Citation
Genetic Effects of Rooting Ability and Early Growth Traits in Loblolly Pine Clones

Material Information

Title:
Genetic Effects of Rooting Ability and Early Growth Traits in Loblolly Pine Clones
Creator:
BALTUNIS, BRIAN STEPHEN
Copyright Date:
2008

Subjects

Subjects / Keywords:
Analytical estimating ( jstor )
Genetic correlation ( jstor )
Genetic gain ( jstor )
Genetic variance ( jstor )
Genetic variation ( jstor )
Heritability ( jstor )
Phenotypic traits ( jstor )
Population estimates ( jstor )
Seedlings ( jstor )
Statistical discrepancies ( jstor )
City of Gainesville ( local )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Brian Stephen Baltunis. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
4/17/2006
Resource Identifier:
77078672 ( OCLC )

Downloads

This item has the following downloads:

baltunis_b.pdf

baltunis_b_Page_090.txt

baltunis_b_Page_071.txt

baltunis_b_Page_080.txt

baltunis_b_Page_052.txt

baltunis_b_Page_072.txt

baltunis_b_Page_092.txt

baltunis_b_Page_056.txt

baltunis_b_Page_085.txt

baltunis_b_Page_027.txt

baltunis_b_Page_102.txt

baltunis_b_Page_100.txt

baltunis_b_Page_015.txt

baltunis_b_Page_078.txt

baltunis_b_Page_020.txt

baltunis_b_Page_018.txt

baltunis_b_Page_091.txt

baltunis_b_Page_043.txt

baltunis_b_Page_063.txt

baltunis_b_Page_038.txt

baltunis_b_Page_044.txt

baltunis_b_Page_014.txt

baltunis_b_Page_110.txt

baltunis_b_Page_103.txt

baltunis_b_Page_050.txt

baltunis_b_Page_029.txt

baltunis_b_Page_054.txt

baltunis_b_Page_032.txt

baltunis_b_Page_037.txt

baltunis_b_Page_042.txt

baltunis_b_Page_087.txt

baltunis_b_Page_083.txt

baltunis_b_Page_099.txt

baltunis_b_Page_061.txt

baltunis_b_Page_108.txt

baltunis_b_Page_016.txt

baltunis_b_Page_007.txt

baltunis_b_Page_057.txt

baltunis_b_Page_048.txt

baltunis_b_Page_002.txt

baltunis_b_Page_098.txt

baltunis_b_Page_035.txt

baltunis_b_Page_101.txt

baltunis_b_Page_017.txt

baltunis_b_Page_106.txt

baltunis_b_Page_028.txt

baltunis_b_Page_051.txt

baltunis_b_Page_040.txt

baltunis_b_Page_031.txt

baltunis_b_Page_070.txt

baltunis_b_Page_026.txt

baltunis_b_Page_021.txt

baltunis_b_Page_073.txt

baltunis_b_Page_006.txt

baltunis_b_Page_066.txt

baltunis_b_Page_074.txt

baltunis_b_Page_064.txt

baltunis_b_Page_025.txt

baltunis_b_Page_024.txt

baltunis_b_Page_022.txt

baltunis_b_Page_060.txt

baltunis_b_Page_001.txt

baltunis_b_Page_047.txt

baltunis_b_Page_097.txt

baltunis_b_Page_055.txt

baltunis_b_Page_081.txt

baltunis_b_Page_112.txt

baltunis_b_Page_053.txt

baltunis_b_Page_076.txt

baltunis_b_Page_005.txt

baltunis_b_Page_039.txt

baltunis_b_Page_077.txt

baltunis_b_Page_082.txt

baltunis_b_Page_013.txt

baltunis_b_Page_065.txt

baltunis_b_Page_012.txt

baltunis_b_pdf.txt

baltunis_b_Page_089.txt

baltunis_b_Page_011.txt

baltunis_b_Page_030.txt

baltunis_b_Page_109.txt

baltunis_b_Page_095.txt

baltunis_b_Page_105.txt

baltunis_b_Page_079.txt

baltunis_b_Page_009.txt

baltunis_b_Page_033.txt

baltunis_b_Page_041.txt

baltunis_b_Page_075.txt

baltunis_b_Page_094.txt

baltunis_b_Page_010.txt

baltunis_b_Page_045.txt

baltunis_b_Page_096.txt

baltunis_b_Page_046.txt

baltunis_b_Page_104.txt

baltunis_b_Page_111.txt

baltunis_b_Page_086.txt

baltunis_b_Page_003.txt

baltunis_b_Page_084.txt

baltunis_b_Page_059.txt

baltunis_b_Page_019.txt

baltunis_b_Page_049.txt

baltunis_b_Page_088.txt

baltunis_b_Page_069.txt

baltunis_b_Page_068.txt

baltunis_b_Page_008.txt

baltunis_b_Page_062.txt

baltunis_b_Page_023.txt

baltunis_b_Page_058.txt

baltunis_b_Page_067.txt

baltunis_b_Page_004.txt

baltunis_b_Page_107.txt

baltunis_b_Page_034.txt

baltunis_b_Page_093.txt

baltunis_b_Page_036.txt


Full Text












GENETIC EFFECTS OF ROOTING ABILITY AND EARLY GROWTH
TRAITS IN LOBLOLLY PINE CLONES
















By

BRIAN STEPHEN BALTUNIS


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2005

































Copyright 2005

by

Brian Stephen Baltunis















ACKNOWLEDGMENTS

I would like to thank Drs. Timothy White, Dudley Huber, Barry Goldfarb, Hank

Stelzer, John Davis, and Rongling Wu for serving on my advisory committee. Special

thanks go to Tim White and Dudley Huber for their guidance, advice, and support

throughout my entire program.

I also would like to thank the Forest Biology Research Cooperative for financial

support for this research, the members of the FBRC for establishing field trials, and

especially International Paper Company for providing the facilities for the propagation of

the rooted cuttings for this project. Many people were involved with bringing this study

together, and I would like to express my gratitude to all the faculty, staff, workers, and

fellow graduate students who helped. My appreciation especially goes to Brian Roth;

without his effort, this project could never have happened.

Finally, I would like to thank my wife, Jacqueline, for her patience and

encouragement these last five years.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iii

LIST OF TABLES .............. ............................................... ........ vi

L IST O F FIG U R E S .............. ............................ ............. ........... ... ........ viii

A B ST R A C T ................. .......................................................................................... x

CHAPTERS

1 IN T R O D U C T IO N ............................................................................. .............. ...

2 GENETIC EFFECTS OF ROOTING LOBLOLLY PINE STEM CUTTINGS
FROM A PARTIAL DIALLEL MATING DESIGN................................. ...............6.

Introduction ........................... .................................................... .6
M materials and M methods ................................................................. ....................... 8
P population ............................................................................................8
E xperim ental D esign ............................................................. ....................... 9
Statistical A n aly ses........... ................................................ .......... ..... ... .. 10
R results and D iscu ssion ......... ................................................................... 16
A average R ooting ........... ...... .............................................. ...... .......... ....... 16
Observed Variance Components ............... ........ ..................................... 18
C ausal V ariance C om ponents.......... ................. ......................... ........ ....... 21
H eritability E stim ates ............................ ................................ .... ...... ...... 22
Type B Genetic Correlations ............................. ......... ....... ........... 26
Selection for R ooting .......................... .................................. ...... ... ...... 27
C onclu sion ........... ......... ............................................. ... ............. 28

3 GENETIC ANALYSIS OF EARLY FIELD GROWTH OF LOBLOLLY PINE
CLONES AND SEEDLINGS FROM THE SAME FULL-SIB FAMILIES .............30

Intro du action ...................................... ................................................ 3 0
M materials and M methods ....................................................................... ..................33
Population ..................................... .......................... .... ......... 33
P ro p ag atio n ........................................................................................... ..... 3 4
F ield D esign ................................................. .................. 35
Statistical A n aly ses........... ................................................ .......... ..... .... .. 36









Results and Discussion ..................................... ...... .... ...... .. .......... 43
Overall Growth of Clones and Seedlings .................................... .................43
Genetic Components of Variance............................................... .................. 44
H eritability Estim ates ............................................. ... ....... .. .......... 49
Type B Genetic Correlations Between Propagule Types.............. .............. 50
Genotype x Environment Interaction ...................................... ............... 52
C o n c lu sio n ...................................... ............................... ................ 5 4

4 GENETIC GAIN FROM SELECTION FOR ROOTING ABILITY AND EARLY
GROWTH IN VEGETATIVELY PROPAGATED CLONES OF LOBLOLLY
P IN E ..................................................................................................................... 5 5

In tro d u ctio n ........................................................................................5 5
M materials and M methods ....................................................................... ..................57
Population ....................................................................57
R ooting and Field Trials.............................................. ............................ 57
Statistical A n aly ses........... ................................................ .......... ..... .... .. 58
Causal Com ponents of V ariance ................................ ............... ...................60
H eritability E stim ates ............................ ................................ .... ............6 1
Type B G enetic C orrelations ................................................................... ... ..62
G genetic G ain ................................................................... ............ 63
R results and D iscu ssion .............................. ......................... ... ........ .... ............65
Causal Com ponents of V ariance ................................ ............... ...................65
Type B G enetic C orrelations ................................................................... ... ..66
G genetic G ain ................................................................... ............ 68
C conclusion ...................................................................................................... ....... 75

5 C O N C L U SIO N ......... ...................................................................... ......... .. ..... .. 78

APPENDICES

A LOBLOLLY PINE PARTIAL DIALLEL MATING DESIGN. THIRTY-TWO
PARENTS WERE CROSSED TO GENERATE 70 FULL-SIB FAMILIES ............81

B VARIANCE COMPONENT ESTIMATES FOR ROOTING ABILITY .................. 83

C VARIANCE COMPONENT ESTIMATES FOR EARLY GROWTH TRAITS
OF LOBLOLLY PINE CLONES AND SEEDLINGS FROM THE SAME FULL-
SIB FA M IL IE S .......... ... ......... ......... ...................... ................. 86

D VARIANCE COMPONENT ESTIMATES FROM THE BIVARIATE
ANALYSES OF ROOTING AND 2ND YEAR HEIGHT ..........................................88

L IST O F R EFE R E N C E S ............................................................................. ............. 94

BIOGRAPHICAL SKETCH ............................................................. ............... 101
















LIST OF TABLES


Table p

2-1. Experimental design for 5 rooting trials of loblolly pine stem cuttings. All trials
were established in randomized complete block designs with 4 to 6 blocks and 4
to 9 ramets per clone in a row plot within each block..............................................9

2-2. Summary of rooting from 5 loblolly pine trials set over two years and three
se a so n s ................................. ........................................................... ............... 1 7

2-3. Genetic parameter estimates (standard error) for rooting of loblolly pine stem
cuttings across 5 trials. ........................... .................... ......................... 23

2-4. Type B additive and dominance variance correlations among pairs of rooting
trials for loblolly pine stem cuttings (above and below diagonal, respectively).
Standard errors are given in parentheses .................................... ................26

3-1. Location of six field trials, establishment date and total number of test trees for
each test. .......................................................... ................ 3 4

3-2. Total number of clones, full-sib families, half-sib families, average number of
clones per full-sib and half-sib family, and average number of seedlings per full-
sib and half-sib family established at the six field trials. .......................................36

3-3. Mean 1st year height, 2nd year height, height increment, and crown width by
propagule type for each of the six field trials. Although means are expressed in
meters, analyses were conducted using measured traits in centimeters...............44

3-4. Genetic parameter estimates for lst year height, 2nd year height, height increment,
and crown width by propagule type across all six trials. Standard errors are
given in parentheses. ........................ ...................... ... ...... .. .... ........... 46

3-5. Genetic correlations between propagule types for 1st year height, 2nd year height,
height increment, and crown width at the parental ( ro ) and full-sib family

( ropF levels ................................................51

4-1. Means, variance component estimates, heritabilities, and genetic correlations
from the bivariate analysis of rooting ability and 2nd year height. Standard errors
are given in parentheses. ........................................ ............................................67









4-2. The total genetic correlation (?, ) between rooting ability and 2"d year height
from analyses of a single rooting trial and field trial. Shaded values indicate the
rooting trial in which cuttings originated from for their respective field trials.
Standard errors are given in parentheses ....... ........ ................................... 70

4-3. The number of full-sib families (half-sib families) and average number of clones
per full-sib family (half-sib family) selected from selecting 10% or 1% of the
top clones using the combined selection index. ....................................... .......... 77

B-1. Observed variance component estimates for rooting of loblolly pine stem
cuttings from single-trial analyses....................................... ......................... 83

B-2. Observed variance component estimates for rooting of loblolly pine stem
cuttings from pair-wise-trial analyses. ........................................ ............... 84

B-3. Observed variance components for rooting ability from the across-trial analysis
using all five rooting trials. ............................................. ............................. 85

C-1. Observed variance components for loblolly pine clones from the across-trial
analyses of 1st year height, 2nd year height, height increment, and crown width.
A separate error variance was modeled for each trial. ...........................................86

C-2. Observed variance components for loblolly pine seedlings from the across-trial
analyses of 1st year height, 2nd year height, height increment, and crown width.
A separate error variance was modeled for each trial. ...........................................87

D-1. Observed variance component estimates from the bivariate analyses of rooting
ability from SpringO1 and 2nd year height from each of the field trials ..................88

D-2. Observed variance component estimates from the bivariate analyses of rooting
ability from Summer01 and 2nd year height from each of the field trials. ...............89

D-3. Observed variance component estimates from the bivariate analyses of rooting
ability from Winter02 and 2nd year height from each of the field trials .................90

D-4. Observed variance component estimates from the bivariate analyses of rooting
ability from Spring02 and 2nd year height from each of the field trials .................91

D-5. Observed variance component estimates from the bivariate analyses of rooting
ability from Summer02 and 2nd year height from each of the field trials. ...............92

D-6. Observed variance component estimates from the bivariate analysis of rooting
ability using all five rooting trials and 2nd year height using all six field trials. ......93















LIST OF FIGURES


Figure page

2-1. The proportion of the additive (h0 ), dominance (d, ), and epistasis (o )
genetic variances on the observed binary scale for rooting of loblolly pine stem
cuttings from each of the five separate trials (biased heritabilities) and from the
combined analysis of all five trials. Standard error bars for broad-sense
heritability estim ates are included ......................................................................... 19

2-2. Narrow-sense (h2 ) and broad-sense (H ) heritability estimates for rooting of
loblolly pine stem cuttings transformed to the underlying normal scale using the
threshold model of Equation 2-11. Standard error bars are included....................25

2-3. Full-sib family mean (H2 ) and clonal mean (H 2) heritability estimates for
rooting success of 2,200 clones from 70 full-sib families of loblolly pine.
Standard error bars are included........................................ ........................... 25

3-1. The proportion of additive (add) and nonadditive (na) genetic variance
components for clones and seedlings across the six field trials, where h2 = add
and H2 = add + na: a. 1st year height, b. 2nd year height, c. Height increment,
and d. Crown width. Standard error bars for broad-sense heritability estimates
are in clu ded ....................................................... ................. 4 8

3-2. Rank-rank plots showing type B genetic correlations between clones and
seedlings from Test B based on: a. Parental BLUP values, b. Full-sib family
BLUP values, where full-sib BLUP values are equal to the sum of the predicted
general combining ability for each of the two parents plus the predicted specific
com bining ability of the cross. ............. ........... ..... ..... ........ ............... 53

4-1. The genetic gain in rooting ability (%) over the population mean for deployment
of the best half-sib family, full-sib family, best clone from the best ten full-sib
families, and the single best clone when selecting for rooting ability or 2nd year
h e ig h t .................................................................................................................. 7 1

4-2. The genetic gain in 2nd year height (%) over the population mean for deployment
of the best half-sib family, full-sib family, best clone from the best ten full-sib
families, and the single best clone when selecting for rooting ability or 2nd year
height ...............................................................................72









4-3. Responses to selection in rooting ability and 2nd year height with various
selection indices for two clonal deployment options: 10% of clones selected and
1% of clones selected. .............................................. ............... .... ...... ...... 75















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

GENETIC EFFECTS OF ROOTING ABILITY AND EARLY GROWTH
TRAITS IN LOBLOLLY PINE CLONES

By

Brian Stephen Baltunis

December 2005

Chair: Timothy L. White
Major Department: Forest Resources and Conservation

Loblolly pine is the most important commercial tree species in the southern United

States with over 1.1 billion seedlings planted annually. With elite genotypes becoming

available, several forest industry companies in the southeastern United States are

developing rooted cutting and somatic embryogenesis programs aiming towards

deployment of tested clones or families. However, before clones can be deployed,

sufficient data need to be collected on the population in order to have reliable information

about the clones for deployment decisions.

This dissertation reports on the genetic effects of rooting ability and early growth

traits in nearly 2,200 clones of loblolly pine from 70-full-sib families. More than

239,000 stem cuttings were set in five rooting trials over two years. Overall rooting

success across the five trials was 43%, and significant seasonal effects were observed.

Heritability of rooting ability was estimated both on the observed binary scale and on the

transformed underlying normal scale.









Rooted cuttings from these trials along with seedlings from the same full-sib

families were established at several sites, and early growth traits through age two were

compared between propagule types. All growth traits demonstrated genetic variation,

and parental and full-sib family rankings were similar for both propagule types.

However, estimates of dominance genetic variance in the seedling population appear to

be inflated at the expense of additive effects due to a lack of randomization of seedlings

prior to field establishment. Little genotype x environment interaction was observed

across sites for all traits.

A successful clonal forestry program for loblolly pine based on rooted cutting

technology needs to consider selection for both rooting ability and subsequent growth.

There was a positive genetic correlation between rooting ability and 2nd year height at the

parental, full-sib family, and clonal levels indicating that selection for one trait will also

lead to improvement of the other. The genetic gains in rooting ability and 2nd year height

associated with several selection and deployment strategies are discussed. Moderate to

high family and clonal mean heritabilities, moderate to high type B correlations, and

substantial among-family and among-clone genetic variation indicate the potential for

increasing rooting efficiency and improving growth.














CHAPTER 1
INTRODUCTION

Loblolly pine (Pinus taeda L.) is the most important commercial tree species in the

United States with over one billion seedlings planted annually (McKeand et al. 2003).

Most commercially important tree species remain relatively undomesticated, and loblolly

pine is no exception. Genetic improvement of loblolly pine has been occurring since the

1950's in several tree improvement programs. There are three cooperative tree

improvement programs in the southern United States that focus on improvement of

southern pine species including loblolly pine: the Cooperative Forest Genetics Research

Program (CFGRP), North Carolina State University-Industry Cooperative Tree

Improvement Program (NCSUITIP), and the Western Gulf Forest Tree Improvement

Program (WGFTIP). Loblolly pine tree improvement programs in the South are

beginning their 3rd generation of breeding with gains in volume per unit area up to 30%

over unimproved loblolly pine (McKeand et al. 2003).

Long-term tree improvement programs aim to increase the population mean

breeding value of a few key traits such as stem volume, disease resistance, and wood

properties through breeding and selection of superior genotypes. Loblolly pine tree

improvement programs are based on recurrent selection for general combining ability,

which captures only the additive portion of the genetic variance. However, the

nonadditive portion of genetic variation, dominance and epistasis, may be important

components of variation for traits.









Both additive and nonadditive genetic variation can be captured by deploying full-

sib families or clones. However, deployment decisions should be based on reliable

information. Field trials established with full-sib seedlings allow the genetic variation to

be partitioned into additive and nonadditive components. These trials not only provide

ranks of parents or individuals for selection, but also of full-sib families in order to

provide information for making deployment decisions.

In any given generation of breeding, maximum genetic gains can be achieved in the

deployment population by capturing all of the genetic variation through operational

propagation and deployment of selected clones. However, clonal forestry is not a

breeding method to develop better genotypes. Clonal forestry is a method to mass-

produce well-tested genotypes. Short-term genetic gains may be maximized through

deployment of well-tested clones, but long-term gains need to involve both clonal

selection and recurrent selection for additive genetic variation through repeated selection

and breeding.

Clonal tests derived from full-sib families do provide an opportunity to estimate

additive and nonadditive components of variance associated with a particular trait or set

of traits. However, tests should be designed with a sufficient genetic structure in order to

precisely quantify the genetic variation. For example, Frampton and Huber (1995)

reported that they had low power in partitioning the genetic variation because of the lack

of a mating design among the parents of the full-sib crosses in a loblolly pine clonal

study. In another study comparing clones from 30 full-sib families derived from two

disconnected 4x4 factorials, Paul et al. (1997) concluded that future clonal studies should

include more parents in the mating design. Although Isik et al. (2003) estimated the









genetic variances from both clones and seedlings from the same nine full-sib families of

loblolly pine, they identified a weakness of their study in that there were a limited

number of parent trees used in the mating design. Finally, Frampton and Foster (1993)

warned that interpretation of the results may be difficult for studies that only include

seedlings and cuttings from a common checklot to be compared to the clonal propagules

from select parents and families. In this case, any differences in the field performance

because of propagule type may be confounded with the differences in genetic

improvement (Frampton and Foster 1993).

Clonally replicated progeny trials have been suggested as part of a tree

improvement strategy for radiata pine (Jayawickrama and Carson 2000) and for loblolly

pine (Foster and Shaw 1987; Isik et al. 2004; Byram et al. 2004) for a number of reasons.

First, field trials established with clonally replicated progeny allow for further

partitioning of the genetic variation into the additive, dominance, and epistatic genetic

variation (Foster and Shaw 1988). Second, clonally propagated seedlings can provide

genetic information more efficiently and with greater precision than zygotic seedling

progeny (Burdon and Shelbourne 1974; Isik et al. 2004). Finally, clonal testing and

selection strategies can provide greater gain than seedling options (Shaw and Hood 1985;

Mullin and Park 1994; Isik et al. 2004).

Based on current technologies, several forest industries in the southeastern United

States are pursuing clonal forestry programs with loblolly pine (Weber and Stelzer 2002).

In the initial stages of these clonal forestry programs, forest managers needed assurance

that the clonal propagules' growth corresponded to that of seedlings. Therefore, most of

the earlier studies were designed to test whether cuttings grew similarly to seedlings.









Based on those results, it is generally accepted that cuttings rooted from juvenile stock

plants grow and perform comparably to seedlings. For example, Foster et al. (1987)

reported that loblolly pine rooted cuttings should perform comparably to seedlings when

the cuttings come from vigorous juvenile stock plants. In addition, McRae et al. (1993)

concluded that for loblolly pine there were no significant differences between seedlings

and rooted cutting propagules from common checklots through five years of growth.

Similar results were obtained by Frampton et al. (2000) where they reported no

significant differences between the means of rooted cuttings and seedlings for height,

diameter at breast height, and volume through six years in the field.

Trials established with clones and seedlings from the same families provide an

opportunity for comparing both half-sib and full-sib family performances for both

propagules. Genetic correlations between propagule types can provide further assurance

that selections made through traditional tree improvement activities for recurrent

selection for general combining ability can also be used successfully in breeding families

to test in a clonal forestry program. Although a number of studies have been reported

comparing rooted cutting and seedlings, very few have been designed to estimate the

genetic correlation for a trait between propagule types.

Two main criteria need to be met prior to operational deployment of loblolly pine

clones. First, loblolly pine clones must perform well, e.g., meet the selection criteria for

the desired traits. This involves the accumulation of reliable data for the clones from

greenhouse screening, field trials, etc. Second, the selected clones have to be propagated

in large enough numbers for deployment. For a rooted cutting based clonal program, this

involves bulking up the number of hedges (ramets) of a particular clone or group of









selected clones through serial propagation and then producing reforestation stock

efficiently from the bulked-up clones. Only those clones that can be propagated easily

and in sufficient numbers will be economically feasible for deployment. A clone that

grows well in the field but roots poorly may not be economically feasible to include in a

clonal program based on rooted cutting technology.

This dissertation is unique in that a complex genetic structure was utilized in order

to increase the power in quantifying the genetic variation associated with several traits in

loblolly pine clones and seedlings. In Chapter 2 rooting ability was assessed for nearly

2,200 clones of loblolly pine from 70 full-sib families derived from a partial diallel

mating design in order to estimate genetic parameters associated with rooting. More than

1,200 of these clones along with zygotic seedlings from the same full-sib families were

established together on multiple sites across the southeastern United States. Genetic

parameter estimates are compared between propagule types for early growth traits in

Chapter 3. Finally, selection for both rooting ability and field growth is addressed in

Chapter 4.














CHAPTER 2
GENETIC EFFECTS OF ROOTING LOBLOLLY PINE STEM CUTTINGS FROM A
PARTIAL DIALLEL MATING DESIGN

Introduction

Loblolly pine is the most important commercial tree species in the southern United

States with over 1.1 billion seedlings planted annually (McKeand et al. 2003). There are

three cooperative tree improvement programs in the southern United States that focus on

improvement of southern pine species including loblolly pine: the Cooperative Forest

Genetics Research Program (CFGRP), North Carolina State University-Industry

Cooperative Tree Improvement Program (NCSUITIP), and the Western Gulf Forest Tree

Improvement Program (WGFTIP). Loblolly pine tree improvement programs in the

South are beginning their 3rd generation of breeding with gains in volume per unit area up

to 30% over unimproved loblolly pine (McKeand et al. 2003).

Long-term tree improvement programs aim to increase the population mean

breeding value of a few key traits through breeding and selection of superior genotypes.

These programs are based on recurrent selection for general combining ability which

captures only the additive portion of the genetic variance. In any given generation of

breeding, maximum gains can be achieved in the deployment population by capturing all

of the genetic variation (additive and nonadditive components) through operational

propagation of selected clones. With elite genotypes becoming available, several forest

industries in the southeastern United States are developing rooted cutting programs for









loblolly pine aiming towards deployment of tested clones or families (Weber and Stelzer

2002).

Clonal tests derived from full-sib families provide an opportunity to estimate

additive and nonadditive genetic components of variance associated with a particular trait

or set of traits (Isik et al. 2003; Isik et al. 2004). In clonal field trials traits of interest

may include height, volume, wood quality, and disease resistance. However, in order to

establish clonal field trials the clones must first be propagated. Therefore, clonal rooting

trials are important for estimating genetic variance components associated with rooting.

Previous rooting studies of loblolly pine have been relatively small in size ranging

from several hundred (Goldfarb et al. 1998; Foster et al. 2000) to several thousand

cuttings (Foster 1990; Anderson et al. 1999). Many studies contained a small number of

families from factorial mating designs (Goldfarb et al. 1998; Anderson et al. 1999;

Cooney and Goldfarb 1999; Frampton et al. 1999) and few have been designed to

estimate genetic parameters associated with rooting in loblolly pine (Foster 1978; Foster

1990; Anderson et al. 1999). The current study is unique in that a large number of

cuttings were set in each trial (> 34,000), and rooting was assessed on nearly 2,200 clones

of loblolly pine from 70 full-sib families derived from a partial diallel mating design in

order to estimate genetic parameters associated with rooting. The objectives of the study

were to (i) evaluate the rooting ability of stem cuttings from nearly 2,200 loblolly pine

clones, (ii) determine the causal components of variance in rooting of stem cuttings, (iii)

assess heritability estimates for rooting from five trials both on the observed binary scale

and the underlying normal scale, and (iv) determine the Type B genetic correlations for









both additive and dominance genetic effects to measure the correspondence in rooting

performance across five setting dates.

Materials and Methods

Population

The parental population for this study was selected from the Loblolly Pine Lower

Gulf Elite Population (LPLGEP) which consists of selections from all three southern pine

tree improvement cooperatives: CFGRP, NCSUITIP, and WGFTIP. Twenty 1st

generation and ten 2nd generation selections representing the Atlantic Coastal Plain,

Florida, and Lower Gulf provenances were selected from this population. Two additional

slow-growing parents were included to provide linkage with another study. These

parents were crossed in a circular diallel mating design (Appendix A) with some

additional off-diagonal crosses, resulting in a total of 70 full-sib families. On average

each parent was involved in approximately four crosses.

