1986. Individual variation in pathogen attack and differential reproductive
success in the annual legume, Amphicarpaea bracteata. Oecologia 69:253-259.

Parker MA. 1987. Pathogen impact on sexual vs. asexual reproductive success in
Arisaema triphyllum. American Journal of Botany 74:1758-1763.

Parker MA. 1988. Genetic uniformity and disease resistance in a clonal plant. American
Naturalist 132:538-549.

Parker MA. 1991. Nonadaptive evolution of disease resistance in an annual legume.
Evolution 45: 1209-1217.

Pascarella JB & Horvitz CC. 1998. Hurricane disturbance and the population dynamics of
a tropical understory shrub: Megamatrix elasticity analysis. Ecology 79:547-563.

Paul CP & Fulbright DW. 1988. Double-stranded RNA molecules from Michigan
hypovirulent isolates of Endothia parasitica vary in size and sequence homology.
Phytopathology 78:751-755.

Paul ND & Ayres PG. 1986a. The impact of a pathogen (Puccinia lagenophorae) on
populations of groundsel (Senecio vulgaris) overwintering in the field. 1. Mortality,
vegetative growth and the development of size hierarchies. Journal of Ecology
74: 1069-1084.

Paul ND & Ayres PG. 1986b. The impact of a pathogen (Puccinia lagenophorae) on
populations of groundsel (Senecio vulgaris) overwintering in the field. 11.
Reproduction. Journal of Ecology 74:1085-1094.

Paul ND & Ayres PG. 1987. Survival, growth and reproduction of groundsel (Senecio

vulgaris) infected by rust (Puccinia lagenophorae) in the field during summer. Journal
of Ecology 75:61-71.

151

Peever TL, Liu Y-C & Milgroom MG. 1997. Diversity of hypoviruses and other double-
stranded RNAs in Cryphonectria parasitica in North America. Phytopathology
87:1026-1033.

Peto R & Peto J. 1972. Asymptotically efficient rank invariant procedures. Journal of the
Royal Statistical Society, Series A 135:185-207.

Pifiero D, Martinez-Ramos M & Sarukhan J. 1984. A population model of Astrocaryum

mexicanum and a sensitivity analysis of its finite rate of increase. Journal of Ecology
72:977-991.

Pyke DA & Thompson JN. 1986. Statistical analysis of survival and removal rate
experiments. Ecology 67:240-245.

Pyke DA & Thompson JN. 1987. Erratum. Ecology 68:232.

Roane MK, Griffin GJ & Elkins JR. 1986. Chestnut Blight, Other Endothia Diseases app
_tfi Genus Endothiq. APS Press, St. Paul.

 

Roy BA & Bierzychudek P. 1993. The potential for rust infection to cause natural
selection in apomictic Arabis holboellii (Brassicaceae). Oecologia 95:533-541.

Russell EWB. 1987. Pre-blight distribution of Castanea dentata (Marsh) Borkh. Bulletin
of the Torrey Botanical Club 114:183-190.

Russin J S & Shain L. 1985. Disseminative fitness of Endothia parasitica containing
different agents for cytoplasmic hypovirulence. Canadian Journal of Botany 63:54-57.

SAS Institute, Inc. 1997. SAS/STAT Software: Changes and Enhancements Through
Release 6.12. SAS Institute, Inc., Cary, NC.

Schaffer B, Hawksworth FG & Jacobi WR. 1983. Effects of comandra blister rust and
dwarf mistletoe on cone and seed production of lodgepole pine. Plant Disease 67:215-
217.

Scheiner SM. 1993. MANOVA: Multiple response variables and multispecies
interactions. In: Design and Analysis of Ecological Exmriments (Ed. by SM Scheiner
& J. Gurevitch), pp 94-112. Chapman & Hall, Inc., New York.

Schemske DW, Husband BC, Ruckelshaus MH, Goodwillie C, Parker 1M & Bishop JG.

1994. Evaluating approaches to the conservation of rare and endangered plants.
Ecology 75:584-606.

152

Silvertown J, Franco M & Menges E. 1996. Interpretation of elasticity matrices as an aid
to the management of plant populations for conservation. Conservation Biology
10:591-597.

Silvertown J, Franco M, Pisanty I & Mendoza A. 1993. Comparative plant demography -
relative importance of life-cycle components to the finite rate of increase in woody and
herbaceous perennials. Journal of Ecology 81:465-476.

Silvertown JW & Lovett Doust J. 1993. Introduction 19 plant population biology.
Blackwell Science, Oxford.

Sokal RR & Rohlf F J . 1981. Biomefl, 2nd Ed. W.H. Freeman and Co., New York.

Stephenson SL, Adams HS & Lipford M. 1991. The present distribution of chestnut in
the upland forest communities of Virginia. Bulletin of the Torrey Botanical Club
118224-32.

Taylor DR, Jarosz AM, Lenski RE & Fulbright DW. 1998. The acquisition of
hypovirulence in host-pathogen systems with three trophic levels. American Naturalist
151 :343-355.

Thrall PH & J arosz AM. 1994. Host-pathogen dynamics in experimental populations of
Silene alba and Ustilago violacea. 11. Experimental tests of theoretical models. Journal
of Ecology 82:561-570.

Valverde T & Silvertown J. 1998. Variation in the demography of a woodland
understorey herb (Primula vulgaris) along the forest regeneration cycle: projection
matrix analysis. Journal of Ecology 86:545-562.

Van Alfen NK, Jaynes RA, Anagnostakis SL & Day PR. 1975. Chestnut blight:
Biological control by transmissible hypovirulence in Endothia parasitica. Science
189:890-891.

von Ende CN. 1993. Repeated-measures analysis: Growth and other time-dependent
measures. In: Design and Analysis of Ecological Exp_eriments (Ed. by SM Scheiner &
J. Gurevitch), pp 113-137. Chapman & Hall, Inc., New York.

Wennstrom A & Ericson L. 1990. The interaction between the clonal herb Trientalis
europaea and the host specific smut fungus Urocystis trientalis. Oecologia 85:238-240.

Wennstrom A & Ericson L. 1991. Variation in disease incidence in grazed and ungrazed
sites for the system Pulsatilla pratensis-Puccinia pulsatillae. Oikos 60:35-39.

Weste G. 1980. Vegetation changes as a result of invasion of forest on krasnozem by
Phytophthora cinnamomi. Australian Journal of Botany 28:139-150.

153

Weste G. 1981. Changes in the vegetation of sclerophyll shrubby woodland associated
with invasion by Phytophthora cinnamomi. Australian Journal of Botany 29:261-276.

Yao JM, McElreath SD & Tainter PH. 1997. Double stranded RNA in isolates of
Discula destructiva from the Pacific northwesten United States and British Columbia,
Canada. Current Microbiology 34:67-69.

Yao JM & Tainter PH. 1996. Virus-like particles in Discula destructiva. Canadian
Journal of Plant Pathology 18:433-438.

154

 

                

   

   

HICHIGRN STATE UN V.

1111111141 1111111111

312 Sfl 04 84

 

 

 

 

 

 

 

 

 

 

Rnnrss
llllllll
115