Seeds from the 70 families were sown in March 2000 into Ray Leach SuperCells

(Stuewe and Sons, Corvallis, OR). The seedlings were grown in a greenhouse at

International Paper Company's (IPC) facility in Jay, FL, and after three months of growth

the seedlings were pruned back to a height of about 10-12 cm. Approximately 32

seedling hedges (ortets) per full-sib family were transplanted into 3-gallon containers and

given unique clonal identifications in September 2000. The hedges were repeatedly

sheared in order to minimize the effects of maturation and increase the number of shoots

available for the rooting trials. The ortets were randomized in a containerized hedge

orchard in order to reduce spurious C effects at the family level. However, C effects at

the clonal level could not be accounted for, because all cuttings originated from a single

seedling ortet.









Table 2-1. Experimental design for 5 rooting trials of loblolly pine stem cuttings. All
trials were established in randomized complete block designs with 4 to 6
blocks and 4 to 9 ramets per clone in a row plot within each block.
No. of No. of
o No. of No. of Total
Tnal Date set plots per cuttings
Trial Date set families clones lotsper cuttings cuttings
clone per plot
SpringOl May 7- 70 2194 4 4 34,707
11, 2001
SummerO July 2-6, 70 2157 5 4 43,048
2001
Jan 14-
Winter02 Jan61 1648 6 5 49,315
18, 2002
Apr 29-
Spring02 May 3, 61 1254 6 9 67,059
2002
June 24-
Summer02 Ju 24- 61 947 6 9 45,108
28, 2002


Experimental Design

Two rooting trials were conducted in 2001 and three were conducted in 2002. Stem

cuttings between 3 and 8 cm in length were harvested from the seedling ortets in May

2001, July 2001, January 2002, April/May 2002, and June 2002, for trials SpringOl,

SummerOl, Winter02, Spring02, and Summer02, respectively. Cutting size was

relatively consistent within any trial, and the cuttings set in Winter02 were the smallest.

The experimental design differed among the trials due to number of families, clones and

available cuttings from each ortet (Table 2-1). The reduction in the number of families

and clones between the first and last rooting trials was a result of a number of factors.

First, random hedge mortality was a major factor: disease, repeated severe pruning, and

uneven watering (inadequate) all contributed to the random loss of hedges. Since

cuttings of a clone originated from a single ortet, by default mortality resulted in a

truncated population for future rooting trials. Second, not all hedges produced an

adequate number of cuttings at every harvest. The primary objective of the three settings









in 2002 was to produce propagules for field trials. Therefore, clones that were not

producing enough shoots to ultimately be planted across six field sites were culled

regardless of rooting frequency. This resulted in the reduction in the number of families

in the last three rooting trials because of too few surviving clones in some of the families

to meet the goal for field designs. We were striving for a balanced field design with 15

clones from each of 61 full-sib families.

Cuttings were randomly set in 4-, 5-, or 9-cutting clonal row plots (Table 2-1) into

pre-formed plugs consisting of peat moss, perlite, and a binding resin. Each plug was

approximately 13 mL in volume and was held by the V-13 HIKO tray (135 cells; Stuewe

and Sons, Corvallis, OR). Cuttings were either treated prior to setting with a 1.0%

indole-3-butyric acid and 0.5% napthalene-1-acetic acid (NAA) basal dip or after setting

using a foliar NAA application according to IPC protocols. Trays contained 15 to 30

clones depending on the trial and were randomly placed in an environmentally controlled

greenhouse. There were 4 to 6 complete replications depending on trial (Table 2-1).

Rooting assessments were made 9-weeks after setting for both the SpringOl and

SummerOl trials. Cuttings were measured for presence (1) or absence (0) of roots.

Cuttings with a root > 1 mm were considered rooted (Goldfarb et al. 1998, Foster et al.

2000). For trials Winter02, Spring02, and Summer02 assessments were at 11-weeks

following setting. Cuttings in these trials were also scored for presence or absence of

roots. However, only cuttings that had at least one visible root on the exterior of the plug

regardless of length were considered rooted.

Statistical Analyses

For binomial traits, such as rooting, the unit of analysis can be the individual

observations (Huber et al. 1994; Dieters et al. 1996) or plot means combined with a









transformation such as arcsin or logistic (Sohn and Goddard 1979; De Souza et al. 1991).

We chose to analyze the observed 0,1 data for several reasons. First, REML estimation

of variance components has been shown to be robust to violations of the underlying

normality assumptions (Banks et al. 1985; Westfall 1987) suggesting that analyses using

individual observations of binary data yields satisfactory results. Second, simulation

studies have shown that the use of individual observations is superior to the use of plot

means in REML, and that these variance component estimates perform well across

mating designs and imbalanced data (Huber et al. 1994). Huber et al. (1994) showed that

a lower variance among estimates was obtained using individual observations as

compared to plot means and that this advantage increased with increasing imbalance.

Third, when heritability is low and the incidence close to 50%, there is little difference

between heritability estimates on the binary and transformed scale (Dempster and Lerner

1950). In fact these two estimates are equivalent when the incidence is exactly 50% for

low heritability traits. Lopes et al. (2000) demonstrated that the Dempster and Lerner

(1950) threshold model closely estimates the true underlying heritability at incidences

between 25% and 75% for traits with low heritability (h2 < 0.3). Finally, Lopes et al.

(2000) also demonstrated (for traits with low heritability) that heritability estimates from

the observed binary data without transformation of data result in predicted gain close to

the realized gain, while transformations can suffer from issues of back transformation

when one wishes to predict gains on the original scale.

All variance components for rooting ability of loblolly pine stem cuttings were

estimated using the individual binary observations using REML estimation for each of

the 5-rooting trials using GAREML (Huber 1993). However, upwardly biased estimates









of genetic variances result when variance components are estimated from single-site

(trial) analyses since the estimated genetic variance also contains the genotype x

environment interaction (Comstock and Moll 1963). Therefore, across-trial analyses

were performed to separate the genotype x environment interactions in order to remove

this bias.

Yklmno =/ + + R) + trayk(j()) + gca1 + gcam + sCalm + c(fam)n(lm)
[2-1]
+ t gca', + t gca ,, + t scale + t c(fam),n(lm) + r famjn(,) + errorlklmno

where yzjklmno is the rooting response (0 or 1) of the 0th ramet of the nth clone within the

Imth full-sib family in the kth tray within thejth rep of the ith trial

,u is the population mean

T, is the fixed effect of trial

R,j( is the fixed effect of rep

tiv ,, ,, is the random variable tray (incomplete block) ~ IID(0, RA )

gcal and m is the random variable female (/) and male (m) general combining ability (gca)

^2
~IID(0, crGCA)

scalm is the random variable specific combining ability (sca) IID(0, ^SA )

c(fam)n(m) is the random variable clone within family ~ IID(0, ^ONE^ )

t'giI 'im is the random variable test by female gca and test by male gca interaction ~

IID(O, TESTxGCA)

t*scalm is the random variable test by full-sib family interaction ~ IID(O, ^ESTxFM )

t*c(fam),n(Im) is the random variable test by clone interaction IID(0, ESTxCLONE)

S"t,11n, is the random variable rep by family interaction ~ IID(0, ^rEPxFA









errorklmno is the random error which includes among plot and within plot ~

IID(O, OR OR).

The single trial model is identical except all model factors with subscript 'i' are removed

(sources involving test).

Genetic parameters were estimated and standard errors were calculated according

to Foster and Shaw (1988) using the appropriate variance components from the individual

or across trial model. Estimates of additive and dominance genetic variance are upwardly

biased because they are confounded with fractions of epistasis (Cockerham 1954).

Epistastic genetic variance is also only approximated because it contains only a fraction

of the total epistasis plus any C effects, if they exist.

[2-2] V, = 48 = + V +-V V +... = estimate of additive genetic variance

[2-3] VD = 4 VD +V + AD + DD +... = estimate of dominance genetic

variance

[2-4] VI =2 cLE-2 2 -3C2 = AA+ D DD + ... = estimate of epistatic

genetic variance

[2-5] V = 2 -cA + A +C L = estimate of total genetic variance


[2-6]

p 2 22 2 2 2 "2 "2 "2
S= 2G + SCA + CLONE + 2TESTxGCA +TESTxFAI TESTxCLONE -F + O REPxFAMI F O'ERROR

phenotypic variance for across trial model (the phenotypic variance from the individual

trial model is the same but drop 2ESTxGCA, TESTXFAM 2 and TESTXCLONE

Biased and unbiased heritability estimates for rooting based on observed 0,1 data

were derived using the estimated variance components from the single and across trial











models, respectively. In addition the proportion of dominance (d2) and epistasis (I2)

were estimated. Standard errors of these estimates were calculated using a Taylor series

expansion (Kendall and Stuart 1963; Namkoong 1979; Huber et al. 1992; Dieters 1994).

[2-7]


4&2
j2 4 GCA
0,1 ^2 2 2 2 ^ 2 2 ^2 ^2 ^2
2cGCA SCA CLONE + 2"TESTxGCA + (TESTxFAM + (TESTxCLONE + "REPxFAM -+ ERROR


across-trial narrow-sense hertibility based on observed binary data

[2-8]

^2 ^2 ^2
2 2 GCA + SCA +CLONE
0,1 2 2 ^2 ^2 2 ^2 + 2 + 2 + 2 ^2
2cGCA SCA + -CLONE + (7TESTxGCA (TESTxFAM (7TESTxCLONE (REPxFAM -V ERROR


across-trial broad sense heritability based on observed binary data

[2-9]

4 ^ 2
21 __ SCA
0,1 2 ^2 ^2 ^2 ^2 ^2 ^2 ^2
2cGCA SCA CLONE + TESTxGCA TESTxFAM TESTxCLONE + REPxFAM ERROR

across-trial dominance proportion

[2-10]

^2 ^2 ^2
'2 C_ CLONE 2 GCA SCA
0,1 22 2 2 2 2 "2 +2 "2
2a GCA + SCA CLONE + (7 TESTxGCA + (TESTxFAM (7TESTxCLONE REPxFAM ERROR


across-trial epistatic proportion.

The main problem with calculating heritability on the observed 0,1 data is that the


relationship between h2 on the observed scale and h2 on the underlying normal scale

depends on the mean incidence (e.g., % survival, % infected individuals, rooting

percentage), and therefore the conversion results in a biased estimate (Van Vleck 1972).

However, this is not a problem for low heritability traits with intermediate incidences










(Lopes et al. 2000). In order to make valid comparisons to heritability estimates in other

rooting trials that have different mean rooting percentages h0, needs to be transformed to


h2 on the underlying normal scale. Therefore, narrow- and broad-sense heritability

estimates on the observed 0,1-scale were transformed using a threshold model to an

underlying normal scale (Dempster and Lerner 1950).


[2-11] h2 ()( p), where
z

h2 is the heritability on the underlying normal scale

p is the rooting percent

z is the ordinate of the normal density function which corresponds to probability p.

Full-sib family mean heritability and clonal mean heritability for rooting were

estimated for both the single- and across-trial analyses. Standard errors for these

estimates were calculated using Dickerson's Method which assumes the phenotypic

variance (denominator) is a known constant (Dickerson 1969).

[2-12]

2 2&c +GCA + SCA
FS 22 2 2 2 2
22 ( 2 + 'CLONE 'TESTxGCA TESTxFAM TESTxCLONE REPxFAM ERROR
2 GCA + SCA + +
c t t tc tr ctrn
= across-trial family mean heritability, where, c = harmonic mean number of clones per

family, t = number of trials, r = harmonic mean number of reps per test, and n = harmonic

mean number of ramets per clone per plot.










[2-13]

22 + 2 +(2
2 GCA +SCA CLONE
CL 2 22 2 12 .2
2 (j 2 + 2 + S j2 | CO 'TESTxGCA + TESTxFAM UTESTxCLONE 'REPxFAM + ERROR
-GCA + USCA -CLONE t + + t
t t t tr trn

= across-trial clonal mean heritability.

Type B genetic correlations for rooting across all 5 trials were estimated for both

additive and nonadditive components (Yamada 1962; Burdon 1977). Standard errors of

Type B correlations were calculated using the Taylor Series Expansion method.

"2
[2-14] rB 2 g ,
g ge

where rB is the estimate of the Type B genetic correlation, !^2 is the genetic variance


component (either additive or dominance), and ^ 2 is the G x E interaction (additive and


dominance).

Results and Discussion

Average Rooting

A total of over 239,000 stem cuttings from nearly 2,200 clones of loblolly pine

were set in five rooting trials. Overall rooting across the five trials was 43% and is

comparable to other rooting studies involving loblolly pine. Goldfarb et al. (1998)

reported 44% rooting of loblolly pine stem cuttings from 400 seedling hedges from one

open-pollinated family. Over four rooting trials, Anderson et al. (1999) reported 33%

rooting from 90 clones of loblolly pine from 9 full-sib families. Foster (1990) reported

overall rooting from three settings of 42% for 546 clones of loblolly pine derived from 54

full-sib families. However, studies with fewer families of loblolly pine have yielded









substantially higher rooting percentages (Cooney and Goldfarb 1999; Murthy and

Goldfarb 2001; LeBude et al. 2004).

Rooting of loblolly pine cuttings was variable over the five trials (Table 2-2).

Spring cuttings rooted at greater than 50%, while summer cuttings in the two summer

trials averaged 38% and 24% respectively. Winter cuttings were intermediate at 45%

rooting. Broad inference linear contrasts were constructed using the estimates of the

fixed effects in order to test seasonal rooting responses. Cuttings set in the two spring

trials rooted at a significantly greater frequency then cuttings set in the summer trials (p <

0.0001). This implies that we would always expect a greater rooting percent in spring

settings than in summer settings under this propagation system.

Table 2-2. Summary of rooting from 5 loblolly pine trials set over two years and three
seasons.
Trial Rooting Half-sib Full-sib
Tnal r. ir. Clone range
% family range family range
SpringO1 54% 36-70% 28-77% 0-100%

SummerO1 38% 24-54% 18-69% 0-100%

Winter02 45% 17-60% 17-67% 0-100%

Spring02 51% 36-67% 28-75% 0-98%

Summer02 24% 9-43% 8-53% 0-89%



Seasonal rooting responses for loblolly pine stem cuttings have been observed in

other studies. Early rooting trials of loblolly pine cuttings reported best rooting from

cuttings set from September through January (Cech 1958; Reines and Bamping 1960;

Grigsby 1962; Marino 1982). These early experiments concluded that increased

temperatures in the greenhouse in spring and summer trials decreased rooting. In fact,









Cech (1958) reported a 3-fold increase in rooting under cool conditions rather than under

warm conditions. However, Foster et al. (2000) observed an overall rooting of 50% for a

rooting trial established in March, while only 20% rooting for a trial established in

September. They hypothesized that the reduction in September rooting was due to a

decrease in metabolic activity due to the decrease in photoperiod. Rowe et al. (2002a;

2002b) reported trends in rooting similar to those of the current study. They observed

59% rooting for spring cuttings versus 40% rooting for winter cuttings and 35% rooting

for summer cuttings. Cooney and Goldfarb (1999) also reported high rooting percentages

for spring cuttings (62% and 83% in two successive years). In contrast, Murthy and

Goldfarb (2001) reported higher rooting percentages for winter cuttings (85%) than for

spring cuttings (60%). Winter cuttings often take longer to root but overall rooting may

not be different than spring cuttings. Perhaps the slight reduction in rooting frequency

for the Winter02 setting versus the two spring settings was a function of rate of rooting.

The reduction in rooting seen here for summer settings may be a result of increased

temperatures during the collection and propagation phases of the experiment. The higher

temperatures and humidity experienced during summer months may have resulted in an

increased abundance or activity of pathogens, and hence a higher rate of decay was

observed in the two summer trials. The time delay between collection and setting of

cuttings may also have contributed to this reduction in rooting. Murthy and Goldfarb

(2001) reported a decline in rooting percentage with increasing drying time of cuttings.

Observed Variance Components

Variance components were estimated for all single- and across-trial analyses

(Appendix B). Even with the reduction in the number of clones and families throughout

the study, there was no apparent reduction in the variance component estimates over time









as evidenced by parameter estimates, e.g., additive genetic variance estimates were

relatively constant over time (Figure 2-1). The variance associated with general

combining ability was 2% of the total phenotypic variation associated with rooting. Half-

sib family rooting percentages ranged from a high of 36-70% for SpringO1 to a low of 9-

43% for Summer02 (Table 2-2). The proportion of the total variation in rooting that was

accounted for by specific combining ability was only 0.3-1.1%. Full-sib family means

for rooting for the two spring settings ranged from 28-76%. Although overall rooting

was greater for these two spring trials, the net difference in the range of family means

was approximately the same (-45-51%). Similar ranges in family mean rooting

percentages have been reported (Foster 1990; Anderson et al. 1999).



0.25

0.2 T I-lo2

0.15 -
H 0,1 0
0.1 -

0.05 1


SpringOl SummerOl Winter02 Spring02 Summer02 All
Trial


Figure 2-1. The proportion of the additive (0 ), dominance (d, ), and epistasis (1 )
genetic variances on the observed binary scale for rooting of loblolly pine
stem cuttings from each of the five separate trials (biased heritabilities) and
from the combined analysis of all five trials. Standard error bars for broad-
sense heritability estimates are included.









Clones within families accounted for approximately 10-17% of the total variation

in rooting in the five trials (Appendix B). Foster (1990) reported that the clone within

family source of variation in rooting of 546 clones of loblolly pine was 3.7 % of the total

variation. However, in another study involving cuttings of loblolly pine, the among clone

source of variation accounted for nearly 22% of the total variation (Foster 1978). In the

current study, the variance associated with clone within family was 4.5-8.4 times greater

than the gca variance and 13-56 times greater than the sea variance based on single-trial

analyses.

Just as there was a large range in rooting among families, there was also a large

range in rooting among clones within families (Table 2-2). Rooting for clones within

families ranged from 0-100% for the first three trials, and ranged from 0-98% and 0-89%

in the last two trials, respectively. Anderson et al. (1999) observed ranges in rooting

frequency for clones within family similar to those observed in this study. Foster (1990)

reported significant variation for clones within family with rooting percentages ranging

from 6.7-85.0%.

A large variation in the rooting environment has been reported in many rooting

experiments of loblolly pine and other species, with an error variance ranging from 46-

71% (Foster 1978; Foster 1990; Sorensen and Campbell 1980; Cunningham 1986). In

the current study, the majority of the observed variance in rooting was attributable to the

error variance. The error variance accounted for 76.3-83.7% of the total variance

observed in the five trials. Each rooting trial was spread over an entire greenhouse due to

the large size of the experiments. Apparently, the rooting environment was not uniform

throughout the greenhouse. There are many factors that can contribute to a variable









rooting environment. Differential temperature gradients, unequal airflow, unequal

misting of cuttings, edge effects, and disease incidence can all contribute to a variable

rooting environment.

Causal Variance Components

The observed variance components were used to estimate additive, dominance and

epistatic variances using equations 2-2, 2-3, and 2-4, respectively. Additive genetic

variance was approximately 2 to 6 times larger than the dominance genetic variation for

rooting in the five trials (Figure 2-1). In Foster (1990), the additive genetic variance was

also about 6 times larger than the dominance genetic variance. However, Anderson et al.

(1999) found three times and Foster (1978) found 2.2 times greater dominance genetic

variance in rooting compared to the additive variance in loblolly pine stem cuttings. The

epistatic genetic variance was also estimated in the current study and was approximately

0.44 to 1.49 times as much as the additive genetic variance. The lowest amount of

epistasis was observed for the winter setting and was the only setting that had a

nonadditive to additive ratio less than one.

C effects can lead to upwardly biased estimates of total genetic and nonadditive

genetic components of variance when analyzing clonal data (Libby and Jund 1962). If C

effects are present, then total genetic variation associated with clones will be

overestimated (Libby and Jund 1962). Significant C effects are likely to occur in traits

that are measured soon after propagation (such as rooting traits, early shoot growth, etc.),

but apparently lessen for traits measured at later times (Libby and Jund 1962). Foster et

al. (1984) used a secondary cloning approach to separate C effects from the genetic

variance in rooting of western hemlock cuttings. They found significant C effects

associated with rooting and that these non-genetic effects biased the genotypic values of









clones. However, low or non-significant C effects for rooting of tamarack and balsam

poplar cuttings have been reported (Farmer et al. 1989; Farmer et al. 1992). In the

current study, estimates of epistatic genetic variance components are confounded with C

effects, because ramets of a clone came from the original seedling ortet. However,

estimates of additive and dominance genetic effects are not confounded with C effects,

because the clones within families were randomized in the hedge orchard.

Heritability Estimates

Rooting of loblolly pine stem cuttings was weakly controlled by additive effects.

Individual tree narrow-sense heritability using the observed variance components (h1 )

ranged from 0.075 to 0.089 in the five separate trials (Figure 2-1). These estimates are in

agreement with Foster (1978) who reported h2 as 0.07 for rooting percentage of loblolly

pine. A slightly higher h2 was reported by Foster (1990). However, Anderson et al.

(1999) reported h2 of 0.26 for rooting percentage in loblolly pine stem cuttings.

However, all of these previous studies of loblolly pine rooted cuttings analyzed data

based on plot rooting percentage, as opposed to using 0,1 data as in this study, and this

leads to an increased estimate of heritability due to the reduction of the impact of the

within plot portion of the error variance in any plot means or plot percentage analysis for

rooting. When the data in the current study were analyzed based on rooting percentage,

2 was 0.18 (from SpringO1, data not shown), which falls in the middle of the range

previously reported.

The unbiased estimates of individual tree narrow-sense heritability (h ) using the

0,1 data from the pairwise test analyses ranged from 0.045 to 0.074. When all of the data









from the 5 trials were analyzed together, hk, was 0.051 (Table 2-3; Figure 2-1). The

proportion of dominance (d2 ) was estimated for each of the trials. The upwardly-biased

estimates for this parameter ranged from 0.014 to 0.044 (Figure 2-1). When all of the

data were analyzed together from the 5 trials, d2 was 0.018 (Table 2-3; Figure 2-1). The

epistatic proportion (i, ) was also estimated and ranged from 0.032 to 0.095 (Table 2-3;

Figure 2-1).

Table 2-3. Genetic parameter estimates (standard error) for rooting of loblolly pine stem
cuttings across 5 trials.
Parameter Estimate
Narrow-sense heritability on observed binary scale (h, ) 0.051 (0.017)
Broad-sense heritability on observed binary scale (H0, ) 0.101 (0.008)
Narrow-sense heritability on the underlying normal scale
1(2) 0.08 (0.027)
(h )
Broad-sense heritability on the underlying normal scale
(/2) 0.16 (0.013)

Narrow-sense family mean heritability (!H ) 0.833 (0.24)
Broad-sense clonal mean heritability (HE) 0.815 (0.074)
Additive genetic variance (VA) 0.0117 (0.004)
Dominance genetic variance (VD) 0.0042 (0.002)
Epistatic genetic variance (V) 0.0074 (0.002)
Phenotypic variance (Vp ) 0.2297 (0.002)
Type B additive genetic variance correlation 0.68 (0.23)
Type B dominance genetic variance correlation 0.61 (0.27)
Type B total genetic variance correlation 0.53 (0.048)

Broad-sense heritability (H,12 ) ranged from approximately 0.15 to 0.22 (Figure 2-

1). Broad-sense heritability was 0.101 when all of the data from the five trials were

combined (Table 2-3; Figure 2-1). Foster (1990) observed very little nonadditive genetic









variance and reported H 2 of 0.13 for rooting percentage. In another study H' 2 was

reported as 0.23 for rooting percentage (Foster 1978). However, Anderson et al. (1999)

reported a much higher 2 (0.63) for rooting percentage of loblolly pine. When the data

in this study were analyzed based on rooting percentage, then H2 was 0.47.

Narrow- and broad-sense heritability estimates on the observed 0,1-scale were

transformed to an underlying normal scale assuming a threshold model (Equation 2-11).

Narrow-sense heritability based on the underlying normal scale (h2 ) ranged from 0.12 to

0.16 (Figure 2-2). Broad-sense heritability based on the underlying normal scale (H )

ranged from 0.24 to 0.36 (Figure 2-2). When all of the data from the five settings were

analyzed together, then h2 was 0.08 and H2 was 0.16 (Table 2-3; Figure 2-2). The

transformation of h0 and Hj, to the underlying normal scale allows direct comparisons

of heritability estimates among rooting studies when the rooting percentages are different.

Family mean heritability ranged from 0.84 to 0.9, while clonal mean heritability

ranged from 0.82 to 0.92 (Figure 2-3). When all of the data from the five trials were

combined then H2 was 0.83 and H2 was 0.82 (Table 2-3; Figure 2-3). Foster (1990)
FS CL

reported lower estimates of both family-mean heritability and clonal-mean heritability for

a rooting study consisting of 540 clones of loblolly pine from 54 full-sib families, 0.46

and 0.40, respectively. In another loblolly pine rooting study with 27 full-sib families

consisting of 10 clones each, family-mean heritability was reported as 0.31, while the

clonal-mean heritability was 0.87 (Anderson et al. 1999).

















42N
AT


0.4

0.35

0.3

o 0.25
0O
-o 0.2
E
o 0.15
Z
0.1

0.05

0


SpringOl SummerOl Winter02 Spring02 Summer02 All

Trial


Figure 2-2. Narrow-sense (1) and broad-sense (/f ) heritability estimates for rooting
of loblolly pine stem cuttings transformed to the underlying normal scale
using the threshold model of Equation 2-11. Standard error bars are included.



1.2


a,

E 0.6
o,
LJ 0.4

0.2

0


SpringOl SummerOl Winter02 Spring02 Summer02
Trial


FS

CL


Figure 2-3. Full-sib family mean ( ) and clonal mean (H ) heritability estimates for
rooting success of 2,200 clones from 70 full-sib families of loblolly pine.
Standard error bars are included.









Type B Genetic Correlations

The type B correlations for additive effects were moderately high and ranged from

0.53 to 0.91 (Table 2-4) indicating that parental rankings of the 32 parents were

moderately to strongly correlated among pairs of trials. When all of the data were

analyzed together from the five trials, the type B additive correlation was 0.68. The

highest type B correlation was observed between the two spring settings which were one

year apart, while the lowest correlations were observed between the winter setting

(Winter02) and other trials. Spring and summer cuttings were, in general, actively

growing, succulent material, while winter cuttings were generally smaller and more

lignified. Perhaps, some of the genes controlling rooting of dormant winter cuttings are

different than those controlling spring and summer cuttings.

Table 2-4. Type B additive and dominance variance correlations among pairs of rooting
trials for loblolly pine stem cuttings (above and below diagonal, respectively).
Standard errors are given in parentheses.

SpringO1 SummerO 1 Winter02 Spring02 Summer02

SpringO1 0.701 (0.11) 0.59(0.15) 0.911 (0.07) 0.851 (0.08)


SummerOl 0.823 (0.19) 0.634 (0.12) 0.673 (0.12) 0.634 (0.13)


Winter02 0.192 (0.30) 0.629 (0.34) 0.656 (0.12) 0.525 (0.16)


Spring02 0.493 (0.24) 0.533 (0.35) 0.724 (0.24) 0.907 (0.06)


Summer02 0.777 (0.22) 0.973 (0.37) 0 0.703 (0.24)



Estimated type B correlations for dominance effects measure the correspondence of

dominance across pairs of trials and were estimated to be moderate to high (Table 2-4)









with two exceptions. First, the type B dominance correlation between the SpringOl and

Winter02 trials was 0.192. Also, no dominance genetic variance was detected in the

pairwise analysis of theWinter02 and Summer02 trials. However, when all of the data

were analyzed together, the type B dominance correlation was 0.61.

Selection for Rooting

Before clones of loblolly pine can be deployed operationally, two things must

occur. First, the clones have to be field-tested and selected for desirable traits, e.g.,

growth, disease resistance, and wood properties. Second, the selected clones have to be

be propagated in large enough numbers for deployment. For a rooted cutting-based

program, this involves bulking up the number of hedges of a particular clone or group of

clones through serial propagation and then producing reforestation stock efficiently from

the bulked-up clones. Only those clones that can be propagated easily will be

economically feasible for deployment. Therefore, selection of clones for rootability as

well as field performance should be considered as part of a clonal forestry program based

on rooted cutting technology (Foster et al. 1985; Foster et al. 2000). In the current

study, selection of the top 10% of clones for rooting (-220 clones) would result in a gain

of about 24% in rooting, where

Gain =
I 22 2 2 ^2 ^2 ^2 .2
i 2 2 j 2 2 2"TESTxGCA 'TESTxSCA + TESTxCLONE 'REPxFAM F ERROR
iL ^GCA SCA +CLONE
L t t t tr trn

(Foster 1990).

Selecting the top 1% of clones for rooting would result in a gain of nearly 37% in

rooting of loblolly pine stem cuttings in the current generation. Alternatively, genetic










gain in rooting success in the next generation can be achieved through selection and

breeding.

Gain =
2 2^2 ^2 ^2 ^2 ^2



(j 2
^f 2 G +2 2 + -CLONE 2s-- --xGCA TESTxSCA TESTxCL+ONE -REPxFA+M + RROR



+ il 2 ^CA&_
i, 2p 2r mO
t /2 ^2 ^2
CL2 TESTxCLONE REPxFAM RROR
V/JCLONE + +
t tr trn


By selecting the best rooting clone from the top 25 out of 70 families, gain in

rooting success of 16.8% can be expected in the next generation by breeding these 25

selections.

Conclusion

Loblolly pine is the most important commercial tree species in the southeastern

United States. Several forest industry companies are developing rooted cutting programs

for loblolly pine in order to maximize genetic gains through deployment of tested clones.

With rooting data from 2,200 clones from 70 full-sib families, the current study gives

better estimates of genetic components of variance for rooting than several previous

studies. These results show a great deal of genetic variation for rooting among families

and clones of loblolly pine. Only those clones that can be propagated easily will be

economically feasible for deployment. Therefore, selection of clones for rooting ability,

as well as field performance should be considered as part of a clonal forestry program

based on rooted cutting technology. Combined with moderate to high estimates of

family- and clonal-mean heritabilities and type B correlations, these results indicate the






29


potential for increasing rooting efficiency by selecting good rooting families and clones

or culling poor rooters.














CHAPTER 3
GENETIC ANALYSIS OF EARLY FIELD GROWTH OF LOBLOLLY PINE CLONES
AND SEEDLINGS FROM THE SAME FULL-SB FAMILIES

Introduction

Loblolly pine (Pinus taeda L.) is the most important commercial tree species in the

United States with over one billion seedlings planted annually (McKeand et al. 2003).

Most commercially important tree species remain relatively undomesticated, and loblolly

pine tree improvement programs are only now beginning their 3rd generation of breeding

and testing (McKeand and Bridgewater 1998). Tree improvement programs for loblolly

pine have relied on recurrent selection for general combining ability for improvement of

a few key traits. These programs have historically utilized seedling progeny trials in

order to predict breeding values for these traits. Traditional tree improvement programs

using open-pollinated seed orchard seedlings for deployment only capture additive

genetic variation. However, nonadditive genetic variation may be an important

component of variation for some traits, and additive and nonadditive genetic variation

can be captured by deploying full-sib families or clones.

More efficient field-testing has been implemented in order to gain information on

full-sib families. For example, the Cooperative Forest Genetics Research Program at the

University of Florida is using a partial diallel mating design to cross slash pine selections

to generate full-sib seedlings for progeny trials (Gezan et al. 2004). Tests established

with full-sib seedlings allow the genetic variance to be partitioned into additive and

dominance components (Falconer and Mackay 1996). These trials not only provide ranks









of parents or individuals for selection, but also of full-sib families in order to provide

information for making deployment decisions.

For a number of reasons, clonally replicated progeny trials have been suggested as

part of a tree improvement strategy for radiata pine (Jayawickrama and Carson 2000) and

for loblolly pine (Foster and Shaw 1987; Isik et al. 2004; Byram et al. 2004). First, field

trials established with clonally replicated progeny allow for further partitioning of the

genetic variation into the additive, dominance, and epistatic genetic components (Foster

and Shaw 1988). Second, clonally propagated seedlings can provide genetic information

more efficiently and with greater precision than zygotic seedling progeny (Burdon and

Shelbourne 1974; Isik et al. 2004). Finally, clonal testing and selection strategies can

provide greater gain for operational deployment than seedling options (Shaw and Hood

1985; Mullin and Park 1994; Isik et al. 2004).

Based on current technologies, several forest industries in the southeastern United

States are pursuing clonal forestry programs with loblolly pine (Weber and Stelzer 2002).

In the initial stages of these clonal forestry programs, forest managers needed assurance

that the clonal propagules' growth corresponded to that of seedlings. Therefore, earlier

studies were designed to test whether cuttings grew similarly to seedlings. Based on

those results, it is generally accepted that cuttings rooted from juvenile stock plants grow

and perform comparably to seedlings. For example, Foster et al. (1987) reported that

loblolly pine rooted cuttings should perform comparably to seedlings when the cuttings

come from vigorous juvenile stock plants. In addition, McRae et al. (1993) concluded

that for loblolly pine there were no significant differences between seedlings and rooted

cutting propagules from common checklots through five years of growth. Similar results









were obtained by Frampton et al. (2000) where they reported no significant differences

between the means of rooted cuttings and seedlings for height, diameter at breast height,

and volume through six years in the field.

Trials established with clones and seedlings from the same families provide an

opportunity for comparing both half-sib and full-sib family performances across

propagule types. Genetic correlations between propagule types can provide further

assurance that selections made through traditional tree improvement activities for

recurrent selection for general combining ability in seedling tests can also be used

successfully in breeding families to test in a clonal forestry program. Although a number

of studies have compared rooted cutting and seedlings, very few have been designed to

estimate the genetic correlation between propagule types for a trait.

While clonal tests derived from full-sib families provide an opportunity to estimate

additive and nonadditive components of variance, tests should be designed with sufficient

genetic structure in order to precisely quantify the genetic variation. Several clonal

studies have reported deficiencies in mating designs, number of parents and families

(Frampton and Huber 1995; Paul et al. 1997; Isik et al. 2003). For example, Frampton

and Huber (1995) reported that they had low power in partitioning the genetic variation

because of the lack of a mating design among the parents of the full-sib crosses in a

loblolly pine clonal study. In addition, Frampton and Foster (1993) warned that

interpretation of the results may be difficult for studies that only include seedlings and

cuttings from a common checklot to be compared to the clonal propagules from select

parents and families. In this case, any differences in the field performance because of









propagule type may be confounded with the differences in genetic improvement

(Frampton and Foster 1993).

The current study employs a complex genetic structure to increase the power in

quantifying the genetic variation associated with several growth traits in loblolly pine.

More than 1,200 clones together with zygotic seedlings from the same 61 full-sib families

were tested on multiple sites across the southeastern United States. Because of the test

and mating designs, genetic correlations can be directly calculated between propagule

types and within propagule types across sites. The specific objectives of this study are to

1) determine heritability estimates for various growth traits for loblolly pine clones and

seedlings, 2) compare the performance between parents and full-sib families when grown

as rooted cuttings and seedlings, and 3) determine the extent of genotype x environment

interaction by looking at the genetic correlations for parents, families, and clones across

multiple sites.

Materials and Methods

Population

The parental population consisted of twenty first-generation and ten second-

generation selections, subset from the Loblolly Pine Lower Gulf Elite Population. Two

additional first-generation, slow-growing parents were included. The parental selections

represent the Atlantic Coastal Plain, Florida, and Lower Gulf provenances of loblolly

pine (see FBRC 2000 for details). Briefly, these thirty-two loblolly pine parents were

mated in a partial diallel design and created 70 full-sib families from which more than

2,000 seedling hedges were generated and given unique clonal identifications (Appendix

A). On average, each parent was involved in approximately four crosses.









Propagation

The propagation of the rooted cuttings for the field trials has been previously

described (Chapter 2; Baltunis et al. 2005). But briefly, the seedling hedges were

repeatedly sheared to slow down the effects of maturation and increase the number of

shoots available for collection. Cuttings were collected from seedling hedges from 61

full-sib families, placed randomly in clonal-row plots, and replicated six times in a

greenhouse in January 2002, April 2002, and June 2002. At the time of collection, the

hedges were 22, 25, and 27 months old from seed, respectively. Cuttings were assessed

for rooting at 11-weeks after setting (Baltunis et al. 2005). Rooted cuttings were then

transplanted into Ray Leach Supercells (Steuwe and Sons, Corvalis, Oregon) and

randomized into their designated field planting order and grown to size. The clonal

propagules for a field trial came from a single rooting trial (Table 3-1).

Table 3-1. Location of six field trials, establishment date and total number of test trees
for each test.

Test Sticking Location Latitude Date Total No.
Test Location
Date Longitude Planted Test Trees

A January Worth County, 31'44'20"N October 9,216
2002 Georgia 83'55'50"W 2002

B April Morgan County, 3324'55"N November 8,960
2002 Georgia 83'29'45"W 2002

April Putnam County, 2938'N November
2002 Florida 8146'W 2002

D April Nassau County, 30'45'23"N February 8,960
2002 Florida 81'54'27"W 2003

E April Randolph County, 3148'03"N December 4,400
2002 Georgia 84'41'30"W 2002

F June Santa Rosa County, 30'50'05"N April
2002 Florida 87'11'57"W 2003









A single crop of seedlings, on the other hand, was used to produce all the seedlings

for the field trials. Loblolly pine seed from the same full-sib families that the clones were

derived from were stratified for about 30 days and then sown in May 2002. The

seedlings were grown in family blocks in a different greenhouse than the cuttings. The

seedlings were then moved outdoors under shade cloth and kept in their family blocks in

a separate area from where the rooted cuttings were growing. The seedlings were not

randomized into their designated field order until just prior to planting.

Field Design

In total 47,408 measurement trees were established at six field sites across the

southeastern United States (Table 3-1). Three trials each were established in Florida and

Georgia (Table 3-1). An additional field trial was established in Virginia with a subset of

the clones but was not included in any of these analyses. There were four replications in

each of two cultural treatments (high and low intensity) in each test, except for Test E

where there was only one cultural treatment and four replications. The goal for the high

intensity treatment was to push the trees to their utmost potential by reducing competition

and providing a non-limiting supply of nutrients, while the low intensive culture provides

insights into family and clonal performance under a less optimal cultural regime (FBRC

2000). Both cultural intensities were treated similarly during the first year with cultural

differences implemented at the beginning of the second growing season.

Each trial contained 756-974 clones with approximately 15 clones from each of 61

full-sib families (Table 3-2). In total, more than 1,200 clones were planted in field trials

(Table 3-2). The trials were designed to have four zygotic seedlings from each of the

same 61 full-sib families within each replication. However, because of poor germination

for some of the families or mortality in the nursery, each full-sib family is represented, on









average, by approximately 27 zygotic seedlings per test (Table 3-2). Both the rooted

cuttings and seedlings were planted in single-tree plots utilizing a resolvable alpha

incomplete block design (Williams et al. 2002) in which incomplete block size ranged

from 10-14 trees. The variables measured were 1st year height, 2nd year height, height

increment, and crown width.

Table 3-2. Total number of clones, full-sib families, half-sib families, average number of
clones per full-sib and half-sib family, and average number of seedlings per
full-sib and half-sib family established at the six field trials.

Test A Test B Test C Test D Test E Test F Total


Total # clones 974 941 942 956 868 756 1,212


Total # FS families 61 61 61 61 61 61a 61


Total # HS families 32 32 32 32 32 32b 32

Ave. # clones/FS
.# cl 16 15.4 15.4 15.7 14.2 12.4 19.9
family

Ave. # clones/HS
Ave. # conesHS 60.9 58.8 58.9 59.5 54.2 47.2 75.7
family

Ave. # seedlings/
Ave. # seedlings/ 27.9 27 27 27 15.2 35.1 151.2
FS family

Ave. # seedlings/
Ave. # seedlings/ 106.5 103 103 103 58 106.3 576.5
HS family
a Only 47 full-sib families and b 31 half-sib families are represented in the seedling
population of Test F.


Statistical Analyses

All growth variables, 1st year height, 2nd year height, height increment, and crown

width, were analyzed in ASREML (Gilmour et al. 2002) using a parental model in order

to estimate the genetic variance components associated with those traits. Analyses were

conducted for each trial separately and across all six trials for each propagule type. The









across-trial analyses assumed a different error variance for each trial. For the clonal

population, the following across-trial model was used.

[3-1]

Yjklmnop = + + T C + Rk() + incbkl(Qk) + gcam + gcan + scamn + cloneo(mn)
+ t gcat + t gca,, + t fammn + t* cloneo(mn) + t c gc + t c gca ,, + t c fammn
+t* c coIe + r g / )m +r* g )+ r famk()mn + ekmnop



where,

yjkmnop is the measured growth trait of the pth ramet of the oth clone within the mnth

full-sib family in the lth incomplete block within the kth replication of thejth cultural

treatment in the ith test

/u is the clonal population mean

T, is the fixed effect of trial, i = 1,...,6

T C, is the fixed effect of the interaction between trial and culture, j = 1,2


Rk(U) is the fixed effect of replication, k = 1,2,3,4

incbkl(,k) is the random variable incomplete block associated with each test NIID(0,

2,
INC,

gcam and gcaa are the random variables female (m) and male (n) general combining

ability, respectively NIID(0, G^CA


scamn is the random variable specific combining ability ~ NIID(0, !CA)

cloneo(mn) is the random variable clone within full-sib family ~ NIID(0, ^LONE









t ge t,, and t gca ., are the random variables test by female general combining ability

and test by male general combining ability interactions, respectively NIID(0, 2ESTxG )

t famm,. is the random variable test by full-sib family interaction NIID(0, &~ESTxFM

t chie ,, is the random variable test by clone within full-sib family interaction ~

NIID(0, cr 2)
NIID(O, TESTxCLONE

t c gca ,, and t c gca ., are the random variables test by culture by female general

combining ability and test by culture by male general combining ability, respectively ~

NIID(0, TXGC4)
NtC ,TxCxGCA)

t c famj, is the random variable test by culture by full-sib family ~ NIID(0, & CF

t c clone is the random variable test by culture by clone within full-sib family

interaction ~ NIID(0, &^ TCLNE

r gII .)m and r gII ), are the random variables replication by female general

combining ability and replication by male general combining ability, respectively ~

NIID(0, REPxGCA)

r fak(,,,,, is the random variable replication by full-sib family ~ NIID(0, 62EPxiMI)

ekm,,op is the random error associated with each test ~ NIID(0, 'ER )OR.

The single-site model for the clonal population is identical, except that all model

factors with subscript i are removed (sources involving test).

All traits were also analyzed for the seedling population assuming a randomized

complete block design (incomplete block dropped from model). Both single-trial









analyses and an across-trial analysis were performed again assuming heterogeneous

errors across sites.

[3-2]

zkimn = P + Rk() +gc +gc + T sea,, + ga +t gca c,, + t* faml,
+t*c*g I +t*c* gca,, +t*c* fam;i+m +r*gL )L +r* g ./ ) +r* famk(U)lm
+ eijklmn

where

ZUklmn is the measured growth trait of the nth seedling from the Imth full-sib family in the

kth replication of thejth cultural treatment in the ith test

/j is the seedling population mean and the other variables are defined as above (just

adjusting the appropriate subscripts).

Genetic parameters were estimated and standard errors were calculated according

to Foster and Shaw (1988) using the variance components from the appropriate model.

2= 1 1
4 16

is the estimate of additive genetic variance.
2[3-4] 4 1 1 1
[3-4] D = 4 = + AA + V-A + D-VDD
2 2 4

is the estimate of dominance genetic variance.

S= (2 (2 2 1 1 3
[3-5] V CLONE -2, GCA 0SCA = VAA + AD DD ..


is the estimate of epistatic genetic variance for the clonal population.
[3-6] = 2&c2 +s2 + o
[3-6] VG = GCA +JCSCA CLONEE

is the estimate of total genetic variance for the clonal population.










[3-7] = 4'!2 + 4A 2
GCA SCA

is the estimate of total genetic variance for the seedling population (assuming no

epistasis).

P 282 2 2 2 2 2 2
S = 2c GCA SCA + CLONE + 2 TESTxGCA TESTxFAM + TESTxCLONE + 2TxCxGCA
6
[3-8] y'-2
[3-8] ERROR,
^2 + ^2 + ^2 ^2 1=1
STxCxFAM 'TxCxCLONE + 2REPxGCA 'REPxFAM +
6

is the estimated phenotypic variance for the across-trial model for the clonal population.

P j ^2 ^ 2 ^ 2 12 ^ 2 12 ^ 2
S= 2 GC S +SCA 2 TESTxGCA + 7TESTxFAM + 2- TxCxGCA + -TxCxFAMI + 2-REPxGCA
6
[3-9] -- -2
[39] ERROR,
"2 1-i
+ -'REPxFAM + 6
6

is the estimated phenotypic variance for the across-trial model for the seedling

population.


Individual tree narrow-sense heritability (h2 ) and broad-sense heritability (H )

were derived using the estimated variance components for all the growth traits and each

propagule type at each site and across sites. Standard errors were calculated using Taylor

series expansion (Kendall and Stuart 1963; Namkoong 1979; Huber et al. 1992; Dieters

1994). The following heritability formulae were used for the clonal data.


[3-10] h2 = GCA
V V
P PC

is the across-trial estimate of individual tree narrow-sense heritability for the clonal


population, where Vp is from [3-8].


V (22 +'2 + '2
[3-111 H =V2 G GCA + CSCA + CCLONE
3- 77-









is the across-trial estimate of individual tree broad-sense heritability for the clonal

population, where V is from [3-8].

Heritability estimates were also obtained for the seedling data for each trial and

across-trials.


[3-12] h2 =A ~~C
PS PS

is the across-trial estimate of individual tree narrow-sense heritability for the seedling

population, where V is from [3-9].


[3-13] H2_ GCA + 2CA
VPS VPS

is the across-trial estimate of individual tree broad-sense heritability for the seedling

population, where Vp is from [3-9].

The various growth variables at each site were also analyzed with a bivariate mixed

model with the growth of the clones and seedlings as two dependent variables. Type B

genetic correlations for general combining ability (Br ) and full-sib family value


(BpropF ) between cuttings and seedlings were estimated in order to compare parental and

full-sib family performance between propagule types. The genetic correlation between

propagule types for additive effects, for example, gives us an indication of whether

parental ranks are dependent upon whether their progeny are grown as cuttings or

seedlings, while a type B genetic correlation at the full-sib family level measures the

performance of full-sib families across propagule types.










[3-14] Cov(GCAcrGCASED )
r-" )Var(GCAcu ) x Var (GCASEED)


is the type B genetic correlation of additive effects between propagule types using genetic

variance component estimates from the bivariate analysis.

2Cov(GCAcur GCAED) + Cov(SCAc SCARED)
[3-15] r
)(2Var(GCAcu ) + Var(SCAcu )) x (2Var(GCAED ) + Var(SCAEED))


is the type B genetic correlation of full-sib family values between propagule types using

genetic variance component estimates from the bivariate analysis.

The extent of genotype x environment interaction was investigated by analyzing

data across trials for each propagule type using the variance components from the

appropriate model. Type B genetic correlations were calculated for additive effects, full-

sib family, and the total genetic or clonal value across the trials (Yamada 1962; Burdon

1977). Standard errors of type B correlations were calculated using the Taylor series

expansion method.

^2
O~-zc
[3-16] B = 2 GCA
GCA "2 +"2
S GCA TESTxGCA

is the type B genetic correlation for additive effects across trials.

^2 ^ 2
[3-17] 'rB =~2 2 2 2
[3-17] ^ 2GCA SCA
GCA TESTxGCA SCA TESTxFAM

is the type B genetic correlation for full-sib families across trials.

j 2 '^2 ^2
[3-18] 2 GCA + SCA + CLONE
S^2 ^2 2 ^2 ^2 ^2
2-GCA + 2-TESTxGCA + -SCA + TESTxFArM + 'CLONE + UTESTxCLONE

is the type B genetic correlation for total genetic or clonal value across trials for the

clonal population.









Ranging from 0 to 1, a value of i- near one indicates little genotype x

environment interaction and that the parents ranked the same across the trials, while a

low r (near zero) indicates that parental ranks were not stable across the sites and

hence, genotype x environment interaction exists. A high B, indicates that full-sib

families performed similarly across the sites, while a high? B indicates that the total

genetic values of the clones were stable across the trials.

Results and Discussion

Overall Growth of Clones and Seedlings

Survival of both clonal and zygotic propagules was high across all of the field

trials. Survival of the rooted cuttings ranged from 91.7-98.3% at the six field trials, while

seedling survival ranged from 86.5-97.4%. At the time of planting, seedlings were

generally taller than rooted cuttings, and this trend has continued through year two (Table

3-3). For example, after the first and second growing seasons seedlings were on average

11 cm and 10 cm taller than rooted cuttings, respectively. It has been suggested that

propagule size differences at the time of planting may create difficulties in the analysis

and interpretation of subsequent growth (Frampton and Foster 1993). However, these

differences in initial propagule size may not be a problem in the current study because

height increments were very similar between both propagule types, with site means

ranging between 1.0-2.0 m and 1.0-1.9 m for rooted cuttings and seedlings, respectively

(Table 3-3).

The initial and first year treatments were the same within a site during test

establishment, and there were no cultural differences between propagule types for 1st year

height. There were some differences in mean height due to the effects of cultural









intensity during the second growing season at some of the sites. However, these effects

were more a function of scale. Although the overall means of the growth variables

differed by cultural treatment, the ranks of parents, families, or clones were not affected.

Type B genetic correlations exceeded 0.85 indicating little cultural treatment x genetic

effect interaction, and therefore, cultural effects are ignored for the purposes of this study.

This implies that the rankings of parents, families, and clones were robust across multiple

management regimes through age two.

Table 3-3. Mean 1st year height, 2nd year height, height increment, and crown width by
propagule type for each of the six field trials. Although means are expressed
in meters, analyses were conducted using measured traits in centimeters.
1st Year 2nd Year Height Crown Width
Height (m) Height (m) Increment (m) (m)
Clones 1.0 2.1 1.2 0.9
Test A
Seedlings 1.1 2.3 1.2 1.0
Clones 0.8 1.8 1.0 0.9
Test B
Seedlings 0.9 1.8 1.0 0.9
Clones 1.0 2.1 1.1 1.2
Test C
Seedlings 1.1 2.3 1.2 1.3
Clones 0.7 2.0 1.3 1.0
Test D
Seedlings 0.8 2.1 1.3 1.1
Clones 1.2 3.3 2.0 2.0
Test E
Seedlings 1.3 3.2 1.9 2.0
Clones 0.5 1.7 1.3 1.1
Test F
Seedlings 0.6 1.8 1.3 1.1


Genetic Components of Variance

All of the early growth traits demonstrated genetic variation (Figure 3-1; Table 3-4;

Appendix C). However, the genetic variation partitioned differently for the two

propagule types. In all cases, the estimate of additive genetic variation was greater for









the clones than the seedlings. Within a single trial, for instance, the estimate of the

additive genetic variation for 2nd year height based on seedlings was 0 to 0.58 that of

additive genetic variation based on clonal data. Similar trends were observed from the

across-trial analyses. The estimate of additive genetic variation for the clonal material

was 4.7, 3.3, 2.9, and 2 times greater than the additive genetic variation for seedlings for

1st year height, 2nd year height, height increment, and crown width, respectively (Table 3-

4).

Based on single-trial analyses, the majority of the genetic variation associated with

all of the growth variables in the clonal population was additive, while in the seedling

population the trend was towards nonadditive genetic variation (Figure 3-1). For

example, at Test D all of the genetic variation associated with 2nd year height of clones

was additive, while for the seedlings it was dominance genetic variation (Figure 3-1).

When all of the data were analyzed together, then the estimates of dominance genetic

variation were approximately equivalent for both propagules (Table 3-4) suggesting a

large test by dominance interaction for the seedlings. Epistasis was negative for all

growth variables. As a result, estimates of additive and dominance genetic variance for

these traits might not be upwardly biased as indicated by the expected portions of

epistatic interactions defined in Equations 3-3 and 3-4 (Foster and Shaw 1988). Isik et al.

(2003) reported similar trends in the partitioning of the genetic variation for clones and

seedlings including negative epistasis estimates for height, diameter, and volume through

age six. They also reported that the additive genetic variation for growth traits was

always greater for clones than the estimate for seedlings, while dominance genetic

variation was greater in the seedling population (Isik et al. 2003).









Table 3-4. Genetic parameter estimates for 1st year height, 2nd year height, height
increment, and crown width by propagule type across all six trials. Standard
errors are given in parentheses.
1st Year Height 2nd Year Height Height Increment Crown Width

Clone Seed Clone Seed Clone Seed Clone Seed
69.8 14.7 401.5 119.9 152.6 51.8 110.3 55
VA (21.3) (7.8) (122) (50.8) (47) (24.2) (34) (20)

12 17.5 86.4 83.3 30.3 41.9 19.7 20
D (6.2) (9.3) (36.5) (41.6) (13.6) (21.5) (10.1) (11.7)

-11.5 -107.6 -39.7 -13.5
(11.4) (64.3) (24.8) (18.1)
70.3 32.1 380.3 203.1 143.2 93.65 116.5 75
G (10.8) (9.5) (61.1) (53.8) (23.6) (26.1) (17.2) (20.7)

442.8 405.2 1806 1703 902.6 972 640.9 649.5
p (11.3) (7.6) (62.3) (36.2) (24.5) (19.4) (17.8) (14.3)

2 0.16 0.04 0.22 0.07 0.17 0.05 0.17 0.08
(0.04) (0.02) (0.06) (0.03) (0.05) (0.02) (0.05) (0.03)
S 0.16 0.08 0.21 0.12 0.16 0.1 0.18 0.11
(0.02) (0.02) (0.03) (0.03) (0.02) (0.02) (0.02) (0.03)

S0.81 0.71 0.88 0.78 0.83 0.6 0.82 0.78
erBC (0.06) (0.2) (0.04) (0.12) (0.05) (0.15) (0.06) (0.12)

B 0.8 0.5 0.88 0.66 0.83 0.58 0.82 0.67
rBsF (0.05) (0.1) (0.04) (0.09) (0.05) (0.1) (0.05) (0.09)
rB 0.69 0.77 0.76 0.76
(0.04) (0.03) (0.04) (0.03)


Genetic causes are not the only source for similarity among relatives. Non-genetic

factors, such as C effects, can lead to upwardly biased estimates of total genetic and

nonadditive genetic components of variance when analyzing clonal data (Libby and Jund

1962). C effects are often assumed negligible when estimating epistatic variance (Foster

and Shaw 1988). If C effects are present then the total genetic variation associated with









clones will be overestimated (Libby and Jund 1962). Significant C effects are likely to

occur in traits that are measured soon after propagation, but apparently lessen for traits

measured at later times (Libby and Jund 1962). In the current experiment, if C effects

exist, then estimates of epistatic genetic variation will be confounded with C effects

because ramets of a clone came from a single, non-replicated hedge. However, the

estimates for epistasis were negative for all of the growth traits (Table 3-4), which

suggest that interloci interactions are not an important source of genetic variation for

early growth traits, and that C effects may not be a major contributing factor. Further,

estimates of additive and dominance should be free of confounding C effects in the

current experiment since randomization of clones occurred at all stages including hedge

establishment, propagation, and growth prior to and after test establishment.

As Falconer and Mackay (1996) point out, relatives of all sorts may resemble one

another because of sharing a common environment. The variance attributed to a common

environment occurs more frequently and contributes greater to the covariance of full-sibs

than to the covariance of any other sort of relatives (Falconer and Mackay 1996). When

a common environment effect exists, then the differences between means of families

become greater than when these non-genetic factors are not present. In the case of full-

sib families, then this will lead to biased or inflated estimates of dominance genetic

variation.

The results seen in the current experiment for dominance genetic variance estimates

in the seedling population appear to be a result of a partitioning problem in that estimates

of the variance due to dominance are inflated at the expense of additive effects (Figure 3-

1). Seeds are often sown and seedlings grown in full-sib family blocks in the greenhouse
















a. 1styear height
0.7
0.6
0.5
- 0.4 I na


0.2
0.1








0.45
0
o (1 0 (D 0 (D 0 (D 0 (D 0 w
-5 U) O U) 0 U) 5 U) 5 U) O U)
TestA TestB TestC TestD TestE TestF


c. Height increment
0.45 -
f nlA


b. 2nd year height
0.6

S0.5

0.4
IU na
0.3
[ add
q .2

0.1



0 () 0 () 0 (1) 0 (1) 0 (1) 0 (

TestA Test B Test C Test D TestE Test F


d. Crown width
0.6
0.3 1~05-


0.3 0.4
0.25 na d ai
0 0.3
^ 0.2 Iadd ^ T add
0.15 0.2 I I
0.1
~ os 8o0.1
0 0

W ( 0 0 (W 0 (0 0 (D 0 (0 0 (W 0 (W 0 (1 0 (1 0 (1 0 (1 0 (1
-O U) -0 U) -O U) -O U) > U) > U) U) -0 U) -O U) -O U) -O U) (n U)
Test A TestB Test C Test D Test E TestF TestA TestB Test C Test D TestE TestF


Figure 3-1. The proportion of additive (add) and nonadditive (na) genetic variance components for clones and seedlings across the six
field trials, where h2 = add and H2 = add + na: a. 1st year height, b. 2nd year height, c. Height increment, and d. Crown
width. Standard error bars for broad-sense heritability estimates are included.









or nursery for progeny testing or other field testing. In order to eliminate the common

environment effect at the family level, then seedlings within a full-sib family should be

randomized over the environment in which they are being grown and tested. Although

rooted cuttings were randomized at all stages of propagation, the seedlings were in fact

grown in full-sib blocks, and apparently a common environment effect has carried over to

the field through age two measurements.

Heritability Estimates

First year height, 2nd year height, height increment, and crown width were all

influenced by additive genetic variation. Individual tree narrow-sense heritability was

always greater for the clones than for seedlings for all of the traits (Figure 3-1, Table 3-

4). Generally, h2 estimates of total height increased from age one to age two for both

clones and seedlings (Figure 3-1, Table 3-4). Narrow-sense heritability for total height

increased from 0.16 at age one to 0.22 at age two for the clones, while in the seedling

population, h increased from 0.04 to 0.07 (Table 3-4). In addition, h estimates for all

of the growth traits from the seedling data had larger standard errors associated with them

than the estimates from clones (Table 3-4). Yet, the differences in heritability estimates

between propagule types may be related to C effects. If C effects are present, then H2

estimates from the clonal population may be inflated. However, estimates of h2 from the

clonal population may not be upwardly biased since clones within full-sib families were

randomized in the hedge orchard and throughout the study, thus reducing spurious C

effects at the parental and full-sib family levels.

Heritability estimates have been reported for several loblolly pine seedling and

clonal populations. In a study reported by Isik et al. (2003), individual tree narrow-sense









heritability for volume at age six was 0.3 for clones, while for seedlings h2 was 0.06. In

another loblolly pine clonal test, heritability estimates for total height growth through age

five for two factorials were similar to the values reported here (Paul et al. 1997). They

reported h2 for 1st and 2nd year height as 0.08 and 0.17, respectively (mean of two

factorials). Unfortunately, they did not include seedlings in their study. Analogous

results have been reported in Eucalyptus globulus Labill. for heritability estimates for

diameter for both clones and seedlings in that h2 was always greater for the clonal

population than the seedling population (Costa e Silva et al. 2004).

Broad-sense heritability estimates for the various growth traits were always larger

for the clonal material than the seedlings except for the estimates of H2 for Ist year

height at Test E and for all four variables at Test D (Figure 3-1). However, H2 estimates

from the across-trial models were always greater for the clones than seedlings (Table 3-

4). Because of the negative estimates of epistasis in the clonal population, 2 estimates

were equivalent to the estimates for individual tree narrow-sense heritability for all traits.

As was the case with 2, seedling estimates of H2 had higher standard errors associated

with them (Figure 3-1).

Type B Genetic Correlations Between Propagule Types

In order to further compare the clonal rooted cuttings and zygotic seedlings, type B

genetic correlations for general combining ability and full-sib family value between

propagule types were estimated. For all growth traits measured, the type B genetic

correlation between propagule types for additive effects was high with values exceeding

0.72 (Table 3-5). For example, ro for Ist year height ranged from 0.93-0.99, while









for height increment Bprop ranged from 0.72-0.99 (Table 3-5). These high genetic

correlations imply that parental rankings for early growth traits are stable regardless of

whether their progeny are being tested as zygotic seedlings or rooted cuttings (Figure 3-

2a).

Table 3-5. Genetic correlations between propagule types for 1st year height, 2nd year
height, height increment, and crown width at the parental (rop ) and full-sib
family (ro F) levels.

1st Year Height 2nd Year Height Height Increment Crown Width

Test rBpropGCA BpropFS rBpropGCA rBpropFS rBpropGCA rBpropFS propGCA propFS

A 0.99 0.62 0.99 0.65 0.99 0.38 0.86 0.86
B 0.99 0.74 0.99 0.86 0.99 0.99 0.99 0.81

C 0.93 0.74 0.99 0.75 0.99 0.99 0.99 0.97
D 0.93 0.44 0.99 0.64 0.99 0.92 0.93 0.92

E 0.99 0.55 0.89 0.71 0.72 0.77 0.99 0.69

F 0.95 0.49 0.96 0.89 0.95 0.83 0.99 0.83



The type B genetic correlations for additive effects between propagule types

observed in the current study were consistent with the expectations reported by Borralho

and Kanowski (1995). In a simulation study comparing the performance of clones and

seedlings from the same half-sib family, Borralho and Kanowski (1995) reported that the

expected correlations between propagule types exceeded 0.8 when greater than 100

seedlings or propagules were tested. Additionally, in a field study comparing rooted

cuttings and seedlings from four half-sib families, Foster et al. (1987) reported that

family rank correlations between propagule types were positive and significant for 1st

year height (0.52), 3rd year height (0.66) and 6th year height (0.70).









Full-sib families also ranked relatively similar regardless of propagule type at most

of the sites (Table 3-5; Figure 3-2b). For example, type B genetic correlations at the full-

sib family level ranged from 0.64-0.89 for 2nd year height. However, B, was more

variable from site to site for 1st year height and height increment. The results observed

here are in accordance with those reported by Frampton et al. (2000) in that full-sib

families or parental trees selected based on seedling genetic trials should also perform

well as rooted cuttings.

Genotype x Environment Interaction

The stability of parents and full-sib families across sites was compared for the

clonal and seedling populations (Table 3-4). The type B genetic correlation for additive

effects across sites was always greater for the clones, although these estimates were

moderately high for both populations. For example, ?B, for 2nd year height was 0.88

and 0.78 for the clonal and seedling populations, respectively (Table 3-4). Full-sib

families also ranked comparably across sites for the clonal population with i- values

exceeding 0.8 for all of the early growth traits (Table 3-4). These across site genetic

correlations were estimated with more precision using clonal replicates as evidenced by

lower standard errors for the estimates from the clonal population.

An additional genotype x environment interaction between test and total genetic

value can be estimated using clonal tests. The total genetic values of the clones were

fairly stable when all of the data was considered from all sites, indicating that a good

clone at one site is good at all the sites (Table 3-4). However, based on analyses from

pairs of trials (data not shown), there appears to be a propagation effect relating to the

season the cuttings were rooted. For example, the worst genetic correlations were







53


observed between the trial established with rooted cuttings from the winter setting and

any of the other trials, while the best genetic correlations were obtained from the field

trials that contained rooted cuttings originating from the spring setting. Similar results


35
30
25
r 20
1 15
C,
10
5
0





70
60
50


a,
- 40
30
CO


a. Parental ranks for 2nd year height

*
rB =0.99










0 5 10 15 20 25 30 35
Test B Parental Ranks Clones


b. Full-sib family ranks for 2nd year height


r =0.86
rBpropFS *

0 10* *3


-*I
0 10 20 30 40 50 60 70





0 10 20 30 40 50 60 70


Test B Full-Sib Family Ranks Clones


Figure 3-2. Rank-rank plots showing type B genetic correlations between clones and
seedlings from Test B based on: a. Parental BLUP values, b. Full-sib family
BLUP values, where full-sib BLUP values are equal to the sum of the
predicted general combining ability for each of the two parents plus the
predicted specific combining ability of the cross.









were also observed with rooting with this same population in that poor genetic

correlations were observed between rooting ability in winter and spring (Chapter 2;

Baltunis et al. 2005).

Conclusion

Several forest industries in the southeastern United States are deploying full-sib

families of loblolly pine operationally. In addition, many of these companies are

pursuing clonal forestry programs with loblolly pine. Genetic field trials established with

clones and seedlings from the same full-sib families provide an opportunity for

comparing both half-sib and full-sib family performances for both propagules. Based on

the current study, several conclusions can be drawn. First, clonally replicated seedling

trials of loblolly pine provide genetic information with greater precision than zygotic

seedlings. Second, genetic correlations between estimates of the genetic effects

associated with these growth traits between these propagule types were highly favorable.

These high genetic correlations between propagule types reassure that parental and full-

sib family rankings are stable regardless of propagule type. This implies that parental

and full-sib family rankings based on existing seedling progeny trials could be used to

select parents and families that perform well when they are deployed as rooted cuttings.

Third, little genotype x environment interaction was observed across sites at the parental,

family, and clonal level for all traits. However, there appears to be a carry-over effect

relating to the season in which the cuttings were rooted for the clonal material. Finally,

randomization is essential at all stages in testing when estimating genetic parameters.

The lack of randomization for the seedling population apparently resulted in a problem

with partitioning of the genetic variance, causing full-sib families to appear more

different and inflating estimates of dominance genetic variation.
















CHAPTER 4
GENETIC GAIN FROM SELECTION FOR ROOTING ABILITY AND EARLY
GROWTH IN VEGETATIVELY PROPAGATED CLONES OF LOBLOLLY PINE

Introduction

Loblolly pine (Pinus taeda L.) is the most important commercial tree species in the

southeastern United States (McKeand et al. 2003). Genetic improvement of loblolly pine

has been occurring since the 1950's in several tree improvement programs. These

programs have aimed to increase the population mean breeding value of a few key traits,

such as stem volume, disease resistance, and wood properties, through breeding and

selection of superior genotypes. Tree improvement programs are based on recurrent

selection for general combining ability and capture only a portion of the additive genetic

variation with open-pollinated seedlings for deployment. However, the nonadditive

portion of genetic variation, dominance and epistasis, may be important components of

variation for traits. For example, Stonecypher and McCullough (1986) reported that the

estimates of nonadditive variance were approximately equal to those of additive variance

for growth traits in Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) through age six.

Additionally, Paul et al. (1997) reported that both additive and dominance genetic

variance increased from age one to five for height growth for loblolly pine clones. The

only manner in which to capture the total genetic variation is through operational

deployment of clonal propagules.









Based on current technologies, several forest industries in the southeastern United

States are pursuing clonal forestry programs for loblolly pine using either somatic

embryogenesis or rooted cuttings (Weber and Stelzer 2002). Two main criteria need to

be met prior to operational deployment of loblolly pine clones. First, loblolly pine clones

must perform well, e.g., meet the selection criteria for the desired traits. This involves

the accumulation of reliable data for the clones from greenhouse screening, field trials,

etc. Second, the selected clones have to be propagated in large enough numbers for

deployment. For a rooted-cutting-based clonal program, this involves bulking up the

number of hedges (ramets) of a particular clone or group of selected clones through serial

propagation and then producing reforestation stock efficiently from the bulked-up clones.

Only those tested clones that can be propagated easily in sufficient numbers will be

economically feasible for deployment.

Loblolly pine is considered a difficult to root species (Wise and Caldwell 1994). In

populations that have not experienced any selection pressure for rooting ability, loblolly

pine has been reported to root near 50% (Foster 1990; Baltunis et al. 2005). Previous

rooting studies with loblolly pine have demonstrated substantial family and clonal

variation for rooting ability (Foster 1990; Baltunis et al. 2005), indicating the potential

for increasing rooting efficiency for both clonal deployment and through recurrent

selection and breeding. Selection for both rooting ability and field growth will be

necessary for a successful loblolly pine clonal forestry program based on rooted cutting

technology (Foster et al. 1985; Foster et al. 2000; Baltunis et al. 2005).

The objectives of this study were to (i) determine the genetic correlation between

rooting ability and 2nd year height, (ii) predict the genetic gain associated with selection









for rooting ability, (iii) predict the genetic gain associated with selection for 2nd year

height, and (iv) predict the genetic gain from combined selection for rooting ability and

2nd year height using a Monte Carlo selection index.

Materials and Methods

Population

The parental population consisted of twenty first-generation and ten second-

generation selections from the larger Loblolly Pine Lower Gulf Elite Population. Two

additional first-generation, slow-growing parents were included. The parental selections

represent the Atlantic Coastal Plain, Florida, and Lower Gulf provenances of loblolly

pine. These thirty-two loblolly pine parents were mated in a partial diallel design and

created 70 full-sib families from which approximately 2,200 seedling hedges were

generated and given unique clonal identifications (Appendix A). On average, each parent

was involved in about four crosses.

Rooting and Field Trials

The propagation of the rooted cuttings for this study has previously been described

(Baltunis et al. 2005; Chapter 2). But briefly, the seedling hedges were repeatedly

sheared to slow down the effects of maturation and increase the number of shoots

available for collection. Cuttings were set in five rooting trials over two years in May

2001, July 2001, January 2002, April 2002, and June 2002 in trials SpringOl, SummerOl,

Winter02, Spring02, and Summer02, respectively. Four to nine ramet clonal row plots

were set in four to six replications depending on trial (Chapter 2). At the time of

collection, the hedges were 13, 15, 22, 25, and 27 months old, respectively. Cuttings

were assessed for rooting nine to eleven weeks after they were set (Baltunis et al. 2005;

Chapter 2).









Six field trials (A, B, C, D, E, F) were established with rooted cuttings from the

three latter rooting trials. Three field trials each were established in Georgia and Florida

between October 2002 to April 2003 (see Table 3-1). The clonal propagules for each

field trial, however, came from a single rooting trial. Single-tree-plots of a clone were

established in four replications in each of two cultural treatments (high and low intensity)

in each test, except for Test E where there was only one cultural treatment and four

replications. The goal for the high intensity treatment was to push the trees to their

utmost potential by reducing competition and providing a non-limiting supply of

nutrients, while the low intensive culture provides insights into family and clonal

performance under a less optimal cultural regime (FBRC 2000). Previous analyses of

growth for this population had shown that type B genetic correlations exceeded 0.85

indicating little cultural treatment x genetic effect interaction, and therefore, cultural

treatment effects were ignored for the purposes of this study (Chapter 3).

Statistical Analyses

A bivariate parental linear mixed-effects model was used to obtain estimates of

variance and covariance components for rooting ability and 2nd year height using

ASREML (Gilmour et al. 2002):

[4-1] y, =X,b, +Zn, +Z,,u, +Zff +Zcc, +Z0o, +Zp, p +Zqq, + e,,

where y, is the vector of observations indexed (i) by rooting ability and 2nd year height,

b, is the vector of fixed effects (i.e. mean, trials and replications within trials) and X, is

the known incidence matrix relating the observations in y, to the fixed effects in b,


where X,b, roo
w e 0 X height b height











n, is the vector of random incomplete blocks nested within replication and test effects


-MVN 0, rootNC
0


0
Shegh
'heightC\


u, is the vector of random parent (female and male) general combining ability effects


-MVN (0, G 0 A) where G


F "2
'GCAroot
7GCAroot~giht


GCAro t Jght and A

0*GCAhelght


numerator relationship


matrix,


f, is the vector of random specific combining ability effects -MVN (0, S 0 Is) where


SCAro

C SCAroothight


SSCA"""" and I is an identity matrix of size equal to the number of
SCAeight _


full-sib families,


c, is the vector of random clones within full-sib family effects -MVN (0, C 0 Ic) where


C= CLONEot
CLONErooteght


SCLONEroo' and Ic is an identity matrix of size equal to the number
'CLONE ght


of clones,


o, is the vector of random test by full-sib family effects



~MVN 0, i r oot a
0 I height 7 i


p, is the vector of random test by clone within full-sib family effects


-MVN 0, orroot TESTxCLONE t
I 0


0
S 2 'TESTxLONE
height TESTxCLONE~h









q, is the vector of random replication within test by full-sib family effects


~MVN 0, i'
O height T



e, is the random vector of residual terms -MVN 0, roo R he ght
0 height 'ERRORigi _

Z,Z Z, Z Zf,, ZCI, Zo, ZP, and Zq, are the known incidence matrices relating the

observations in y, to effects in n,, u,, f,, c,, o,, p,, and q,, respectively, and I, is the

identity matrix of dimension equal to the number of observations for rooting ability or 2nd

year height.

Causal Components of Variance

Genetic parameters were estimated and standard errors were calculated for rooting

ability and 2nd year height from the bivariate analysis according to Foster and Shaw

(1988).
[4-2] ~2 V 1 1
[4-2] A = VA + +- VAA ...
4 16

is the estimate of additive genetic variance.
4-3 V 2 V 1 1
[4-3] D = SCA =D + VAA + AD -VDD
2 2 4

is the estimate of dominance genetic variance.
2[4-4] V 2& C 1 1 3
[4-4] V = C^LONE 2 ^GCA -3 SC = -V +- VAD DD "
4 2 4

is the estimate of epistatic genetic variance.

[4-5] VG = 2<^cA + -SCA + UCLONE

is the estimate of total genetic variance.










j^ 2 +(2 N 2 +j2 12 1 2
[4-6] P = 2 GCA SCA C cLONE TESTxFAM TESTxCLONE REPxFAM ERROR

is the estimate of the phenotypic variance.

^2 ^2 ^2 ^2 ^2 ^2
[7 = 2 SCA CLONE TESTxFAM 'TESTxCLONE REPxFAM ERROR
[4-7] VP- dGC + + + + + +
Sf fc tf tfc tr trfcn

is the phenotypic variance of half-sib family means, f= harmonic mean number of full-

sib families per half-sib family, c = harmonic mean number of clones per full-sib family,

t = number of trials, r = harmonic mean number of replications per test, and n = harmonic

mean number of ramets per clone per replication per test (n = 1 for field trials).

^2 ^2 ^2 ^2 ^2
[A- v 2 2 CLONE TESTxFAM TESTxCLONE REPxFAM ERROR
[4-8] Vp = 2CGA + CSCA + + + + +
c t tc tr ctrn

is the phenotypic variance of full-sib family means.

^2 "2 ^2 ^2
1 2 (2 j 2 2 'TESTxFAM TESTxCLONE 'REPxFAM ERROR
[4-9] V GC = 2 A + SCA +CLONE + + + +
CL t t tr trn

is the phenotypic variance of clonal means.

Heritability Estimates

Heritability estimates have previously been reported for this population for rooting

ability (Baltunis et al. 2005; Chapter 2) and 2nd year height (Chapter 3). However, since

rooting ability and 2nd year height were analyzed using a bivariate model in the current

study, heritabilities were estimated based on the genetic parameter estimates from the

bivariate model and differed slightly from those reported previously. Standard errors

were calculated using the Taylor series expansion method (Kendall and Stuart 1963;

Namkoong 1979; Huber et al. 1992; Dieters 1994).











[4-10] h2 =A GCA
S 2 +"2 +2 +E2 2 +"2 2
S2cGCA SCA CLONE TESTxFAM TESTxCLONE REPxFAM ERROR

the individual tree narrow-sense heritability.


[4-111 2 G GCA ^SCA CLONE
^-12 2 2 2 2 12 '2 2
VP 2 GCA SCA 'CLONE + TESTxFAM + TESTxCLONE + UREPxFAM + RROR

the individual tree broad-sense heritability.

^2
[4-12] H = CA is the half-sib family mean heritability.
HS


^2 "2
[4-13] i2 2= GCA SCA i the full-sib family mean heritability.
Vp-
FS


r2 ^2 ^2
[4-14] 2 = GCA SCA CLONE is the clonal mean heritability.
CL
vpi

Type B Genetic Correlations

Although the same genotypes, or clones, were tested in the rooting and field trials,

these measurements were not necessarily taken on the same ramet. Therefore, type B

genetic correlations between rooting ability and 2nd year height for general combining

ability, full-sib family value, and the total genetic value were calculated. Standard errors

of these estimates were calculated using the Taylor series expansion method (Kendall and

Stuart 1963; Namkoong 1979; Huber et al. 1992; Dieters 1994).


OGCAotg
[4-15] ; 2 is the type B genetic correlation between rooting ability
GCAroot X GCAeIght


and 2nd year height for additive effects, and GCA is the covariance between general


combining ability effects for rooting ability and 2nd year height..










[4-16] 2r GCA4h," +SCA,,g is the type B genetic correlation
"2 2 ^2 ^ 2
S^GCA ot + SCA t 2GCAheght + 4SCA, ,

between rooting ability and 2nd year height for full-sib families, and

2 ^GCAoot&,g + SCAh is the covariance between the full-sib family effects for rooting
^ootheigh ^ footheght

ability and 2nd year height.


[2 -17] s te 2 ty2 2
I2I GCAo + SCA, + CLONE 21cGCA, + SCAg + t CLONE )

B genetic correlation between rooting ability and 2nd year height for the total genetic

value of clones, and 2,GCAooh + i -SCArooti ,g + i 'CLONErooet is the covariance between the


total clonal value effects for rooting ability and 2nd year height.

Genetic Gain

Genetic gain was estimated for a number of deployment options based on various

selection scenarios using the BLUP values from the bivariate analysis. All deployed

populations were assumed to be propagated as rooted cuttings. An additional assumption

for gain calculations was that the seventy full-sib families and all clones within families

were available for deployment. The deployment strategies considered were 1) half-sib

family deployment, 2) full-sib family deployment, and 3) clonal deployment. For a half-

sib family deployment option, the predicted value for each parent (general combining

ability) was used to select the best half-sib family for deployment as rooted cuttings.

Full-sib family values were calculated by summing the predicted values for the female

general combining ability, male general combining ability, and the full-sib family

specific combining ability. The genetic gain over the trait mean was determined for

deployment of the best full-sib family.









The total genetic value of a clone was determined by summing the predicted values

for the female general combining ability, male general combining ability, the full-sib

family specific combining ability, and clone within full-sib family. Several clonal

deployment strategies were compared. First, the genetic gain was determined for

deployment of the single best clone for each trait. A second clonal deployment strategy

was considered by selecting the top ten full-sib families and then deploying the single

best clone from each of these ten families. Two additional clonal deployment strategies

were compared by selecting the top 10% and 1% of clones (out of 2200 possible clones)

using an unrestricted selection index for 2nd year height and rooting ability. The total

genetic values of the clones were weighted with the following weights: 1:0, 0.9:0.1,

0.8:0.2, ..., 0.2:0.8, 0.1:0.9, 0:1 for 2nd year height and rooting ability, respectively, in the

Monte Carlo index (Cotterill and Dean 1990). All gains were expressed as the

percentage gain over the mean of the trait:


[4-18] %Gain = x 100%, where


x5 is the average predicted value for trait i of the selected population, and

7, is the population mean for either rooting ability or height.

In addition genetic gain was calculated using theoretical gain formulae assuming an

unrelated population (Falconer and Mackay 1996), and these estimates were compared

with the predicted genetic gain based on BLUP values for the half-sib family, full-sib

family, and best clone deployment options. The following formulae were used:









[4-19] Gain = ihs V is the genetic gain associated with selection of the best half-


sib family, and ihs is the selection intensity corresponding to selecting one out of thirty-

two half-sib families (i = 2.07) and assuming all half-sib families are unrelated,

[4-20] Gain = if,Hi V is the genetic gain associated with selection of the best full-


sib family, and if, is the selection intensity corresponding to selecting one out of seventy

full-sib families (i = 2.38) and assuming all full-sib families are unrelated,

[4-21] Gain = i -2~p is the genetic gain associated with clonal selection, and i, is

the selection intensity corresponding to selecting one out of 2,206 clones (i = 3.58) and

assuming all individuals are unrelated.

Results and Discussion

Causal Components of Variance

The genetic parameter estimates from the bivariate analysis of rooting ability and

2nd year height in loblolly pine clones were consistent with the estimates from their

respective univariate analyses, and both traits showed genetic variation (Table 4-1;

Appendix D). Additive genetic variation accounted for the majority of the genetic

variation associated with rooting ability and 2nd year height (Table 4-1). Dominance

genetic variation contributed a minor portion to the total genetic variation for both traits.

Epistasis appears to be more important for rooting ability than 2nd year height as

evidenced by a negative estimate of epistasis for 2nd year height (Table 4-1). However,

estimates of epistasis may be confounded with C effects since rooted cuttings of a clone

originated from a single ortet.









Type B Genetic Correlations

The bivariate analysis of five rooting trials and six field trials also allowed for

estimation of the genetic covariance between rooting ability and 2nd year height for

parental effects, full-sib family effects, and the total genetic value of clones within full-

sib family. There was a positive genetic relationship between rooting ability and 2nd year

height at all three genetic levels (Table 4-1). The genetic correlation at the parental level

between rooting ability and 2nd year height (~C ) was 0.32. At the full-sib family level,

the genetic correlation between traits (BFS ) was 0.39. The correlation of total genetic

values of clones for rooting ability and 2nd year height (mB ) was 0.29.

Previous studies have also reported positive correlations between rooting traits and

growth. For example, Paul et al. (1993) reported that 1st year and 5th year height growth

had strong genetic correlations with rooting (0.61 and 0.69, respectively) in western

hemlock clones (Tsuga heterophylla (Raf.) Sarg.). In another study, Foster et al. (1985)

reported that the genetic correlation between rooting ability and growth of western

hemlock clones in a growth chamber was 0.37. A weak, but positive, relationship

between the number of roots on loblolly pine cuttings and subsequent field growth has

been reported (Foster et al. 2000). On the other hand, Goldfarb et al. (1998) reported that

the phenotypic correlation between the number of roots and first year field growth in

loblolly pine rooted cuttings was negligible.

C effects relating to the season in which the cuttings were set has previously been

discussed in this population (Chapter 3). Further evidence of the presence of C effects

can be seen from the genetic correlations between the total genetic value of clones for

rooting ability and 2nd year height based on analyses involving a single rooting and field









trial (Table 4-2). Rooted cuttings planted in each field trial originated from a single

sticking date, and the highest genetic correlations (rBTG ) occurred between traits from

these trials. For example, rooted cuttings from the Winter02 rooting trial were planted

only in the Field A trial. The total genetic correlation between rooting ability from

Winter02 and 2nd year height from Field A was 0.24, while 2nd year height from Field A

had much lower correlations with rooting ability from any of the other sticking dates

Table 4-1. Means, variance component estimates, heritabilities, and genetic correlations
from the bivariate analysis of rooting ability and 2nd year height. Standard
errors are given in parentheses.
Rooting Ability 2nd Year Height
S42 % 210 cm

VA 0.0135 (0.004) 399.3 (119.6)
VD 0.0014 (0.002) 77.3 (39.6)
V, 0.0085 (0.002) -94.9 (63.7)
VG 0.0235 (0.002) 381.7 (59.9)

VP 0.2299 (0.002) 1836 (61.0)
Vp- 0.0038 (0.001) 108.1 (29.6)
Vp- 0.0085 (0.002) 231.1 (59.1)
VP~ 0.0286 (0.002) 434.1(59.9)
h2 0.059 (0.02) 0.22 (0.06)
H2 0.102 (0.01) 0.21 (0.03)

Hf2 0.88 (0.04) 0.92 (0.03)
HS
fi2 0.84 (0.04) 0.95 (0.01)
FS
fi2 0.82 (0.01) 0.88 (0.02)
CL
rB 0.32(0.19)
rB, 0.39 (0.17)
rB, 0.29 (0.08)









(Table 4-2). This indicates that the vigor of the hedge at the time of rooting may have

influenced rooting and subsequent field growth.

The biological significance of positive correlations between rooting and 2nd year

height in the current study is unclear. Goldfarb et al. (1998) reported that root

morphological traits were not meaningful for subsequent field growth after one year of

growth for loblolly pine rooted cuttings. However, it has generally been acknowledged

that stock plant vigor is related to rooting. Perhaps the vigor of the stock plant is an

indicator of the metabolic activity of cuttings during root initiation. The more

metabolically active genotypes may have rooted quicker and formed better root systems.

In the current study, rooting was assessed at nine to eleven weeks, and selection may

have indirectly been for rate of rooting as opposed to strictly rooting ability. Clones that

rooted quickly during the rooting period may have had increased metabolic activity. A

positive genetic correlation between rooting and growth implies that some of the same

genes are responsible for the expression of each trait. There may be genes in common for

rooting and height relating to metabolic activity. Genotypes that have a tendency for

increased rates of metabolic activity may root at higher frequencies and grow larger than

genotypes that do not have these alleles.

Genetic Gain

The importance of selecting for rooting ability in a clonal forestry program based

on rooted cuttings has long been recognized (Foster et al. 1984; Foster 1990; Baltunis et

al. 2005). Genetic gains in field traits will not be realized if the clones can not be

propagated efficiently for deployment. Therefore, increases in rooting ability could have

broad economic impacts on a clonal forestry program (Foster et al. 1984). Positively

correlated traits imply that selection for one trait should also lead to improvement in the









second trait. In the case of rooting ability and growth, positive genetic correlations can

lead to substantial gains for both traits in a clonal forestry program based on rooted

cuttings.

The genetic gains in rooting ability and 2nd year height based on BLUP values were

compared for a number of deployment strategies. Selecting the top half-sib family for

rooting ability would result in a gain of 36% in rooting ability (Figure 4-1). Deployment

of the top half-sib family selected for rooting ability would result in a genetic gain of

5.4% in 2nd year total height (Figure 4-2). Selecting the highest ranking half-sib family

for 2nd year height would result in a gain of 14.8% and 8.1% in rooting ability and 2nd

year height, respectively (Figure 4-1 and Figure 4-2). The genetic gain predictions using

theoretical calculations were slightly lower than the gain predictions based on BLUP

values. The gain in rooting ability associated with selecting the best half-sib family for

rooting was 26.7% (using Equation 4-19). The gain in height by selecting the top

growing half-sib family was 9.4% (using Equation 4-19).

Slightly higher gains can be achieved by selecting the top full-sib family.

Propagation of the top full-sib family selected for rooting ability results in genetic gains

in rooting ability of about 43% over the population mean (Figure 4-1). Gains of nearly

9% in 2nd year total height could be expected by deploying the top rooting full-sib family

(Figure 4-2). If the best growing full-sib family was propagated and deployed, then the

genetic gains in rooting ability (Figure 4-1) and 2nd year height (Figure 4-2) would be

8.6% and 10.1%, respectively. The theoretical gain in rooting ability for full-sib

selection was equivalent to the estimate based on BLUP values. However, the theoretical









gain for full-sib selection for 2nd year height was 16.3% which was higher than the

estimate based on BLUPs.

Table 4-2. The total genetic correlation (?B ) between rooting ability and 2nd year height
from analyses of a single rooting trial and field trial. Shaded values indicate
the rooting trial in which cuttings originated from for their respective field
trials. Standard errors are given in parentheses.
Field A Field B Field C Field D Field E Field F

l 0.002 0.17 0.17 0.21 0.16 0.25
(0.07) (0.07) (0.07) (0.07) (0.07) (0.07)

0.08 0.07 0.17 0.21 0.14 0.12
SummerO1
(0.06) (0.07) (0.07) (0.07) (0.07) (0.07)

0.24 0.16 0.23 0.27 0.24 0.18
Winter02
(0.07) (0.07) (0.07) (0.08) (0.08) (0.08)

0.06 0.25 0.26 0.34 0.16 0.24
png (0.07) (0.07) (0.08) (0.07) (0.07) (0.08)

0.10 0.21 0.16 0.21 0.13 0.28
Summer02
(0.07) (0.07) (0.08) (0.08) (0.08) (0.08)



Clearly, the most gain for either trait would be achieved by selecting the single best

clone (Figure 4-1 and Figure 4-2). For instance, by selecting the top clone for rooting

ability, the genetic gain based on clonal predicted values in rooting was 96% (Figure 4-1)

indicating that approximately 82% of the ramets would root from this clone (42% mean

rooting plus 0.96 x 42% = 82%). However, selecting the top rooting clone from this

population would result in a decrease (genetic loss) in overall 2nd year height (Figure 4-

2). On the other hand, selecting the top clone for 2nd year height would result in a gain of

nearly 27% in 2nd year height (Figure 4-2) and 43% in rooting ability (Figure 4-1) over

the population mean for these traits. Genetic gain in rooting ability based on theoretical







71


calculations was greater than those based on BLUP values for selection of the single best

clone and was 118%, while the gain in 2nd year height for the top growing clone was

31.3% (using Equation 4-21).


100

90 0% Gain in Rooting Ability when
>h Selecting for 2nd Year Height
80 -
0% Gain in Rooting Ability when
S70- Selecting for Rooting Ability


0
60

o 50
40

30
o 20
10 .. -.


Best Half-Sib Best Full-Sib Best Clone from Best Clone
Family Family Best 10 Full-Sib
Families


Figure 4-1. The genetic gain in rooting ability (%) over the population mean for
deployment of the best half-sib family, full-sib family, best clone from the
best ten full-sib families, and the single best clone when selecting for rooting
ability or 2nd year height.


In order to address genetic diversity issues and the risk associated with deploying a

single clone, a second clonal deployment option was considered by selecting the best

clone from each of the ten highest ranking full-sib families. This strategy resulted in

genetic gains nearly double that of the full-sib family deployment strategy. When the

trait selected is rooting ability, the best clone in the top ten full-sib families would result

in gains of 77% in rooting ability (Figure 4-1) and 5.6% in 2nd year height (Figure 4-2).








When the trait selected is for 2nd year height, then the genetic gains would be 37.6% in

rooting ability (Figure 4-1) and 18.3% for 2nd year height (Figure 4-2).


E % Gain in 2nd Year Height when
Selecting for Rooting Ability
0% Gain in 2nd Year Height when
Selecting for 2nd Year Height


rf]


ifi


. . . . . .


Best Half-Sib
Family


Best Full-Sib
Family


Best Clone from
Best 10 Full-Sib
Families


Pest:


'lone


Figure 4-2. The genetic gain in 2nd year height (%) over the population mean for
deployment of the best half-sib family, full-sib family, best clone from the
best ten full-sib families, and the single best clone when selecting for rooting
ability or 2nd year height.

Deployment of well-tested clones can result in genetic gains in both rooting ability

and 2nd year total height. Both traits should be considered for a successful loblolly pine

clonal forestry program based on rooted cutting technology. The responses to selection

when considering both rooting ability and 2nd year height were compared for a number of

selection indices (Figure 4-3). The gain in rooting ability associated with selecting the

top 10% of clones ranged from 23.8% when only 2nd year height was considered to

57.6% when only rooting ability was considered (Figure 4-3). Higher gains can be









achieved by increasing the selection intensity. When only the top 1% of clones was

selected, the gain in rooting ability ranged from 33.4% when only 2nd year height was

considered to 80.6% when only rooting ability was considered (Figure 4-3). The genetic

gain in 2nd year height associated with deployment of the best 10% of clones ranged from

4.8% when only rooting ability was considered to 12.6% gain when only 2nd year height

was considered (Figure 4-3). Similarly, genetic gain in 2nd year height ranged from 3.8%

to 18.5% when the best 1% of clones were selected (Figure 4-3).

Clonal forestry programs that consider multiple traits need to optimize their

selection strategies for deployment populations. Arbitrarily setting the optimum selection

weights on 2nd year height and rooting ability to 90% of the maximum genetic gain

obtainable for a single trait, then the optimum selection weights can be compared. For

example, if the top 10% of clones are selected and 90% of the maximum gain in rooting

ability is considered, then the optimum weights on 2nd year height and rooting ability

correspond to weights of 0.7 and 0.3, respectively. These weights result in genetic gains

of 53.6% in rooting ability and 8.7% in 2nd year height (Figure 4-3). If 1% of the clones

are selected, then the optimum weights are 0.6 and 0.4 on 2nd year height and rooting

ability, respectively, and correspond to gains of 75.9% in rooting ability and 11% in 2nd

year height (Figure 4-3).

If 90% of the maximum gain obtainable for 2nd year height is the criterion, then the

optimum weights for both deployment options are 0.9 and 0.1 on 2nd year height and

rooting ability, respectively. The genetic gains associated with selecting the top 10% of

clones utilizing these weights results in gains of 40.4% in rooting ability and 11.9% in 2nd

year height (Figure 4-3). If only the top 1% of clones are deployed using these selection









weights, then genetic gains of 54.2% in rooting ability and 17.4% in 2nd year height are

obtainable (Figure 4-3). The maximum change in genetic gain of rooting ability results

from increasing the weight on rooting ability from 0 to 0.1. In fact, an additional 17-21%

gain in rooting ability can be obtained in the deployment population by increasing the

weight on rooting ability from 0 to 0.1.

When 10% of the clones were selected (220), then the number of full-sib families

represented in the deployment population ranged from 38 to 45 depending on selection

index (Table 4-3). When only the top 1% of clones were selected (22), the number of full

sib families represented in the deployment population varied from 9 to 15 (Table 4-3).

Maximum genetic gains can be achieved by ignoring relatedness among selections.

However, if no constraints are placed on the relatedness of selections, then there is a

tendency to make many selections from the better families. Although the approximate

average number of clones selected per full-sib family was five when 10% of the clones

were selected, nearly 45% of these selections came from only five families when only

rooting ability was selected. Similar trends can be seen when selecting 1% of the clones.

For example, when all of the selection weight is on rooting ability, then there was an

average of 2.44 clones per full-sib family selected (Table 4-3). However, 15 of the 22

selections came from three families. On the other hand, when considering the number of

half-sib families represented in the selected population, 26 to 30 out of 32 possible

parents are represented when 10% of the clones are selected. Thirteen to 17 parents are

represented in the deployment population when 1% of the clones are selected (Table 4-3).

Therefore, at least in this population, there may be sufficient genetic diversity in the

deployment population even with higher selection intensities. Isik et al. (2005) reported










genetic gains in growth near 30% for loblolly clones from a different population

irregardless of any restrictions on the relatedness of selections.

Conclusion

Rooting ability and 2nd year height are heritable traits in loblolly pine, and both

traits showed substantial clonal variation. There was a positive genetic correlation

between rooting and height at the parental, full-sib family, and clonal levels. Genetic

gains in rooting ability and 2nd year height are possible as demonstrated by a number of

deployment strategies. For hard-to-root species, like loblolly pine, a successful clonal


90%

80% ....

70% .

60%-

7 50% -

4 40%
S/
30%

20%

10% -


^' <^ C' O' 00 b'

Weight on 2nd Year Height I Weight on Rooting Ability
% Gain in 2nd Year Height (10%) % Gain in 2nd Year Height (1%)
% Gain in Rooting Ability (10%) -% Gain in Rooting Ability (1%)



Figure 4-3. Responses to selection in rooting ability and 2nd year height with various
selection indices for two clonal deployment options: 10% of clones selected
and 1% of clones selected.









forestry program based on rooted cuttings must consider both rooting ability and growth

when making selections. Because here these traits are positively correlated, selection for

one trait should lead to positive gains in the other. Or, early culling of clones based on

poor rooting will not negatively effect selection of clones for growth when field data

becomes available.

Clonal forestry is not a breeding method to develop better genotypes. However,

clonal forestry is a method to mass-produce well-tested genotypes. Short-term genetic

gains may be maximized through deployment of well-tested clones, but long-term gains

need to involve both clonal selection and recurrent selection for additive genetic variation

through repeated selection and breeding. Restrictions on the relatedness of selections

will be necessary when making selections for a breeding population or when deployment

will be traditional zygotic seedlings from seed orchards in order to reduce detrimental

effects of inbreeding depression.









Table 4-3. The number of full-sib families (half-sib families) and average number of
clones per full-sib family (half-sib family) selected from selecting 10% or 1%
of the top clones using the combined selection index.
10% Clones Selected 1% Clones Selected
nd Ave. # Ave. #
Weight on 2nd # FS # FS#
S Clones per Clones per
Year Height: Families FS Famiy Families Fs F
FS Family FS Family
Weight on (# HSS (# HS
Rooting Ability families) Fam ) families) F
Family) Family)
1:0 38 (26) 5.8(16.9) 14 (17) 1.6 (2.6)
0.9:0.1 41 (28) 5.3 (15.7) 15 (17) 1.5 (2.6)
0.8:0.2 43 (28) 5.1 (15.7) 12 (15) 1.8 (2.9)
0.7:0.3 43 (30) 5.1 (14.7) 13 (16) 1.7 (2.8)
0.6:0.4 45 (30) 4.9(14.7) 10 (12) 2.2 (3.7)
0.5:0.5 45 (29) 4.9(15.2) 12 (15) 1.8 (2.9)
0.4:0.6 45 (29) 4.9(15.2) 11 (15) 2.0 (2.9)
0.3:0.7 45 (29) 4.9(15.2) 11 (15) 2.0 (2.9)
0.2:0.8 45 (29) 4.9(15.2) 10(14) 2.2 (3.1)
0.1:0.9 45 (29) 4.9(15.2) 9 (13) 2.4 (3.4)
0:1 45 (29) 4.9(15.2) 9 (13) 2.4 (3.4)














CHAPTER 5
CONCLUSION

Loblolly pine is the most important commercial tree species in the southeastern

United States. Several forest industries in the southeastern United States are deploying

full-sib families of loblolly pine operationally. In addition, many of these companies are

pursuing clonal forestry programs with loblolly pine. Clones need to be well-tested

before they can be deployed. This involves the accumulation of reliable data from

propagation, growth traits, disease resistance, etc. Those well-tested clones that meet the

selection criteria can lead to substantial genetic gain.

Several key results can be concluded from this research. With rooting data from

2,200 clones from 70 full-sib families, the current study gives better estimates of genetic

components of variance for rooting than several previous studies. These results show a

great deal of genetic variation for rooting among families and clones of loblolly pine.

Combined with moderate to high estimates of family- and clonal-mean heritabilities and

type B correlations for rooting ability in different rooting trials, these results indicate the

potential for increasing rooting efficiency by selecting good rooting families and clones

or culling poor rooters.

Field testing of clones is a necessary component to clonal forestry programs based

on rooted cuttings. In addition, genetic field trials established with clones and seedlings

from the same full-sib families provide an opportunity for comparing both half-sib and

full-sib family performances for both propagules. Based on the results in Chapter 3,

several conclusions can be drawn. First, clonally replicated seedling trials of loblolly









pine provide genetic information with greater precision than zygotic seedlings. Second,

genetic correlations between propagule types for the growth traits were highly favorable.

These high genetic correlations between propagule types reassure that parental and full-

sib family rankings are stable regardless of propagule type. This implies that parental

and full-sib family rankings based on existing seedling progeny trials could be used to

select parents and families that perform well when they are deployed as rooted cuttings.

Third, little genotype x environment interaction was observed across sites at the parental,

family, and clonal level for all traits. However, there appears to be a carry-over effect

relating to the season in which the cuttings were rooted for the clonal material. Finally,

randomization is essential at all stages in testing when estimating genetic parameters.

The lack of randomization for the seedling population apparently resulted in a problem

with partitioning of the genetic variance, causing full-sib families to appear more

different and inflating estimates of dominance genetic variation.

Rooting ability and 2nd year height are heritable traits in loblolly pine, and both

traits showed substantial clonal variation. There was a positive genetic correlation

between rooting and height at the parental, full-sib family, and clonal levels. Genetic

gains in rooting ability and 2nd year height are possible as demonstrated by a number of

deployment strategies. For difficult-to-root species, like loblolly pine, a successful clonal

forestry program based on rooted cuttings must consider both rooting ability and growth

when making selections. Because here these traits are positively correlated, selection for

one trait should lead to positive gains in the other. Or, early culling of clones based on

poor rooting will not negatively affect selection of clones for growth when field data

become available.









Clonal forestry is not a breeding method to develop better genotypes. However,

clonal forestry is a method to mass-produce well-tested genotypes. Short-term genetic

gains may be maximized through deployment of well-tested clones, but long-term gains

need to involve both clonal selection and recurrent selection for additive genetic variation

through repeated selection and breeding. Restrictions on the relatedness of selections

will be necessary when making selections for a breeding population or when deployment

will be traditional zygotic seedlings from seed orchards in order to reduce detrimental

effects of inbreeding depression.



























APPENDIX A
LOBLOLLY PINE PARTIAL DIALLEL MATING DESIGN. THIRTY-TWO
PARENTS WERE CROSSED TO GENERATE 70 FULL-SIB FAMILIES.





















Parent 2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
1 X X X X
2 X X X X X
3 X X X
4 X X X
5 X X X X
6 X X X
7 X X
8 XX X
9 X X
10 X
11 X X
12 X X X
13 X X
14 X X X X X
15 X X
E 16 X X
17 X
18 X
19
20 X X X t
21 X X
22 X X X
23 X X
24 X
25 X X
26 X X
27 X
28 X X
29 X
30 X X
31 X
32

















APPENDIX B
VARIANCE COMPONENT ESTIMATES FOR ROOTING ABILITY

Table B-1. Observed variance component estimates for rooting of loblolly pine stem
cuttings from single-trial analyses.

SpringO 1 SummerO 1 Winter02 Spring02 Summer02
"2
TRAY 0.003977 0.005357 0.002891 0.010084 0.004796
"2
OGCA 0.005018 0.004629 0.005383 0.004451 0.003817
"2
SCA 0.002683 0.000771 0.001363 0.001868 0.00187
"2
aCLONE 0.041427 0.039236 0.024317 0.026364 0.023738
"2
-REPxFAM 0.000836 0.000559 0.000637 0.000661 0.001043
S 0.190089 0.177991 0.204829 0.200698 0.142
ErMOR 0.190089 0.177991 0.204829 0.200698 0.142
















Table B-2. Observed variance component estimates for rooting of loblolly pine stem cuttings from pair-wise-trial analyses.

SpringO 1 SpringO 1 SpringO 1 SpringO 1 SummerO 1 SummerO 1 SummerO 1 Winter02 Winter02 Spring02
SummerO 1 Winter02 Spring02 Summer02 Winter02 Spring02 Summer02 Spring02 Summer02 Summer02

-2 0.004751 0.003637 0.007712 0.004474 0.004115 0.007929 0.005059 0.006905 0.003774 0.007824
TRAY


(GCA 0.003434 0.002923 0.00433 0.003929 0.003191 0.002844 0.002557 0.003065 0.002347 0.003915

"2
(SCA 0.001409 0.000417 0.001262 0.001876 0.00067 0.000654 0.001027 0.001107 0 0.001147

"2
'CLONE 0.021605 0.011304 0.017015 0.014235 0.012645 0.014791 0.014151 0.011209 0.010850 0.013135
J^j 2


T-ESxGCA 0.001464 0.002034 0.000422 0.000688 0.001846 0.001380 0.001472 0.001609 0.002123 0.0004


OESTxFAM 0.000302 0.001758 0.001299 0.000539 0.000396 0.000572 0.000029 0.000423 0.001480 0.000485
00
2
TESTxCLONE 0.018862 0.021637 0.017991 0.022116 0.019564 0.018869 0.020504 0.014919 0.014083 0.012163


"2
REpxFAM 0.000674 0.000764 0.000703 0.000956 0.000598 0.000613 0.000797 0.000646 0.000825 0.000767


ROR 0.183321 0.198958 0.197306 0.162528 0.192458 0.192031 0.159335 0.202419 0.174684 0.177164
0E OR 0.183321 0.198958 0.197306 0.162528 0.192458 0.192031 0.159335 0.202419 0.174684 0.177164











Table B-3. Observed variance components for rooting ability from the across-trial
analysis using all five rooting trials.

Variance Component Estimate
^2
oRAY 0.005785
^2
GCA 0.002921
^2
aSCA 0.001062
^2
rCLONE 0.01638
^2
TESTrGCA 0.001364
^2
-TESTxFAM 0.000691
^2
OTESTxCLONE 0.017349
^2
(REPxFAM 0.000713
2E
ERROR 0.184907

















APPENDIX C
VARIANCE COMPONENT ESTIMATES FOR EARLY GROWTH TRAITS OF
LOBLOLLY PINE CLONES AND SEEDLINGS FROM THE SAME FULL-SIB
FAMILIES

Table C-1. Observed variance components for loblolly pine clones from the across-trial
analyses of 1st year height, 2nd year height, height increment, and crown width.
A separate error variance was modeled for each trial.

1st Year Height 2nd Year Height Height Crown Width
Increment

INc 53.4293 300.188 123.143 105.75
"2
GCA 17.4525 100.363 38.141 27.5703
"2
(SCA 2.9992 21.601 7.5808 4.93
"2
CLONE 32.3599 157.931 59.3416 56.4437
"2
TESTxGCA 4.1225 13.3866 7.9053 5.9029
"2
TESTxFAM 1.2003 3.7766 1.2451 1.3288
"2
TESTxCLONE 22.4253 82.7545 29.0673 23.4036
"2
TxCxGCA 0.4069 4.5692 2.942 2.2437
"2
TxCxFAM 0.0000008 0.3873 0.0000005 1.1495
"2
xCxCLONE 0.0000008 17.1395 14.179 10.7818
"2
REPxGCA 0.3275 1.0281 1.5638 0.6583
"2
REPxFAM 1.9242 5.511 1.5806 0.8574
^2
jEROR 512.142 1700.04 692.262 610.815
"2
ERRORB 279.24 853.321 427.247 306.51
"2
-ERRORc 527.513 2061.14 928.903 843.627

ERRORD 264.114 1110.76 864.549 329.026
"2
ERRORE 313.481 1124.11 682.649 423.413

OR 123.318 804.457 527.064 296.22
L-RORF 123.318 804.457 527.064 296.22











Table C-2. Observed variance components for loblolly pine seedlings from the across-
trial analyses of 1st year height, 2nd year height, height increment, and crown
width. A separate error variance was modeled for each trial.

1st Year Height 2nd Year Height Height Crown Width
Increment
"2
rGCA 3.6682 29.9688 12.9402 13.7566
"2
SCA 4.3654 20.8173 10.4732 5.0053
"2
TESTxGCA 1.5235 8.4371 8.6503 3.8956
"2
TESTxFAM 8.5693 25.1585 8.5047 7.9082
"2
TxCxGCA 0.0000007 0.0000004 0.0000002 0.5247
i2
TxCxFAM 0.6036 0.0000001 1.375 1.0372
"2
REPxGCA 0.0000001 0.0000001 0.0000001 1.2412
"2
oREPxFAM 6.3223 55.867 26.1737 15.4417

RROR2 599.229 1621.95 713.814 590.37

ERRORB 361.489 1104.24 567.689 325.987
"2
ERRORc 508.95 2524.33 1246.59 1110.51
"2
ERRRD 294.534 1356.8 1059.94 501.614
"2
ERRORE 379.556 1766.73 1109.52 610.975

OR 101.715 755.187 585.714 341.353
L-RORF 101.715 755.187 585.714 341.353
















APPENDIX D
VARIANCE COMPONENT ESTIMATES FROM THE BIVARIATE ANALYSES OF
ROOTING AND 2ND YEAR HEIGHT

Table D-1. Observed variance component estimates from the bivariate analyses of
rooting ability from SpringOl and 2nd year height from each of the field trials.

SpringO1 Field A Field B Field C Field D Field E Field F

iNCr, 0.0039 0.0039 0.004 0.0039 0.004 0.004

Ui2 219.85 232.03 801.24 390.46 105.53 89.22
"2
GCArt 0.0051 0.0052 0.0052 0.0051 0.0052 0.0051

-GCAoot,, 0.0778 0.219 0.2515 0.3664 0.3268 0.1727
"2
GCA,,g 93.85 77.71 134.98 134.81 144.59 98.15

S-C, 0.0027 0.0026 0.0026 0.0027 0.0026 0.0027

SCAroo,,gh 0.0654 -0.1556 -0.0148 0.0203 0.0815 0.1294
"2
SCA,,,,, 28.34 3.7 8.13 28.18 43.77 30.08

CLONEt 0.0415 0.0415 0.0414 0.0415 0.0414 0.0414

CLONEroo,_tgi, -0.21 0.3412 0.4084 0.3997 0.2865 0.7351
"2
CLONEh,,ght 386.08 191.46 250.52 230.24 426.68 207.67

REPxFAMt 0.0008 0.0008 0.0008 0.0008 0.0008 0.0008
"2 0.0996 5.31 23.68 15.47 24.33 15.44
"2
OERRORo 0.1901 0.1901 0.1901 0.1901 0.1901 0.1901
2OR 1641.17 879.86 2003.72 1127.51 1072.05 834.69
E ORhgt, 1641.17 879.86 2003.72 1127.51 1072.05 834.69











Table D-2. Observed variance component estimates from the bivariate analyses of
rooting ability from SummerO 1 and 2nd year height from each of the field
trials.

Summer01 Field A Field B Field C Field D Field E Field F
"2
JiNC 0.0053 0.0053 0.0053 0.0053 0.0053 0.0053

CU2 220.06 232.4 801.8 389.37 105.55 89.37
"2
GCAot 0.005 0.005 0.005 0.005 0.005 0.0049

'GCAoot, 0.243 0.1731 0.2933 0.3457 0.2579 0.0796
"2
GCA,, t 98.15 79.12 135.89 138.56 147.16 98.3

2sCA 0.0007 0.0007 0.0007 0.0007 0.0007 0.0008

sCAo,,, ,, 0.0397 0.0141 -0.0125 0.0078 0.0616 0.1556
"2
sCA,,,g 26.74 3.33 6.33 24.69 40.85 26.5
^2
CLONEt 0.0393 0.0392 0.0393 0.0393 0.0392 0.0393

CLONErootgi -0.0944 -0.0803 0.3137 0.4131 0.2729 0.2437
2
'CLONE,,gh, 385.69 190.78 250.56 230.58 426.65 206.64
2
REPxFAMt 0.0006 0.0006 0.0006 0.0006 0.0006 0.0006
U2 0.0757 5.3 23.92 15.69 24.4 15.71

"2
OERRORo 0.1779 0.1779 0.1779 0.1779 0.1779 0.1779
OR 1641.1 879.8 2003.35 1127.44 1072.03 834.0
OEMRORho gh 1641.1 879.8 2003.35 1127.44 1072.03 834.0




Full Text

PAGE 1

GENETIC EFFECTS OF ROOTING ABILITY AND EARLY GROWTH TRAITS IN LOBLOLLY PINE CLONES By BRIAN STEPHEN BALTUNIS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Brian Stephen Baltunis

PAGE 3

iii ACKNOWLEDGMENTS I would like to thank Drs. Timothy White Dudley Huber, Barry Goldfarb, Hank Stelzer, John Davis, and Rongling Wu for serving on my advisory committee. Special thanks go to Tim White and Dudley Hube r for their guidance, advice, and support throughout my entire program. I also would like to thank the Forest Bi ology Research Cooperative for financial support for this research, the members of th e FBRC for establishing field trials, and especially International Paper Company for providing the facilities for the propagation of the rooted cuttings for this project. Many pe ople were involved w ith bringing this study together, and I would like to express my gra titude to all the faculty, staff, workers, and fellow graduate students who helped. My a ppreciation especially goes to Brian Roth; without his effort, this proj ect could never have happened. Finally, I would like to thank my wi fe, Jacqueline, for her patience and encouragement these last five years.

PAGE 4

iv TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES.............................................................................................................vi LIST OF FIGURES.........................................................................................................viii ABSTRACT....................................................................................................................... ..x CHAPTERS 1 INTRODUCTION........................................................................................................1 2 GENETIC EFFECTS OF ROOTING LO BLOLLY PINE STEM CUTTINGS FROM A PARTIAL DIALLEL MATING DESIGN...................................................6 Introduction...................................................................................................................6 Materials and Methods.................................................................................................8 Population..............................................................................................................8 Experimental Design.............................................................................................9 Statistical Analyses..............................................................................................10 Results and Discussion...............................................................................................16 Average Rooting..................................................................................................16 Observed Variance Components.........................................................................18 Causal Variance Components..............................................................................21 Heritability Estimates..........................................................................................22 Type B Genetic Correlations...............................................................................26 Selection for Rooting...........................................................................................27 Conclusion..................................................................................................................28 3 GENETIC ANALYSIS OF EARLY FIELD GROWTH OF LOBLOLLY PINE CLONES AND SEEDLINGS FROM THE SAME FULL-SIB FAMILIES.............30 Introduction.................................................................................................................30 Materials and Methods...............................................................................................33 Population............................................................................................................33 Propagation..........................................................................................................34 Field Design.........................................................................................................35 Statistical Analyses..............................................................................................36

PAGE 5

v Results and Discussion...............................................................................................43 Overall Growth of Clones and Seedlings............................................................43 Genetic Components of Variance........................................................................44 Heritability Estimates..........................................................................................49 Type B Genetic Correlations Between Propagule Types....................................50 Genotype x Environment Interaction..................................................................52 Conclusion..................................................................................................................54 4 GENETIC GAIN FROM SELECTION FO R ROOTING ABILITY AND EARLY GROWTH IN VEGETATIVELY PROP AGATED CLONES OF LOBLOLLY PINE...........................................................................................................................55 Introduction.................................................................................................................55 Materials and Methods...............................................................................................57 Population............................................................................................................57 Rooting and Field Trials......................................................................................57 Statistical Analyses..............................................................................................58 Causal Components of Variance.........................................................................60 Heritability Estimates..........................................................................................61 Type B Genetic Correlations...............................................................................62 Genetic Gain........................................................................................................63 Results and Discussion...............................................................................................65 Causal Components of Variance.........................................................................65 Type B Genetic Correlations...............................................................................66 Genetic Gain........................................................................................................68 Conclusion..................................................................................................................75 5 CONCLUSION...........................................................................................................78 APPENDICES A LOBLOLLY PINE PARTIAL DIALLEL MATING DESIGN. THIRTY-TWO PARENTS WERE CROSSED TO GENE RATE 70 FULL-SIB FAMILIES............81 B VARIANCE COMPONENT ESTIMATES FOR ROOTING ABILITY..................83 C VARIANCE COMPONENT ESTIMATES FOR EARLY GROWTH TRAITS OF LOBLOLLY PINE CLONES AND SEEDLINGS FROM THE SAME FULLSIB FAMILIES...........................................................................................................86 D VARIANCE COMPONENT ESTIMATES FROM THE BIVARIATE ANALYSES OF ROOTING AND 2ND YEAR HEIGHT..........................................88 LIST OF REFERENCES...................................................................................................94 BIOGRAPHICAL SKETCH...........................................................................................101

PAGE 6

vi LIST OF TABLES Table page 2-1. Experimental design for 5 rooting trials of loblolly pine stem cuttings. All trials were established in randomized complete block designs with 4 to 6 blocks and 4 to 9 ramets per clone in a row plot within each block................................................9 2-2. Summary of rooting from 5 loblolly pine trials set over two years and three seasons......................................................................................................................17 2-3. Genetic parameter estimates (standard e rror) for rooting of loblolly pine stem cuttings across 5 trials..............................................................................................23 2-4. Type B additive and dominance va riance correlations among pairs of rooting trials for loblolly pine stem cuttings (above and below diagonal, respectively). Standard errors are given in parentheses..................................................................26 3-1. Location of six field trials, establishmen t date and total number of test trees for each test....................................................................................................................34 3-2. Total number of clones, full-sib families, half-sib families, average number of clones per full-sib and half-sib family, and average number of seedlings per fullsib and half-sib family establis hed at the six field trials..........................................36 3-3. Mean 1st year height, 2nd year height, height increment, and crown width by propagule type for each of the six field trials. Although means are expressed in meters, analyses were conducted using measured traits in centimeters...................44 3-4. Genetic parameter estimates for 1st year height, 2nd year height, height increment, and crown width by propagule type across all six trials. St andard errors are given in parentheses.................................................................................................46 3-5. Genetic correlations between propagule types for 1st year height, 2nd year height, height increment, and crow n width at the parental (propGCABr ) and full-sib family (propFSBr ) levels...........................................................................................................51 4-1. Means, variance component estimates, heritabilities, and genetic correlations from the bivariate analysis of rooting ability and 2nd year height. Standard errors are given in parentheses...........................................................................................67

PAGE 7

vii 4-2. The total genetic correlation (TGBr ) between rooting ability and 2nd year height from analyses of a single rooting trial and field trial. Shaded values indicate the rooting trial in which cuttings originated from for their respective field trials. Standard errors are given in parentheses..................................................................70 4-3. The number of full-sib families (half-sib families) and average number of clones per full-sib family (half-sib family) sele cted from selecting 10% or 1% of the top clones using the combined selection index........................................................77 B-1. Observed variance component estimat es for rooting of loblolly pine stem cuttings from single-trial analyses............................................................................83 B-2. Observed variance component estimates for rooting of loblolly pine stem cuttings from pair-wise-trial analyses......................................................................84 B-3. Observed variance components for roo ting ability from the across-trial analysis using all five rooting trials.......................................................................................85 C-1. Observed variance components for l oblolly pine clones from the across-trial analyses of 1st year height, 2nd year height, height incr ement, and crown width. A separate error variance was modeled for each trial..............................................86 C-2. Observed variance components for lobl olly pine seedlings from the across-trial analyses of 1st year height, 2nd year height, height incr ement, and crown width. A separate error variance was modeled for each trial..............................................87 D-1. Observed variance component estimates from the bivariate analyses of rooting ability from Spring01 and 2nd year height from each of the field trials...................88 D-2. Observed variance component estimat es from the bivariate analyses of rooting ability from Summer01 and 2nd year height from each of the field trials................89 D-3. Observed variance component estimates from the bivariate analyses of rooting ability from Winter02 and 2nd year height from each of the field trials...................90 D-4. Observed variance component estimates from the bivariate analyses of rooting ability from Spring02 and 2nd year height from each of the field trials...................91 D-5. Observed variance component estimates from the bivariate analyses of rooting ability from Summer02 and 2nd year height from each of the field trials................92 D-6. Observed variance component estimates from the bivariate analysis of rooting ability using all five rooting trials and 2nd year height using all six field trials.......93

PAGE 8

viii LIST OF FIGURES Figure page 2-1. The proportion of the additive (2 1 0h), dominance (2 1 0d), and epistasis (2 1 0i) genetic variances on the observed binary sc ale for rooting of loblolly pine stem cuttings from each of the five separate tr ials (biased heritabilities) and from the combined analysis of all five trials. Standard error bars for broad-sense heritability estimat es are included............................................................................19 2-2. Narrow-sense (2Nh) and broad-sense (2NH) heritability estimates for rooting of loblolly pine stem cuttings transformed to the underlying normal scale using the threshold model of Equation 2-11. Standard error bars are included......................25 2-3. Full-sib family mean ( 2FSH) and clonal mean ( 2CLH) heritability estimates for rooting success of 2,200 clones from 70 full-sib families of loblolly pine. Standard error bars are included...............................................................................25 3-1. The proportion of additive (add ) and nonadditive (na) genetic variance components for clones and seedlings across the six field trials, where 2h = add and 2 H = add + na: a. 1st year height, b. 2nd year height, c. Height increment, and d. Crown width. Standard error bars for broad-sense heritability estimates are included..............................................................................................................48 3-2. Rank-rank plots show ing type B genetic correla tions between clones and seedlings from Test B based on: a. Pare ntal BLUP values, b. Full-sib family BLUP values, where full-sib BLUP values are equal to the sum of the predicted general combining ability for each of the two parents plus the predicted specific combining ability of the cross..................................................................................53 4-1. The genetic gain in rooting ability (%) over the population mean for deployment of the best half-sib family, full-sib family, best clone from the best ten full-sib families, and the single best clone when selecting for rooting ability or 2nd year height........................................................................................................................71 4-2. The genetic gain in 2nd year height (%) over the population mean for deployment of the best half-sib family, full-sib family, best clone from the best ten full-sib families, and the single best clone when selecting for rooting ability or 2nd year height........................................................................................................................72

PAGE 9

ix 4-3. Responses to selection in rooting ability and 2nd year height with various selection indices for two cl onal deployment options: 10% of clones selected and 1% of clones selected...............................................................................................75

PAGE 10

x Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy GENETIC EFFECTS OF ROOTING ABILITY AND EARLY GROWTH TRAITS IN LOBLOLLY PINE CLONES By Brian Stephen Baltunis December 2005 Chair: Timothy L. White Major Department: Forest Resources and Conservation Loblolly pine is the most important commercial tree species in the southern United States with over 1.1 billion seedlings plan ted annually. With elite genotypes becoming available, several forest industry companie s in the southeastern United States are developing rooted cutting and somatic embryogenesis programs aiming towards deployment of tested clones or families. However, before clones can be deployed, sufficient data need to be co llected on the population in orde r to have reliable information about the clones for deployment decisions. This dissertation reports on the genetic eff ects of rooting abilit y and early growth traits in nearly 2,200 clones of loblolly pine from 70-full-sib families. More than 239,000 stem cuttings were set in five rooti ng trials over two years. Overall rooting success across the five trials was 43%, and si gnificant seasonal effects were observed. Heritability of rooting ability was estimated both on the observed bina ry scale and on the transformed underlying normal scale.

PAGE 11

xi Rooted cuttings from these trials alon g with seedlings from the same full-sib families were established at several sites, and early growth traits through age two were compared between propagule types. All grow th traits demonstrated genetic variation, and parental and full-sib family rankings were similar for both propagule types. However, estimates of dominance genetic va riance in the seedling population appear to be inflated at the expense of additive effect s due to a lack of randomization of seedlings prior to field establishment. Little ge notype x environment interaction was observed across sites for all traits. A successful clonal forestry program for l oblolly pine based on rooted cutting technology needs to consider selection for bot h rooting ability and subsequent growth. There was a positive genetic correlat ion between rooting ability and 2nd year height at the parental, full-sib family, and clonal levels indi cating that selection for one trait will also lead to improvement of the other. Th e genetic gains in rooting ability and 2nd year height associated with several selection and deploymen t strategies are discussed. Moderate to high family and clonal mean he ritabilities, moderate to high type B correlations, and substantial among-family and among-clone gene tic variation indicate the potential for increasing rooting efficien cy and improving growth.

PAGE 12

1 CHAPTER 1 INTRODUCTION Loblolly pine (Pinus taeda L.) is the most important comm ercial tree species in the United States with over one billion seedlings planted annually (McKeand et al. 2003). Most commercially important tree species rema in relatively undomesticated, and loblolly pine is no exception. Genetic improvement of loblolly pine has been occurring since the 1950s in several tree improvement program s. There are three cooperative tree improvement programs in the southern United States that focus on improvement of southern pine species including loblolly pine : the Cooperative Forest Genetics Research Program (CFGRP), North Carolina St ate University-Industry Cooperative Tree Improvement Program (NCSUITIP), and the Western Gulf Forest Tree Improvement Program (WGFTIP). Loblolly pine tree improvement programs in the South are beginning their 3rd generation of breeding with gain s in volume per unit area up to 30% over unimproved loblolly pine (McKeand et al. 2003). Long-term tree improvement programs ai m to increase the population mean breeding value of a few key traits such as stem volume, disease resistance, and wood properties through breeding and se lection of superior genot ypes. Loblolly pine tree improvement programs are based on recurrent selection for general combining ability, which captures only the additive portion of the genetic variance. However, the nonadditive portion of genetic variation, dominance and epistasis, may be important components of variation for traits.

PAGE 13

2 Both additive and nonadditive genetic vari ation can be captured by deploying fullsib families or clones. However, deployment decisions should be based on reliable information. Field trials established with fu ll-sib seedlings allow the genetic variation to be partitioned into additive and nonadditive components. These trials not only provide ranks of parents or individuals for selection, but also of full-sib families in order to provide information for ma king deployment decisions. In any given generation of breeding, maxi mum genetic gains can be achieved in the deployment population by capturing all of th e genetic variation through operational propagation and deployment of selected clones. However, clonal forestry is not a breeding method to develop better genotypes. Clonal forestry is a method to massproduce well-tested genotypes. Short-term genetic gains may be maximized through deployment of well-tested clones, but longterm gains need to involve both clonal selection and recurrent selecti on for additive genetic variat ion through repeated selection and breeding. Clonal tests derived from full-sib familie s do provide an opportunity to estimate additive and nonadditive components of variance associated with a part icular trait or set of traits. However, tests shoul d be designed with a sufficient genetic structure in order to precisely quantify the genetic variation. For example, Frampton and Huber (1995) reported that they had low power in partitioni ng the genetic variation because of the lack of a mating design among the parents of the fu ll-sib crosses in a loblolly pine clonal study. In another study comparing clones from 30 full-sib families derived from two disconnected 4x4 factorials, Paul et al. (1997) concluded that futu re clonal studies should include more parents in the mating design. Although Isik et al. (2003) estimated the

PAGE 14

3 genetic variances from both clones and seedlings from the same nine full-sib families of loblolly pine, they identified a weakness of their study in that there were a limited number of parent trees used in the mating design. Finally, Frampton and Foster (1993) warned that interpretation of the results may be difficult for studies that only include seedlings and cuttings from a common checklot to be compared to the clonal propagules from select parents and families. In this case, any differences in the field performance because of propagule type may be confounded with the differences in genetic improvement (Frampton and Foster 1993). Clonally replicated progeny trials have been suggested as part of a tree improvement strategy for radiata pine (Jay awickrama and Carson 2000) and for loblolly pine (Foster and Shaw 1987; Isik et al. 2004; Byram et al. 2004) for a number of reasons. First, field trials established with clon ally replicated progeny allow for further partitioning of the genetic variation into th e additive, dominance, and epistatic genetic variation (Foster and Shaw 1988). Second, clonally propagated seedlings can provide genetic information more efficiently and w ith greater precision than zygotic seedling progeny (Burdon and Shelbourne 1974; Isik et al. 2004). Finally, clonal testing and selection strategies can provide greater ga in than seedling options (Shaw and Hood 1985; Mullin and Park 1994; Isik et al. 2004). Based on current technologies, several forest industries in the southeastern United States are pursuing clonal forest ry programs with loblolly pine (Weber and Stelzer 2002). In the initial stages of these clonal forestry programs, forest managers needed assurance that the clonal propagules gr owth corresponded to that of se edlings. Therefore, most of the earlier studies were designed to test wh ether cuttings grew similarly to seedlings.

PAGE 15

4 Based on those results, it is generally accepted that cuttings rooted from juvenile stock plants grow and perform comparably to seedlings. For example, Foster et al. (1987) reported that loblolly pine r ooted cuttings should perform co mparably to seedlings when the cuttings come from vigorous juven ile stock plants. In addition, McRae et al. (1993) concluded that for loblolly pine there were no significant differences between seedlings and rooted cutting propagules from common ch ecklots through five years of growth. Similar results were obtained by Frampton et al. (2000) where they reported no significant differences between the means of rooted cuttings and seedlings for height, diameter at breast height, and volum e through six years in the field. Trials established with clones and seedlings from the same families provide an opportunity for comparing both half-sib a nd full-sib family performances for both propagules. Genetic correlati ons between propagule types ca n provide further assurance that selections made through traditional tree improvement activities for recurrent selection for general combining ability can also be used successfully in breeding families to test in a clonal forestry program. Alt hough a number of studies have been reported comparing rooted cutting and seedlings, very few have been designed to estimate the genetic correlation for a tra it between propagule types. Two main criteria need to be met prior to operational deployment of loblolly pine clones. First, loblolly pi ne clones must perform well, e.g., meet the selection criteria for the desired traits. This involves the accumu lation of reliable data for the clones from greenhouse screening, field trials etc. Second, the selected clones have to be propagated in large enough numbers for deployment. For a rooted cutting based clonal program, this involves bulking up the number of hedges (ram ets) of a particular clone or group of

PAGE 16

5 selected clones through serial propagati on and then producing reforestation stock efficiently from the bulked-up clones. Only those clones that can be propagated easily and in sufficient numbers will be economically feasible for deployment. A clone that grows well in the field but roots poorly may not be economically feasible to include in a clonal program based on r ooted cutting technology. This dissertation is unique in that a comp lex genetic structure was utilized in order to increase the power in quantif ying the genetic variation associated with several traits in loblolly pine clones and seedlings. In Chapte r 2 rooting ability was assessed for nearly 2,200 clones of loblolly pine from 70 fullsib families derived from a partial diallel mating design in order to estimate genetic parame ters associated with rooting. More than 1,200 of these clones along with zygotic seed lings from the same full-sib families were established together on multiple sites across the southeastern United States. Genetic parameter estimates are compared between pr opagule types for early growth traits in Chapter 3. Finally, selection for both rooti ng ability and field growth is addressed in Chapter 4.

PAGE 17

6 CHAPTER 2 GENETIC EFFECTS OF ROOTING LOBLO LLY PINE STEM CUTTINGS FROM A PARTIAL DIALLEL MATING DESIGN Introduction Loblolly pine is the most important commercial tree species in the southern United States with over 1.1 billion s eedlings planted annually (McKeand et al. 2003). There are three cooperative tree improvement programs in the southern United States that focus on improvement of southern pine species includ ing loblolly pine: the Cooperative Forest Genetics Research Program (CFGRP), No rth Carolina State University-Industry Cooperative Tree Improvement Program (NCSUI TIP), and the Western Gulf Forest Tree Improvement Program (WGFTIP). Loblolly pine tree improvement programs in the South are beginning their 3rd generation of breeding with gains in volume per unit area up to 30% over unimproved loblolly pine (McKeand et al. 2003). Long-term tree improvement programs ai m to increase the population mean breeding value of a few key traits through breedi ng and selection of superior genotypes. These programs are based on recurrent select ion for general combining ability which captures only the additive portion of the genetic variance. In any given generation of breeding, maximum gains can be achieved in the deployment popul ation by capturing all of the genetic variation (additive and nonadditive components) through operational propagation of selected clones. With elite genotypes becoming available, several forest industries in the southeastern United States are developing rooted cutting programs for

PAGE 18

7 loblolly pine aiming towards deployment of tested clones or families (Weber and Stelzer 2002). Clonal tests derived from full-sib fam ilies provide an opportunity to estimate additive and nonadditive genetic components of variance associated with a particular trait or set of traits (Isik et al. 2003; Isik et al. 2004). In clonal field tr ials traits of interest may include height, volume, wood quality, and disease resistance. However, in order to establish clonal field trials the clones must fi rst be propagated. Th erefore, clonal rooting trials are important for estimating genetic va riance components associated with rooting. Previous rooting studies of loblolly pine have been relatively small in size ranging from several hundred (Goldfarb et al. 1998; Foster et al. 2000) to several thousand cuttings (Foster 1990; Anderson et al. 1999). Many studies contai ned a small number of families from factorial mating designs (Goldfarb et al. 1998; Anderson et al. 1999; Cooney and Goldfarb 1999; Frampton et al. 1999) and few have been designed to estimate genetic parameters associated with r ooting in loblolly pine (Foster 1978; Foster 1990; Anderson et al. 1999). The current study is unique in that a large number of cuttings were set in each trial (> 34,000), and rooting was assessed on nearly 2,200 clones of loblolly pine from 70 full-sib families derived from a partial diallel mating design in order to estimate genetic parameters associated with rooting. The objectives of the study were to (i) evaluate the rooting ability of stem cu ttings from nearly 2,200 loblolly pine clones, (ii) determine the causal components of va riance in rooting of stem cuttings, (iii) assess heritability estimates for rooting from fi ve trials both on the observed binary scale and the underlying normal scale, and (iv) determine the Type B genetic correlations for

PAGE 19

8 both additive and dominance gene tic effects to measure th e correspondence in rooting performance across five setting dates. Materials and Methods Population The parental population for this study was selected from the Loblolly Pine Lower Gulf Elite Population (LPLGEP) which consists of selections from a ll three southern pine tree improvement cooperatives: CFGRP, NCSUITIP, and WGFTIP. Twenty 1st generation and ten 2nd generation selections representing the Atlantic Coastal Plain, Florida, and Lower Gulf provenances were se lected from this populat ion. Two additional slow-growing parents were included to pr ovide linkage with another study. These parents were crossed in a circular dial lel mating design (Appendix A) with some additional off-diagonal crosses, resulting in a total of 70 full-sib families. On average each parent was involved in a pproximately four crosses. Seeds from the 70 families were sown in March 2000 into Ray Leach SuperCells (Stuewe and Sons, Corvallis, OR). The seedlings were grown in a greenhouse at International Paper Companys (IPC) facility in Jay, FL, and after th ree months of growth the seedlings were pruned back to a height of about 10-12 cm. Approximately 32 seedling hedges (ortets) per fu ll-sib family were transplanted into 3-gallon containers and given unique clonal identifica tions in September 2000. The hedges were repeatedly sheared in order to minimize the effects of maturation and increase the number of shoots available for the rooting trials. The ortets were randomized in a containerized hedge orchard in order to reduce spuri ous C effects at the family level. However, C effects at the clonal level could not be accounted for, b ecause all cuttings originated from a single seedling ortet.

PAGE 20

9 Table 2-1. Experimental design for 5 rooting trials of loblolly pine stem cuttings. All trials were established in randomized complete block designs with 4 to 6 blocks and 4 to 9 ramets per clone in a row plot within each block. Trial Date set No. of families No. of clones No. of plots per clone No. of cuttings per plot Total cuttings Spring01 May 711, 2001 70 2194 4 4 34,707 Summer01 July 2-6, 2001 70 2157 5 4 43,048 Winter02 Jan 1418, 2002 61 1648 6 5 49,315 Spring02 Apr 29May 3, 2002 61 1254 6 9 67,059 Summer02 June 2428, 2002 61 947 6 9 45,108 Experimental Design Two rooting trials were conducted in 2001 and three were conducted in 2002. Stem cuttings between 3 and 8 cm in length were harvested from the s eedling ortets in May 2001, July 2001, January 2002, April/May 2002, and June 2002, for trials Spring01, Summer01, Winter02, Spring02, and Summer 02, respectively. Cutting size was relatively consistent within any trial, and th e cuttings set in Winter02 were the smallest. The experimental design differe d among the trials due to numb er of families, clones and available cuttings from each ortet (Table 21). The reduction in the number of families and clones between the first and last rooting tr ials was a result of a number of factors. First, random hedge mortality was a major fact or: disease, repeated severe pruning, and uneven watering (inadequate) all contributed to the random loss of hedges. Since cuttings of a clone originated from a singl e ortet, by default mortality resulted in a truncated population for future rooting tria ls. Second, not all hedges produced an adequate number of cuttings at every harvest. The primary objective of the three settings

PAGE 21

10 in 2002 was to produce propagules for field tr ials. Therefore, cl ones that were not producing enough shoots to ultimately be pl anted across six fiel d sites were culled regardless of rooting frequency. This resulted in the reduction in the number of families in the last three rooting trials because of t oo few surviving clones in some of the families to meet the goal for field designs. We were striving for a balanced field design with 15 clones from each of 61 full-sib families. Cuttings were randomly set in 4-, 5-, or 9cutting clonal row plots (Table 2-1) into pre-formed plugs consisting of peat moss, pe rlite, and a binding resin. Each plug was approximately 13 mL in volume and was held by the V-13 HIKO tray (135 cells; Stuewe and Sons, Corvallis, OR). Cuttings were eith er treated prior to setting with a 1.0% indole-3-butyric acid and 0.5% napthalene-1-a cetic acid (NAA) basal dip or after setting using a foliar NAA application according to IP C protocols. Trays contained 15 to 30 clones depending on the trial and were randomly placed in an environmentally controlled greenhouse. There were 4 to 6 complete re plications depending on trial (Table 2-1). Rooting assessments were made 9-weeks after setting for both the Spring01 and Summer01 trials. Cuttings were measured fo r presence (1) or absence (0) of roots. Cuttings with a root 1 mm were considered rooted (Goldfarb et al. 1998, Foster et al. 2000). For trials Winter02, Spring02, a nd Summer02 assessments were at 11-weeks following setting. Cuttings in these trials we re also scored for presence or absence of roots. However, only cuttings th at had at least one visible r oot on the exterior of the plug regardless of length were considered rooted. Statistical Analyses For binomial traits, such as rooting, the unit of analysis can be the individual observations (Huber et al. 1994; Dieters et al. 1996) or plot means combined with a

PAGE 22

11 transformation such as arcsin or lo gistic (Sohn and Goddard 1979; De Souza et al. 1991). We chose to analyze the observed 0,1 data for several reasons. First, REML estimation of variance components has been shown to be robust to violations of the underlying normality assumptions (Banks et al. 1985; Westfall 1987) suggesti ng that analyses using individual observations of bi nary data yields satisfactory results. Second, simulation studies have shown that the us e of individual observations is superior to the use of plot means in REML, and that these variance component estimates perform well across mating designs and imbalanced data (Huber et al. 1994). Huber et al. (1994) showed that a lower variance among estimates was obtai ned using individual observations as compared to plot means and that this adva ntage increased with increasing imbalance. Third, when heritability is low and the incide nce close to 50%, there is little difference between heritability estimates on the binary and transformed scale (Dempster and Lerner 1950). In fact these two estimates are equiva lent when the inciden ce is exactly 50% for low heritability traits. Lopes et al. (2000) demonstrated that the Dempster and Lerner (1950) threshold model closely estimates the true underlying heritability at incidences between 25% and 75% for traits with low heritability (3 02 h ). Finally, Lopes et al. (2000) also demonstrated (for tr aits with low herita bility) that heritab ility estimates from the observed binary data without transformation of data result in predicted gain close to the realized gain, while tran sformations can suffer from issues of back transformation when one wishes to predict gains on the original scale. All variance components for rooting ability of loblolly pine stem cuttings were estimated using the individual binary obser vations using REML estimation for each of the 5-rooting trials using GAREML (Huber 1993). However, upwardly biased estimates

PAGE 23

12 of genetic variances result when variance components are estimated from single-site (trial) analyses since the estimated genetic variance also contains the genotype x environment interaction (Comstock and Moll 1963). Therefore, across-trial analyses were performed to separate the genotype x e nvironment interactions in order to remove this bias. [2-1] ijklmno lm i j lm in ilm im il lm n lm m l i j k i j i ijklmnoerror fam r fam c t sca t gca t gca t fam c sca gca gca tray R T y ) ( ) ( ) ( )) ( ( ) (* ) ( * * ) ( where yijklmno is the rooting respons e (0 or 1) of the oth ramet of the nth clone within the lmth full-sib family in the kth tray within the jth rep of the ith trial is the population mean Ti is the fixed effect of trial Rj(i) is the fixed effect of rep trayk(j(i)) is the random variable tray (incomplete block) ~ IID(0,2TRAY) m and lgca is the random variable female (l) and male (m) general combining ability (gca) ~ IID(0,2GCA) scalm is the random variable sp ecific combining ability (sca) ~ IID(0,2SCA) c(fam)n(lm) is the random variable clone within family ~ IID(0,2CLONE) t*gcail and im is the random variable test by female gca and test by male gca interaction ~ IID(0,2TESTxGCA) t*scailm is the random variable test by full-sib family interaction ~ IID(0,2TESTxFAM) t*c(fam)in(lm) is the random variable test by clone interaction ~ IID(0,2TESTxCLONE) r*famj(i)lm is the random variable rep by family interaction ~ IID(0,2REPxFAM)

PAGE 24

13 errorijklmno is the random error which include s among plot and within plot ~ IID(0,2ERROR). The single trial model is identical except all model factors with subscript i are removed (sources involving test). Genetic parameters were estimated and st andard errors were calculated according to Foster and Shaw (1988) us ing the appropriate variance co mponents from the individual or across trial model. Estimates of additive and dominance genetic variance are upwardly biased because they are confounded with fr actions of epistasi s (Cockerham 1954). Epistastic genetic variance is also only a pproximated because it co ntains only a fraction of the total epistasis plus any C effects, if they exist. [2-2] ... 4 16 1 4 1 2 AAA AA A GCA AV V V V = estimate of additive genetic variance [2-3] ... 4 4 1 2 1 2 1 2 DD AD AA D SCA DV V V V V = estimate of dominance genetic variance [2-4] ... 3 2 4 3 2 1 4 1 2 2 2 DD AD AA SCA GCA CLONE IV V V V = estimate of epistatic genetic variance [2-5] 2 2 2 2 CLONE SCA GCA GV = estimate of total genetic variance [2-6] 2 2 2 2 2 2 2 2 2 2 ERROR REPxFAM TESTxCLONE TESTxFAM TESTxGCA CLONE SCA GCA PV = phenotypic variance for across tr ial model (the phenotypic va riance from the individual trial model is the same but drop2 2TESTxGCA, 2TESTxFAM, and 2TESTxCLONE). Biased and unbiased heritab ility estimates for rooting based on observed 0,1 data were derived using the estimated variance components from the single and across trial

PAGE 25

14 models, respectively. In addi tion the proportion of dominance (2 d ) and epistasis (2 i) were estimated. Standard errors of these es timates were calculated using a Taylor series expansion (Kendall and Stua rt 1963; Namkoong 1979; Huber et al. 1992; Dieters 1994). [2-7] 2 2 2 2 2 2 2 2 2 2 1 0 2 2 4 ERROR REPxFAM TESTxCLONE TESTxFAM TESTxGCA CLONE SCA GCA GCAh = across-trial narrow-sense hertibility based on observed binary data [2-8] 2 2 2 2 2 2 2 2 2 2 2 2 1 0 2 2 2 ERROR REPxFAM TESTxCLONE TESTxFAM TESTxGCA CLONE SCA GCA CLONE SCA GCAH = across-trial broad sense heritability based on observed binary data [2-9] 2 2 2 2 2 2 2 2 2 2 1 0 2 2 4 ERROR REPxFAM TESTxCLONE TESTxFAM TESTxGCA CLONE SCA GCA SCAd = across-trial dominance proportion [2-10] 2 2 2 2 2 2 2 2 2 2 2 2 1 0 2 2 3 2 ERROR REPxFAM TESTxCLONE TESTxFAM TESTxGCA CLONE SCA GCA SCA GCA CLONEi = across-trial epistatic proportion. The main problem with calculating heritabi lity on the observed 0,1 data is that the relationship between 2h on the observed scale and 2h on the underlying normal scale depends on the mean incidence ( e.g. % survival, % infected individuals, rooting percentage), and therefore th e conversion results in a bias ed estimate (Van Vleck 1972). However, this is not a problem for low heritability traits with intermediate incidences

PAGE 26

15 (Lopes et al. 2000). In order to make valid comparis ons to heritability estimates in other rooting trials that have diffe rent mean rooting percentages 2 1 0 h needs to be transformed to 2h on the underlying normal scale. Therefore, narrowand broad-sense heritability estimates on the observed 0,1-scale were tr ansformed using a threshold model to an underlying normal scale (Dem pster and Lerner 1950). [2-11] 2 2 1 0 2) 1 )( ( z p p h hN where 2Nh is the heritability on the underlying normal scale p is the rooting percent zis the ordinate of the normal density f unction which corresponds to probability p. Full-sib family mean heritability and cl onal mean heritability for rooting were estimated for both the singleand across-tria l analyses. Standard errors for these estimates were calculated using Dicker sons Method which assumes the phenotypic variance (denominator) is a know n constant (Dickerson 1969). [2-12] ctrn tr tc t t c HERROR REPxFAM TESTxCLONE TESTxFAM TESTxGCA CLONE SCA GCA SCA GCA FS 2 2 2 2 2 2 2 2 2 2 2 2 2 2 = across-trial family mean heritability, wh ere, c = harmonic mean number of clones per family, t = number of trials, r = harmonic mean number of reps per test, and n = harmonic mean number of ramets per clone per plot.

PAGE 27

16 [2-13] trn tr t t t HERROR REPxFAM TESTxCLONE TESTxFAM TESTxGCA CLONE SCA GCA CLONE SCA GCA CL 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 = across-trial clonal mean heritability. Type B genetic correlations for rooting acr oss all 5 trials were estimated for both additive and nonadditive components (Yamada 1962; Burdon 1977). Standard errors of Type B correlations were calculated usi ng the Taylor Series Expansion method. [2-14] 2 2 2 ge g g Br where Br is the estimate of the Type B genetic correlation, 2g is the genetic variance component (either additive or dominance), and 2ge is the G x E interaction (additive and dominance). Results and Discussion Average Rooting A total of over 239,000 stem cuttings from nearly 2,200 clones of loblolly pine were set in five rooting trials. Overall rooting across the five trials was 43% and is comparable to other rooting studies involving loblolly pine. Goldfarb et al. (1998) reported 44% rooting of loblolly pine stem cuttings from 400 seedling hedges from one open-pollinated family. Over four rooting trials, Anderson et al. (1999) reported 33% rooting from 90 clones of loblolly pine from 9 full-sib families. Foster (1990) reported overall rooting from three settings of 42% fo r 546 clones of loblolly pine derived from 54 full-sib families. However, studies with fewer families of loblolly pine have yielded

PAGE 28

17 substantially higher rooti ng percentages (Cooney and Goldfarb 1999; Murthy and Goldfarb 2001; LeBude et al. 2004). Rooting of loblolly pine cuttings was vari able over the five trials (Table 2-2). Spring cuttings rooted at greater than 50% while summer cuttings in the two summer trials averaged 38% and 24% respectively. Winter cuttings were intermediate at 45% rooting. Broad inference linear contrasts we re constructed using the estimates of the fixed effects in order to test seasonal rooti ng responses. Cuttings set in the two spring trials rooted at a significantly greater frequency then cuttings set in the summer trials ( p < 0.0001). This implies that we would always expect a greater rooti ng percent in spring settings than in summer setti ngs under this propagation system. Table 2-2. Summary of rooti ng from 5 loblolly pine trials set over two years and three seasons. Trial Rooting % Half-sib family range Full-sib family range Clone range Spring01 54% 36-70% 28-77% 0-100% Summer01 38% 24-54% 18-69% 0-100% Winter02 45% 17-60% 17-67% 0-100% Spring02 51% 36-67% 28-75% 0-98% Summer02 24% 9-43% 8-53% 0-89% Seasonal rooting responses for loblolly pine stem cuttings have been observed in other studies. Early rooting trials of loblolly pine cuttings report ed best rooting from cuttings set from September through Janua ry (Cech 1958; Reines and Bamping 1960; Grigsby 1962; Marino 1982). These early experiments concluded that increased temperatures in the greenhouse in spring and su mmer trials decreased rooting. In fact,

PAGE 29

18 Cech (1958) reported a 3-fold increase in roo ting under cool conditions rather than under warm conditions. However, Foster et al. (2000) observed an overall rooting of 50% for a rooting trial established in March, while on ly 20% rooting for a trial established in September. They hypothesized that the re duction in September rooting was due to a decrease in metabolic activity due to the decrease in photoperiod. Rowe et al. (2002a; 2002b) reported trends in rooting similar to those of the current study. They observed 59% rooting for spring cuttings versus 40% rooting for winter cuttings and 35% rooting for summer cuttings. Cooney and Goldfarb (1999) also reported hi gh rooting percentages for spring cuttings (62% and 83% in two su ccessive years). In contrast, Murthy and Goldfarb (2001) reported higher rooting percen tages for winter cuttings (85%) than for spring cuttings (60%). Winter cuttings often take longer to root but overall rooting may not be different than spring cuttings. Perh aps the slight reduction in rooting frequency for the Winter02 setting versus the two spring settings was a f unction of rate of rooting. The reduction in rooting seen here for su mmer settings may be a result of increased temperatures during the collec tion and propagation phases of th e experiment. The higher temperatures and humidity experienced during summer months may have resulted in an increased abundance or activity of pathogens and hence a higher rate of decay was observed in the two summer trials. The tim e delay between collec tion and setting of cuttings may also have contributed to this reduction in rooting. Murthy and Goldfarb (2001) reported a decline in rooting percentage with increasing drying time of cuttings. Observed Variance Components Variance components were estimated for all singleand acros s-trial analyses (Appendix B). Even with the reduction in th e number of clones and families throughout the study, there was no apparent reduction in the variance component estimates over time

PAGE 30

19 as evidenced by parameter estimates, e.g. additive genetic variance estimates were relatively constant over time (Figure 2-1) The variance associated with general combining ability was 2% of the total phenot ypic variation associ ated with rooting. Halfsib family rooting percentages ranged from a high of 36-70% for Spring01 to a low of 943% for Summer02 (Table 2-2). The proportion of the total variation in rooting that was accounted for by specific combining ability was only 0.3-1.1%. Full-sib family means for rooting for the two spring settings ra nged from 28-76%. Although overall rooting was greater for these two spring trials, the ne t difference in the range of family means was approximately the same (~45-51%). Similar ranges in family mean rooting percentages have been re ported (Foster 1990; Anderson et al. 1999). 0 0.05 0.1 0.15 0.2 0.25 Spring01Summer01Winter02Spring02Summer02AllTrial 2 1 0 h 2 1 0 d 2 1 0 i 2 1 0 H Figure 2-1. The proportion of the additive (2 1 0 h ), dominance (2 1 0 d ), and epistasis (2 1 0 i ) genetic variances on the observed binary scale for rooting of loblolly pine stem cuttings from each of the five separate trials (biased heritabilities) and from the combined analysis of all five trials. Standard error bars for broadsense heritability estimates are included.

PAGE 31

20 Clones within families accounted for appr oximately 10-17% of the total variation in rooting in the five trials (Appendix B). Foster (1990) reported that the clone within family source of variation in rooting of 546 clones of loblolly pine was 3.7 % of the total variation. However, in another study involvi ng cuttings of loblolly pine, the among clone source of variation accounted fo r nearly 22% of the total va riation (Foster 1978). In the current study, the variance associated with clone within family was 4.5-8.4 times greater than the gca variance and 13-56 times greater than the sca variance based on single-trial analyses. Just as there was a large range in root ing among families, there was also a large range in rooting among clones within families (Table 2-2). Rooting for clones within families ranged from 0-100% for the first th ree trials, and ranged from 0-98% and 0-89% in the last two trials, respectively. Anderson et al. (1999) observed ranges in rooting frequency for clones within family similar to those observed in this study. Foster (1990) reported significant variation for clones with in family with rooting percentages ranging from 6.7-85.0%. A large variation in the rooting environm ent has been reported in many rooting experiments of loblolly pine and other spec ies, with an error variance ranging from 4671% (Foster 1978; Foster 1990; Sorens en and Campbell 1980; Cunningham 1986). In the current study, the majority of the observed variance in rooting wa s attributable to the error variance. The error variance account ed for 76.3-83.7% of the total variance observed in the five trials. Each rooting tr ial was spread over an entire greenhouse due to the large size of the experiments. Appare ntly, the rooting environment was not uniform throughout the greenhouse. There are many fact ors that can contribute to a variable

PAGE 32

21 rooting environment. Differential temper ature gradients, unequal airflow, unequal misting of cuttings, edge effects, and diseas e incidence can all cont ribute to a variable rooting environment. Causal Variance Components The observed variance components were used to estimate additive, dominance and epistatic variances using equations 2-2, 2-3, and 2-4, respectively. Additive genetic variance was approximately 2 to 6 times larg er than the dominance genetic variation for rooting in the five trials (Figure 2-1). In Foster (1990), the additive genetic variance was also about 6 times larger than the domina nce genetic variance. However, Anderson et al. (1999) found three times and Foster (1978) found 2.2 times greater dominance genetic variance in rooting compared to the additive va riance in loblolly pine stem cuttings. The epistatic genetic variance was also estimated in the current study and was approximately 0.44 to 1.49 times as much as the additive genetic variance. The lowest amount of epistasis was observed for the winter sett ing and was the only setting that had a nonadditive to additive ratio less than one. C effects can lead to upwardly biased estimates of total genetic and nonadditive genetic components of variance when analyz ing clonal data (Libby and Jund 1962). If C effects are present, then total genetic va riation associated with clones will be overestimated (Libby and Jund 1962). Significant C effects are likely to occur in traits that are measured soon after propagation (such as rooting traits, early shoot growth, etc.), but apparently lessen for traits measured at later times (Libby and Jund 1962). Foster et al. (1984) used a secondary cloning approach to separate C effects from the genetic variance in rooting of western hemlock cu ttings. They found significant C effects associated with rooting and that these non-gene tic effects biased the genotypic values of

PAGE 33

22 clones. However, low or non-significant C e ffects for rooting of tamarack and balsam poplar cuttings have been reported (Farmer et al. 1989; Farmer et al. 1992). In the current study, estimates of epistatic gene tic variance components are confounded with C effects, because ramets of a clone came fr om the original seedling ortet. However, estimates of additive and dominance geneti c effects are not confounded with C effects, because the clones within families were randomized in the hedge orchard. Heritability Estimates Rooting of loblolly pine stem cuttings wa s weakly controlled by additive effects. Individual tree narrow-sense heritability using the observed variance components (2 1 0 h ) ranged from 0.075 to 0.089 in the five separate trials (Figure 2-1). These estimates are in agreement with Foster (1978) who reported 2 h as 0.07 for rooting percentage of loblolly pine. A slightly higher 2 h was reported by Foster (1990). However, Anderson et al. (1999) reported 2 h of 0.26 for rooting percentage in loblolly pine stem cuttings. However, all of these previous studies of l oblolly pine rooted cuttings analyzed data based on plot rooting percentage as opposed to using 0,1 data as in this study, and this leads to an increased estimate of heritability due to the reduction of the impact of the within plot portion of the error variance in a ny plot means or plot percentage analysis for rooting. When the data in the current st udy were analyzed based on rooting percentage, 2 h was 0.18 (from Spring01, data not shown), which falls in the middle of the range previously reported. The unbiased estimates of individual tree narrow-sense heritability (2 1 0 h ) using the 0,1 data from the pairwise test analyses ra nged from 0.045 to 0.074. When all of the data

PAGE 34

23 from the 5 trials were analyzed together, 2 1 0 h was 0.051 (Table 2-3; Figure 2-1). The proportion of dominance (2 1 0 d ) was estimated for each of the trials. The upwardly-biased estimates for this parameter ranged from 0.01 4 to 0.044 (Figure 2-1). When all of the data were analyzed together from the 5 trials, 2 d was 0.018 (Table 2-3; Figure 2-1). The epistatic proportion (2 1 0i) was also estimated and ranged from 0.032 to 0.095 (Table 2-3; Figure 2-1). Table 2-3. Genetic parameter estimates (standard error) for rooting of loblolly pine stem cuttings across 5 trials. Parameter Estimate Narrow-sense heritability on observed binary scale (2 1 0 h ) 0.051 (0.017) Broad-sense heritability on observed binary scale (2 1 0 H ) 0.101 (0.008) Narrow-sense heritability on the underlying normal scale (2Nh) 0.08 (0.027) Broad-sense heritability on the underlying normal scale (2NH) 0.16 (0.013) Narrow-sense family mean heritability ( 2FSH ) 0.833 (0.24) Broad-sense clonal mean heritability ( 2CLH ) 0.815 (0.074) Additive genetic variance (AV ) 0.0117 (0.004) Dominance genetic variance (DV ) 0.0042 (0.002) Epistatic genetic variance (IV ) 0.0074 (0.002) Phenotypic variance (PV ) 0.2297 (0.002) Type B additive genetic variance correlation 0.68 (0.23) Type B dominance genetic va riance correlation 0.61 (0.27) Type B total genetic variance correlation 0.53 (0.048) Broad-sense heritability (2 1 0 H ) ranged from approximately 0.15 to 0.22 (Figure 21). Broad-sense heritability was 0.101 when all of the data from the five trials were combined (Table 2-3; Figure 2-1). Foster (1 990) observed very little nonadditive genetic

PAGE 35

24 variance and reported 2 H of 0.13 for rooting percentage. In another study 2 H was reported as 0.23 for rooting percenta ge (Foster 1978). However, Anderson et al. (1999) reported a much higher2 H (0.63) for rooting percentage of loblolly pine. When the data in this study were analyzed based on rooting percentage, then 2 H was 0.47. Narrowand broad-sense heritability estim ates on the observed 0,1-scale were transformed to an underlying normal scale assuming a threshold model (Equation 2-11). Narrow-sense heritability based on the underlying normal scale (2Nh) ranged from 0.12 to 0.16 (Figure 2-2). Broad-sense heritabili ty based on the underlying normal scale (2NH) ranged from 0.24 to 0.36 (Figure 2-2). When all of the data from the five settings were analyzed together, then 2Nh was 0.08 and 2NH was 0.16 (Table 23; Figure 2-2). The transformation of 2 1 0 h and 2 1 0 H to the underlying normal scale allows direct comparisons of heritability estimates among rooting studie s when the rooting percentages are different. Family mean heritability ranged from 0.84 to 0.9, while clonal mean heritability ranged from 0.82 to 0.92 (Figure 2-3). When all of the data from the five trials were combined then 2FSH was 0.83 and 2CLH was 0.82 (Table 2-3; Figu re 2-3). Foster (1990) reported lower estimates of both family-mean heritability and clonal -mean heritability for a rooting study consisting of 540 clones of loblolly pine from 54 full-sib families, 0.46 and 0.40, respectively. In another loblolly pine rooting study with 27 full-sib families consisting of 10 clones each, family-mean heritability was reported as 0.31, while the clonal-mean heritability was 0.87 (Anderson et al. 1999).

PAGE 36

25 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4Spring01Summer01Winter02Spring02Summer02AllTrialNormal Scale 2Nh 2NH Figure 2-2. Narrow-sense (2Nh) and broad-sense (2NH) heritability estimates for rooting of loblolly pine stem cuttings transf ormed to the underlying normal scale using the threshold model of Equation 2-11 Standard error bars are included. 0 0.2 0.4 0.6 0.8 1 1.2Spring01Summer01Winter02Spring02Summer02AllTrialEstimate 2FSH 2CLH Figure 2-3. Full-sib family mean ( 2FSH ) and clonal mean ( 2CLH ) heritability estimates for rooting success of 2,200 clones from 70 full-sib families of loblolly pine. Standard error bars are included.

PAGE 37

26 Type B Genetic Correlations The type B correlations for additive effect s were moderately high and ranged from 0.53 to 0.91 (Table 2-4) indicating that parental rankings of the 32 parents were moderately to strongly correlated among pairs of trials. When all of the data were analyzed together from the five trials, the type B additive correlation was 0.68. The highest type B correlation was observed between the two spring settings which were one year apart, while the lowest correlations were observed between the winter setting (Winter02) and other trials. Spring and su mmer cuttings were, in general, actively growing, succulent material, while winter cuttings were generally smaller and more lignified. Perhaps, some of the genes contro lling rooting of dormant winter cuttings are different than those controlling spring and summer cuttings. Table 2-4. Type B additive and dominance variance correlations among pairs of rooting trials for loblolly pine stem cuttings (above and below diagonal, respectively). Standard errors are given in parentheses. Spring01 Summer01 Wi nter02 Spring02 Summer02 Spring01 0.701 (0.11) 0.59 (0.15) 0.911 (0.07) 0.851 (0.08) Summer01 0.823 (0.19) 0.634 (0.12) 0.673 (0.12) 0.634 (0.13) Winter02 0.192 (0.30) 0.629 (0.34) 0.656 (0.12) 0.525 (0.16) Spring02 0.493 (0.24) 0.533 (0.35) 0.724 (0.24) 0.907 (0.06) Summer02 0.777 (0.22) 0.973 (0.37) 0 0.703 (0.24) Estimated type B correlations for domina nce effects measure the correspondence of dominance across pairs of trials and were esti mated to be moderate to high (Table 2-4)

PAGE 38

27 with two exceptions. First, the type B do minance correlation between the Spring01 and Winter02 trials was 0.192. Also, no domi nance genetic variance was detected in the pairwise analysis of theWinter02 and Summer02 trials. However, when all of the data were analyzed together, the type B dominance correlation was 0.61. Selection for Rooting Before clones of loblolly pine can be deployed operationall y, two things must occur. First, the clones have to be fiel d-tested and selected for desirable traits, e.g. growth, disease resistance, and wood properties. Second, the selected clones have to be be propagated in large enough numbers for deployment. For a r ooted cutting-based program, this involves bulking up the number of hedges of a particul ar clone or group of clones through serial propagation and then producing reforestation st ock efficiently from the bulked-up clones. Only those clones that can be propaga ted easily will be economically feasible for deployment. Theref ore, selection of clones for rootability as well as field performance should be considered as part of a clonal forestry program based on rooted cutting technology (Foster et al 1985; Foster et al 2000). In the current study, selection of the top 10% of clones for r ooting (~220 clones) would result in a gain of about 24% in rooting, where trn tr t t t H i GainERROR REPxFAM TESTxCLONE TESTxSCA TESTxGCA CLONE SCA GCA CL2 2 2 2 2 2 2 2 2 2 2 (Foster 1990). Selecting the top 1% of clones for rooting w ould result in a gain of nearly 37% in rooting of loblolly pine stem cuttings in the current generation. Alternatively, genetic

PAGE 39

28 gain in rooting success in the next gene ration can be achieved through selection and breeding. trn tr t i ctrn tr tc t t c H i GainERROR REPxFAM TESTxCLONE CLONE GCA I ERROR REPxFAM TESTxCLONE TESTxSCA TESTxGCA CLONE SCA GCA FS F 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 By selecting the best rooting clone from the top 25 out of 70 families, gain in rooting success of 16.8% can be expected in the next generation by breeding these 25 selections. Conclusion Loblolly pine is the most important comm ercial tree species in the southeastern United States. Several forest industry comp anies are developing rooted cutting programs for loblolly pine in order to maximize geneti c gains through deployment of tested clones. With rooting data from 2,200 clones from 70 full-sib families, the current study gives better estimates of genetic components of variance for rooting th an several previous studies. These results show a great deal of genetic variation for rooting among families and clones of loblolly pine. Only those cl ones that can be propagated easily will be economically feasible for deployment. Therefor e, selection of clones for rooting ability, as well as field performance should be consid ered as part of a cl onal forestry program based on rooted cutting technology. Comb ined with moderate to high estimates of familyand clonal-mean heritabilities and type B correlations, these results indicate the

PAGE 40

29 potential for increasing rooting efficiency by selecting good rooting families and clones or culling poor rooters.

PAGE 41

30 CHAPTER 3 GENETIC ANALYSIS OF EARLY FIELD GR OWTH OF LOBLOLLY PINE CLONES AND SEEDLINGS FROM THE SAME FULL-SIB FAMILIES Introduction Loblolly pine ( Pinus taeda L.) is the most important comm ercial tree species in the United States with over one billio n seedlings planted annually (McKeand et al. 2003). Most commercially important tree species rema in relatively undomesticated, and loblolly pine tree improvement programs are only now beginning their 3rd generation of breeding and testing (McKeand and Bridgewater 1998). Tree improvement programs for loblolly pine have relied on recurrent selection for general combining ability for improvement of a few key traits. These programs have hist orically utilized seedling progeny trials in order to predict breeding values for these tr aits. Traditional tree improvement programs using open-pollinated seed orchard seedlings for deployment only capture additive genetic variation. However, nonadditive genetic variation may be an important component of variation for some traits, a nd additive and nonadditive genetic variation can be captured by deploying full-sib families or clones. More efficient field-testing has been impl emented in order to gain information on full-sib families. For example, the Cooperativ e Forest Genetics Research Program at the University of Florida is using a partial dialle l mating design to cross slash pine selections to generate full-sib seedlings for progeny trials (Gezan et al. 2004). Tests established with full-sib seedlings allow the genetic va riance to be partitioned into additive and dominance components (Falconer and Mackay 19 96). These trials not only provide ranks

PAGE 42

31 of parents or individuals for selection, but also of full-sib families in order to provide information for making deployment decisions. For a number of reasons, clonally replicated progeny trials have been suggested as part of a tree improvement strategy for ra diata pine (Jayawickrama and Carson 2000) and for loblolly pine (Foster and Shaw 1987; Isik et al. 2004; Byram et al. 2004). First, field trials established with clonally replicated progeny allow for further partitioning of the genetic variation into the additive, dominan ce, and epistatic genetic components (Foster and Shaw 1988). Second, clonally propagat ed seedlings can provide genetic information more efficiently and with greater precisi on than zygotic seedling progeny (Burdon and Shelbourne 1974; Isik et al. 2004). Finally, clonal testi ng and selection strategies can provide greater gain for op erational deployment than s eedling options (Shaw and Hood 1985; Mullin and Park 1994; Isik et al. 2004). Based on current technologies, several fore st industries in the southeastern United States are pursuing clonal forestry programs w ith loblolly pine (Weber and Stelzer 2002). In the initial stages of these clonal forestry programs, forest managers needed assurance that the clonal propagules growth corresponde d to that of seedlings. Therefore, earlier studies were designed to test whether cuttings grew sim ilarly to seedlings. Based on those results, it is generally accepted that cutt ings rooted from juvenile stock plants grow and perform comparably to seedlings. For example, Foster et al. (1987) reported that loblolly pine rooted cuttings should perform comparably to seedlings when the cuttings come from vigorous juvenile stock plants. In addition, McRae et al. (1993) concluded that for loblolly pine there were no signif icant differences between seedlings and rooted cutting propagules from common checklots thro ugh five years of growth. Similar results

PAGE 43

32 were obtained by Frampton et al. (2000) where they reporte d no significant differences between the means of rooted cuttings and seed lings for height, diamet er at breast height, and volume through six years in the field. Trials established with clones and seedlings from the same families provide an opportunity for comparing both half-sib and full-sib family performances across propagule types. Genetic correlations between propagule types can provide further assurance that selections made through traditional tree improvement activities for recurrent selection for general combining ab ility in seedling tests can also be used successfully in breeding families to test in a clonal forestry program. Although a number of studies have compared rooted cutting a nd seedlings, very few have been designed to estimate the genetic correlation betw een propagule types for a trait. While clonal tests derived from full-sib families provide an opportunity to estimate additive and nonadditive components of varian ce, tests should be designed with sufficient genetic structure in order to precisely quan tify the genetic variation. Several clonal studies have reported deficiencies in ma ting designs, number of parents and families (Frampton and Huber 1995; Paul et al. 1997; Isik et al. 2003). For example, Frampton and Huber (1995) reported that they had lo w power in partitioning the genetic variation because of the lack of a mating design amon g the parents of the full-sib crosses in a loblolly pine clonal study. In addition, Frampton and Foster (1993) warned that interpretation of the results may be difficult for studies that only include seedlings and cuttings from a common checklot to be comp ared to the clonal propagules from select parents and families. In this case, any diffe rences in the field performance because of

PAGE 44

33 propagule type may be confounded with th e differences in genetic improvement (Frampton and Foster 1993). The current study employs a complex geneti c structure to increase the power in quantifying the genetic variation associated with several growth traits in loblolly pine. More than 1,200 clones together with zygotic seedlings from the same 61 full-sib families were tested on multiple sites across the south eastern United States. Because of the test and mating designs, genetic correlations can be directly calculated between propagule types and within propagule types across sites. The specific objectives of this study are to 1) determine heritability estimates for various growth traits for loblolly pine clones and seedlings, 2) compare the performance betw een parents and full-sib families when grown as rooted cuttings and seedlings, and 3) de termine the extent of genotype x environment interaction by looking at the genetic correlations for parents, families, and clones across multiple sites. Materials and Methods Population The parental population consisted of tw enty first-generation and ten secondgeneration selections, subset from the Loblolly Pine Lower Gulf Elite Population. Two additional first-generation, slow-growing pare nts were included. The parental selections represent the Atlantic Coastal Plain, Florid a, and Lower Gulf provenances of loblolly pine (see FBRC 2000 for details). Briefly, th ese thirty-two loblolly pine parents were mated in a partial diallel design and created 70 full-sib families from which more than 2,000 seedling hedges were generated and give n unique clonal identifications (Appendix A). On average, each parent was involved in approximately four crosses.

PAGE 45

34 Propagation The propagation of the rooted cuttings fo r the field trials has been previously described (Chapter 2; Baltunis et al. 2005). But briefly, the seedling hedges were repeatedly sheared to slow down the effects of maturation and incr ease the number of shoots available for collection. Cuttings were collected from seedling hedges from 61 full-sib families, placed randomly in clonal-ro w plots, and replicated six times in a greenhouse in January 2002, April 2002, and Ju ne 2002. At the time of collection, the hedges were 22, 25, and 27 months old from seed, respectively. Cuttings were assessed for rooting at 11-weeks after setting (Baltunis et al. 2005). Rooted cuttings were then transplanted into Ray Leach Supercells (Steuwe and Sons, Corvalis, Oregon) and randomized into their designated field pl anting order and grown to size. The clonal propagules for a field trial came from a single rooting trial (Table 3-1). Table 3-1. Location of six fiel d trials, establishment date and total number of test trees for each test. Test Sticking Date Location Latitude Longitude Date Planted Total No. Test Trees A January 2002 Worth County, Georgia 0 2 4 4 31 N 0 5 5 5 83 W October 2002 9,216 B April 2002 Morgan County, Georgia 5 5 4 2 33 N 5 4 9 2 83 W November 2002 8,960 C April 2002 Putnam County, Florida 8 3 29 N 6 4 81W November 2002 8,960 D April 2002 Nassau County, Florida 3 2 5 4 30 N 7 2 4 5 81 W February 2003 8,960 E April 2002 Randolph County, Georgia 3 0 8 4 31 N 0 3 1 4 84 W December 2002 4,400 F June 2002 Santa Rosa County, Florida 5 0 0 5 30 N 7 5 1 1 87 W April 2003 6,912

PAGE 46

35 A single crop of seedlings, on the other hand, was used to produce all the seedlings for the field trials. Loblolly pine seed from the same full-sib families that the clones were derived from were stratified for about 30 days and then sown in May 2002. The seedlings were grown in family blocks in a different greenhouse than the cuttings. The seedlings were then moved outdoors under shade cloth and kept in their family blocks in a separate area from where the rooted cuttin gs were growing. The seedlings were not randomized into their designated fiel d order until just prior to planting. Field Design In total 47,408 measurement trees were established at six field sites across the southeastern United States (Table 3-1). Three trials each were established in Florida and Georgia (Table 3-1). An additional field trial was established in Virginia with a subset of the clones but was not included in any of these analyses. There were four replications in each of two cultural treatments (high and low intensity) in each test, except for Test E where there was only one cultural treatment a nd four replications. The goal for the high intensity treatment was to push the trees to their utmost potential by reducing competition and providing a non-limiting suppl y of nutrients, while the lo w intensive culture provides insights into family and clonal performa nce under a less optimal cultural regime (FBRC 2000). Both cultural intensities were treated similarly during the first year with cultural differences implemented at the beginning of the second growing season. Each trial contained 756-974 clones with approximately 15 clones from each of 61 full-sib families (Table 3-2). In total, more than 1,200 clones were planted in field trials (Table 3-2). The trials were designed to ha ve four zygotic seedlings from each of the same 61 full-sib families within each replicat ion. However, because of poor germination for some of the families or mortality in the nu rsery, each full-sib family is represented, on

PAGE 47

36 average, by approximately 27 zygotic seedlings per test (Table 3-2). Both the rooted cuttings and seedlings were planted in singl e-tree plots utilizi ng a resolvable alpha incomplete block design (Williams et al. 2002) in which incomp lete block size ranged from 10-14 trees. The variables measured were 1st year height, 2nd year height, height increment, and crown width. Table 3-2. Total number of clones, full-sib families, half-sib families, average number of clones per full-sib and half-sib family, and average number of seedlings per full-sib and half-sib family established at the six field trials. Test A Test B Test C Te st D Test E Test F Total Total # clones 974 941 942 956 868 756 1,212 Total # FS families 61 61 61 61 61 61a 61 Total # HS families 32 32 32 32 32 32b 32 Ave. # clones/FS family 16 15.4 15.4 15.7 14.2 12.4 19.9 Ave. # clones/HS family 60.9 58.8 58.9 59.5 54.2 47.2 75.7 Ave. # seedlings/ FS family 27.9 27 27 27 15.2 35.1 151.2 Ave. # seedlings/ HS family 106.5 103 103 103 58 106.3 576.5 a Only 47 full-sib families and b 31 half-sib families are represented in the seedling population of Test F. Statistical Analyses All growth variables, 1st year height, 2nd year height, height increment, and crown width, were analyzed in ASREML (Gilmour et al. 2002) using a parental model in order to estimate the genetic variance components a ssociated with those traits. Analyses were conducted for each trial separate ly and across all six trials for each propagule type. The

PAGE 48

37 across-trial analyses assumed a different error variance for each trial. For the clonal population, the following across-trial model was used. [3-1] ijklmnop mn ij k n ij k m ij k mn ijo ijmn ijn ijm mn io imn in im mn o mn n m ijk l ij k ij i ijklmnope fam r gca r gca r clone c t fam c t gca c t gca c t clone t fam t gca t gca t clone sca gca gca incbk R C T T y ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (* * * * * * * * where, ijklmnopy is the measured growth trait of the p th ramet of the o th clone within the mn th full-sib family in the l th incomplete block within the k th replication of the j th cultural treatment in the i th test is the clonal population mean iT is the fixed effect of trial, 6 , 1 i ijC T is the fixed effect of the interaction between trial and culture, 2 1 j ) ( ij kR is the fixed effect of replication, 4 3 2 1 k ) ( ijk lincbk is the random variable incomplete block associated with each test ~ NIID(0, 2iINC) mgca and ngca are the random variables female ( m ) and male ( n ) general combining ability, respectively ~ NIID(0, 2GCA) mnsca is the random variable specific combining ability ~ NIID(0, 2SCA) ) (mn oclone is the random variable clone w ithin full-sib family ~ NIID(0, 2CLONE)

PAGE 49

38 imgca t and ingca t are the random variables test by female general combining ability and test by male general combining ability interactions, respectively ~ NIID(0, 2TESTxGCA) imnfam t is the random variable test by full -sib family interaction ~ NIID(0, 2TESTxFAM) ) (*mn ioclone t is the random variable test by clone within full-sib family interaction ~ NIID(0, 2TESTxCLONE) ijmgca c t * and ijngca c t * are the random variables test by culture by female general combining ability and test by culture by male general combining ability, respectively ~ NIID(0, 2TxCxGCA) ijmnfam c t * is the random variable test by cu lture by full-sib family ~ NIID(0, 2TxCxFAM) ) (* *mn ijoclone c t is the random variable test by culture by clone within full-sib family interaction ~ NIID(0, 2TxCxCLONE) m ij kgca r) (* and n ij kgca r) (* are the random variables replication by female general combining ability and replication by male general combining ability, respectively ~ NIID(0,2REPxGCA) mn ij kfam r) (* is the random variable replication by full-sib family ~ NIID(0, 2REPxFAM) ijklmnope is the random error associated with each test ~ NIID(0, 2ERROR). The single-site model for the clonal populati on is identical, except that all model factors with subscript i are removed (sources involving test). All traits were also analyzed for th e seedling population assuming a randomized complete block design (incomplete block dr opped from model). Both single-trial

PAGE 50

39 analyses and an across-trial analysis we re performed again assuming heterogeneous errors across sites. [3-2] ijklmn lm ij k m ij k l ij k ijlm ijm ijl ilm im il lm m l ij k ij i ijklmne fam r gca r gca r fam c t gca c t gca c t fam t gca t gca t sca gca gca R C T T z ) ( ) ( ) ( ) (* * * * * * * where ijklmnz is the measured growth trait of the n th seedling from the lm th full-sib family in the k th replication of the j th cultural treatment in the i th test is the seedling population m ean and the other variables are defined as above (just adjusting the appropriate subscripts). Genetic parameters were estimated and st andard errors were calculated according to Foster and Shaw (1988) us ing the variance com ponents from the appropriate model. [3-3] AAA AA A GCA AV V V V 16 1 4 1 4 2 is the estimate of additive genetic variance. [3-4] DD AD AA D SCA DV V V V V 4 1 2 1 2 1 4 2 is the estimate of dominance genetic variance. [3-5] DD AD AA SCA GCA CLONE IV V V V 4 3 2 1 4 1 3 2 2 2 2 is the estimate of epistatic gene tic variance for the clonal population. [3-6] 2 2 2 2 CLONE SCA GCA GCV is the estimate of total genetic variance for the clonal population.

PAGE 51

40 [3-7] 2 2 4 4 SCA GCA GSV is the estimate of total genetic vari ance for the seedling population (assuming no epistasis). [3-8] 6 2 2 2 2 6 1 2 2 2 2 2 2 2 2 2 2 2 2 i ERROR REPxFAM REPxGCA TxCxCLONE TxCxFAM TxCxGCA TESTxCLONE TESTxFAM TESTxGCA CLONE SCA GCA Pi CV is the estimated phenotypic variance for the across-trial model for the clonal population. [3-9] 6 2 2 2 2 6 1 2 2 2 2 2 2 2 2 2 i ERROR REPxFAM REPxGCA TxCxFAM TxCxGCA TESTxFAM TESTxGCA SCA GCA Pi SV is the estimated phenotypic variance for the across-trial model for the seedling population. Individual tree narrow-sense heritability (2 h) and broad-sense heritability (2 H ) were derived using the estimated variance com ponents for all the growth traits and each propagule type at each site and across sites. Standard errors were calculated using Taylor series expansion (Kendall and Stuart 1963; Namkoong 1979; Huber et al. 1992; Dieters 1994). The following heritability formul ae were used for the clonal data. [3-10] C CP GCA P AV V V h 4 2 2 is the across-trial estimate of individual tree narrow-sense heritability for the clonal population, where CPV is from [3-8]. [3-11] C CP CLONE SCA GCA P GV V V H 2 2 2 2 2

PAGE 52

41 is the across-trial estimate of individual tree broad-sense heritability for the clonal population, where CPV is from [3-8]. Heritability estimates were also obtaine d for the seedling data for each trial and across-trials. [3-12] S SP GCA P AV V V h 4 2 2 is the across-trial estimate of individual tree narrow-sense heritability for the seedling population, where SPV is from [3-9]. [3-13] S SP SCA GCA P GV V V H 4 4 2 2 2 is the across-trial estimate of individual tree broad-sense heritability for the seedling population, where SPV is from [3-9]. The various growth variables at each site we re also analyzed with a bivariate mixed model with the growth of the clones and seedlings as two de pendent variables. Type B genetic correlations for general combining ability (propGCABr ) and full-sib family value (propFSBr ) between cuttings and seedlings were esti mated in order to compare parental and full-sib family performance between propagul e types. The geneti c correlation between propagule types for additive effects, for ex ample, gives us an indication of whether parental ranks are dependent upon whether their progeny are grown as cuttings or seedlings, while a type B genetic correlation at the full-sib family level measures the performance of full-sib families across propagule types.

PAGE 53

42 [3-14] SEED CUT SEED CUT BGCA Var GCA Var GCA GCA Cov rpropGCA is the type B genetic correlation of additive effects between propagule types using genetic variance component estimates fr om the bivariate analysis. [3-15] SEED SEED CUT CUT SEED CUT SEED CUT BSCA Var GCA Var SCA Var GCA Var SCA SCA Cov GCA GCA Cov rpropFS 2 2 , 2 is the type B genetic correlation of full-sib family values between propagule types using genetic variance component estimates from the bivariate analysis. The extent of genotype x environment in teraction was investigated by analyzing data across trials for each propagule type using the variance components from the appropriate model. Type B gene tic correlations were calculated for additive effects, fullsib family, and the total genetic or clonal value across the trials (Yamada 1962; Burdon 1977). Standard errors of type B correlati ons were calculated using the Taylor series expansion method. [3-16] 2 2 2 TESTxGCA GCA GCA BgxeGCAr is the type B genetic correlation for additive effects across trials. [3-17] 2 2 2 2 2 2 2 2 2 TESTxFAM SCA TESTxGCA GCA SCA GCA BgxeFSr is the type B genetic correlation for full-sib families across trials. [3-18] 2 2 2 2 2 2 2 2 2 2 2 2 TESTxCLONE CLONE TESTxFAM SCA TESTxGCA GCA CLONE SCA GCA BgxeCLONEr is the type B genetic correlation for total ge netic or clonal value across trials for the clonal population.

PAGE 54

43 Ranging from 0 to 1, a value of gxeGCABr near one indicates little genotype x environment interaction and that the parents ranked the same across the trials, while a low gxeGCABr (near zero) indicates that parental ra nks were not stable across the sites and hence, genotype x environment interaction exists. A high gxeFSBr indicates that full-sib families performed similarly across the sites, while a highgxeCLONEBr indicates that the total genetic values of the clones were stable across the trials. Results and Discussion Overall Growth of Clones and Seedlings Survival of both clonal a nd zygotic propagules was hi gh across all of the field trials. Survival of the rooted cuttings range d from 91.7-98.3% at the six field trials, while seedling survival ranged from 86.5-97.4%. At the time of planting, seedlings were generally taller than rooted cuttings, and th is trend has continued through year two (Table 3-3). For example, after th e first and second growing seas ons seedlings were on average 11 cm and 10 cm taller than rooted cuttings, respectively. It has been suggested that propagule size differences at the time of plan ting may create difficulties in the analysis and interpretation of subsequent growth (F rampton and Foster 1993). However, these differences in initial propagule size may not be a problem in the current study because height increments were very similar betw een both propagule types, with site means ranging between 1.0-2.0 m and 1.0-1.9 m for r ooted cuttings and seedlings, respectively (Table 3-3). The initial and first year treatments we re the same within a site during test establishment, and there were no cultural differences between propagule types for 1st year height. There were some differences in m ean height due to the effects of cultural

PAGE 55

44 intensity during the seco nd growing season at some of the sites. However, these effects were more a function of scale. Although the overall means of the growth variables differed by cultural treatment, the ranks of pare nts, families, or clones were not affected. Type B genetic correlations exceeded 0.85 i ndicating little cultural treatment x genetic effect interaction, and therefor e, cultural effects ar e ignored for the purposes of this study. This implies that the rankings of parents, families, and clones were robust across multiple management regimes through age two. Table 3-3. Mean 1st year height, 2nd year height, height incr ement, and crown width by propagule type for each of the six fiel d trials. Although means are expressed in meters, analyses were conducted us ing measured traits in centimeters. 1st Year Height (m) 2nd Year Height (m) Height Increment (m) Crown Width (m) Clones 1.0 2.1 1.2 0.9 Test A Seedlings 1.1 2.3 1.2 1.0 Clones 0.8 1.8 1.0 0.9 Test B Seedlings 0.9 1.8 1.0 0.9 Clones 1.0 2.1 1.1 1.2 Test C Seedlings 1.1 2.3 1.2 1.3 Clones 0.7 2.0 1.3 1.0 Test D Seedlings 0.8 2.1 1.3 1.1 Clones 1.2 3.3 2.0 2.0 Test E Seedlings 1.3 3.2 1.9 2.0 Clones 0.5 1.7 1.3 1.1 Test F Seedlings 0.6 1.8 1.3 1.1 Genetic Components of Variance All of the early growth traits demonstrated genetic variat ion (Figure 3-1; Table 3-4; Appendix C). However, the genetic variat ion partitioned differently for the two propagule types. In all cases, the estimate of additive genetic vari ation was greater for

PAGE 56

45 the clones than the seedlings. Within a si ngle trial, for instance, the estimate of the additive genetic variation for 2nd year height based on seedlings was 0 to 0.58 that of additive genetic variation based on clonal data Similar trends were observed from the across-trial analyses. The estimate of add itive genetic variation for the clonal material was 4.7, 3.3, 2.9, and 2 times greater than th e additive genetic vari ation for seedlings for 1st year height, 2nd year height, height increment, and crown width, respectively (Table 34). Based on single-trial analyses, the majority of the genetic varia tion associated with all of the growth variables in the clonal population was a dditive, while in the seedling population the trend was towards nonadditive genetic variation (Figure 3-1). For example, at Test D all of the ge netic variation associated with 2nd year height of clones was additive, while for the seedlings it was dominance genetic variation (Figure 3-1). When all of the data were analyzed togeth er, then the estimates of dominance genetic variation were approximately equivalent for both propagules (Table 3-4) suggesting a large test by dominance interaction for the seedlings. Epistasis was negative for all growth variables. As a result, estimates of additive and dominance genetic variance for these traits might not be upwardly biased as indicated by the expected portions of epistatic interactions defined in Equations 3-3 and 3-4 (Foster and Shaw 1988). Isik et al. (2003) reported similar trends in the partitioning of the ge netic variation for clones and seedlings including negative epistasis estima tes for height, diameter, and volume through age six. They also reported that the add itive genetic variation for growth traits was always greater for clones than the estim ate for seedlings, while dominance genetic variation was greater in the seedling population (Isik et al. 2003).

PAGE 57

46 Table 3-4. Genetic parameter estimates for 1st year height, 2nd year height, height increment, and crown width by propagule t ype across all six tr ials. Standard errors are given in parentheses. 1st Year Height 2nd Year HeightHeight Increment Crown Width Clone Seed Clone Seed Clone Seed Clone Seed AV 69.8 (21.3) 14.7 (7.8) 401.5 (122) 119.9 (50.8) 152.6 (47) 51.8 (24.2) 110.3 (34) 55 (20) DV 12 (6.2) 17.5 (9.3) 86.4 (36.5) 83.3 (41.6) 30.3 (13.6) 41.9 (21.5) 19.7 (10.1) 20 (11.7) IV -11.5 (11.4) ---107.6 (64.3) ---39.7 (24.8) ---13.5 (18.1) --GV 70.3 (10.8) 32.1 (9.5) 380.3 (61.1) 203.1 (53.8) 143.2 (23.6) 93.65 (26.1) 116.5 (17.2) 75 (20.7) PV 442.8 (11.3) 405.2 (7.6) 1806 (62.3) 1703 (36.2) 902.6 (24.5) 972 (19.4) 640.9 (17.8) 649.5 (14.3) 2 h 0.16 (0.04) 0.04 (0.02) 0.22 (0.06) 0.07 (0.03) 0.17 (0.05) 0.05 (0.02) 0.17 (0.05) 0.08 (0.03) 2 H 0.16 (0.02) 0.08 (0.02) 0.21 (0.03) 0.12 (0.03) 0.16 (0.02) 0.1 (0.02) 0.18 (0.02) 0.11 (0.03) gxeGCABr 0.81 (0.06) 0.71 (0.2) 0.88 (0.04) 0.78 (0.12) 0.83 (0.05) 0.6 (0.15) 0.82 (0.06) 0.78 (0.12) gxeFSBr 0.8 (0.05) 0.5 (0.1) 0.88 (0.04) 0.66 (0.09) 0.83 (0.05) 0.58 (0.1) 0.82 (0.05) 0.67 (0.09) gxeCLONEBr 0.69 (0.04) --0.77 (0.03) --0.76 (0.04) --0.76 (0.03) --Genetic causes are not the only source for similarity among relatives. Non-genetic factors, such as C effects, can lead to upw ardly biased estimates of total genetic and nonadditive genetic components of variance when analyzing clonal data (Libby and Jund 1962). C effects are often assumed negligible when estimating epistatic variance (Foster and Shaw 1988). If C effects are present then the total genetic variat ion associated with

PAGE 58

47 clones will be overestimated (Libby and Jund 1962) Significant C eff ects are likely to occur in traits that are measured soon after propagation, but apparent ly lessen for traits measured at later times (Libby and Jund 1962). In the current experiment, if C effects exist, then estimates of ep istatic genetic variation will be confounded with C effects because ramets of a clone came from a singl e, non-replicated hedge. However, the estimates for epistasis were negative for all of the growth traits (Table 3-4), which suggest that interloci interactions are not an important source of genetic variation for early growth traits, and that C effects may not be a major contributing factor. Further, estimates of additive and dominance shoul d be free of confounding C effects in the current experiment since randomization of cl ones occurred at all stages including hedge establishment, propagation, and growth pr ior to and after test establishment. As Falconer and Mackay (1996) point out, re latives of all sorts may resemble one another because of sharing a common environment. The variance attributed to a common environment occurs more frequently and contri butes greater to the c ovariance of full-sibs than to the covariance of any other sort of relatives (Falconer and Mackay 1996). When a common environment effect exists, then the differences between means of families become greater than when these non-genetic fact ors are not present. In the case of fullsib families, then this will lead to biased or inflated estimates of dominance genetic variation. The results seen in the current experiment for dominance genetic variance estimates in the seedling population appear to be a result of a partitioni ng problem in that estimates of the variance due to dominance are inflated at the expense of addi tive effects (Figure 31). Seeds are often sown and seedlings grown in full-sib family blocks in the greenhouse

PAGE 59

48 Figure 3-1. The proportion of additive (a dd) and nonadditive (na) genetic variance components for clones and seedlings across the six field trials, where 2h = add and 2 H = add + na: a. 1st year height, b. 2nd year height, c. Height increment, and d. Crown width. Standard error bars for broad-sens e heritability estimat es are included. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7clone seed clone seed clone seed clone seed clone seed clone seed Test ATest BTest CTest DTest ETest F Broad-Sense Heritabilit y na add a. 1st year height 0 0.1 0.2 0.3 0.4 0.5 0.6clone seed clone seed clone seed clone seed clone seed clone seed Test ATest BTest CTest DTest ETest F Broad-Sense Heritabilit y na add b. 2nd year height 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45clone seed clone seed clone seed clone seed clone seed clone seed Test ATest BTest CTest DTest ETest F Broad-Sense Heritabilit y na add c. Height increment 0 0.1 0.2 0.3 0.4 0.5 0.6clone seed clone seed clone seed clone seed clone seed clone seed Test ATest BTest CTest DTest ETest F Broad-Sense Heritabilit y na add d. Crown width

PAGE 60

49 or nursery for progeny testing or other field testing. In order to eliminate the common environment effect at the family level, then seedlings within a full-sib family should be randomized over the environment in which th ey are being grown and tested. Although rooted cuttings were randomized at all stages of propagation, the seedlings were in fact grown in full-sib blocks, and apparently a co mmon environment effect has carried over to the field through age two measurements. Heritability Estimates First year height, 2nd year height, height increment, and crown width were all influenced by additive genetic variation. Individual tree narrow-sense heritability was always greater for the clones than for seedlings for all of the traits (Figure 3-1, Table 34). Generally, 2 h estimates of total height increased from age one to age two for both clones and seedlings (Figure 3-1, Table 3-4). Narrow-sense heritability for total height increased from 0.16 at age one to 0.22 at ag e two for the clones, while in the seedling population, 2 h increased from 0.04 to 0.07 (Table 3-4). In addition, 2 h estimates for all of the growth traits from the seedling data had larger standard errors associated with them than the estimates from clones (Table 3-4). Ye t, the differences in heritability estimates between propagule types may be related to C effects. If C effects are present, then 2 H estimates from the clonal population may be inflated. However, estimates of 2 h from the clonal population may not be upwardly biased si nce clones within full -sib families were randomized in the hedge orchard and thr oughout the study, thus reducing spurious C effects at the parental and full-sib family levels. Heritability estimates have been reported for several loblolly pine seedling and clonal populations. In a study reported by Isik et al. (2003), individual tree narrow-sense

PAGE 61

50 heritability for volume at age six was 0.3 for clones, while for seedlings 2 h was 0.06. In another loblolly pine clonal te st, heritability estimates for total height growth through age five for two factorials were simila r to the values reported here (Paul et al. 1997). They reported 2 h for 1st and 2nd year height as 0.08 and 0.17, respectively (mean of two factorials). Unfortunately, they did not include seedlings in their study. Analogous results have been reported in Eucalyptus globulus Labill. for heritability estimates for diameter for both clones and seedlings in that 2 h was always greater for the clonal population than the seedling population (Costa e Silva et al. 2004). Broad-sense heritability estimates for the va rious growth traits were always larger for the clonal material than the se edlings except for the estimates of 2 H for 1st year height at Test E and for all four variab les at Test D (Figure 3-1). However, 2 H estimates from the across-trial models were always great er for the clones than seedlings (Table 34). Because of the negative estimates of epistasis in the clonal population, 2 H estimates were equivalent to the estimates for individual tree narrow-sense heritability for all traits. As was the case with2 h, seedling estimates of 2 H had higher standard errors associated with them (Figure 3-1). Type B Genetic Correlations Between Propagule Types In order to further compare the clonal rooted cuttings and zygotic seedlings, type B genetic correlations for general combining ab ility and full-sib family value between propagule types were estimated. For all gr owth traits measured, the type B genetic correlation between propagule types for additiv e effects was high with values exceeding 0.72 (Table 3-5). For example, propGCABr for 1st year height ranged from 0.93-0.99, while

PAGE 62

51 for height increment propGCABr ranged from 0.72-0.99 (Table 3-5). These high genetic correlations imply that parental rankings for ear ly growth traits are stable regardless of whether their progeny are being tested as zygot ic seedlings or rooted cuttings (Figure 32a). Table 3-5. Genetic correlati ons between propagule types for 1st year height, 2nd year height, height increment, and crown width at the parental (propGCABr ) and full-sib family (propFSBr ) levels. 1st Year Height 2nd Year HeightHeight Increment Crown Width Test propGCABr propFSBr propGCABr propFSBr propGCABr propFSBr propGCABr propFSBr A 0.99 0.62 0.99 0.65 0.99 0.38 0.86 0.86 B 0.99 0.74 0.99 0.86 0.99 0.99 0.99 0.81 C 0.93 0.74 0.99 0.75 0.99 0.99 0.99 0.97 D 0.93 0.44 0.99 0.64 0.99 0.92 0.93 0.92 E 0.99 0.55 0.89 0.71 0.72 0.77 0.99 0.69 F 0.95 0.49 0.96 0.89 0.95 0.83 0.99 0.83 The type B genetic correlations for a dditive effects betw een propagule types observed in the current study were consistent with the expectations reported by Borralho and Kanowski (1995). In a simulation study comparing the perfor mance of clones and seedlings from the same half-sib family, Bo rralho and Kanowski ( 1995) reported that the expected correlations between propagule t ypes exceeded 0.8 when greater than 100 seedlings or propagules were tested. Additionally, in a field study comparing rooted cuttings and seedlings from four half-sib families, Foster et al. (1987) reported that family rank correlations between propagule types were positive and significant for 1st year height (0.52), 3rd year height (0.66) and 6th year height (0.70).

PAGE 63

52 Full-sib families also ranked relatively similar regardless of propagule type at most of the sites (Table 3-5; Figure 3-2b). For ex ample, type B genetic correlations at the fullsib family level ranged from 0.64-0.89 for 2nd year height. However, propFSBr was more variable from site to site for 1st year height and height incr ement. The results observed here are in accordance with those reported by Frampton et al. (2000) in that full-sib families or parental trees selected based on seedling genetic trials should also perform well as rooted cuttings. Genotype x Environment Interaction The stability of parents and full-sib families across sites was compared for the clonal and seedling populations (Table 3-4). The type B genetic correlation for additive effects across sites was always greater for the clones, although these estimates were moderately high for both populations. For example, gxeGCABr for 2nd year height was 0.88 and 0.78 for the clonal and seedling populati ons, respectively (Table 3-4). Full-sib families also ranked comparably across sites for the clonal population with gxeFSBr values exceeding 0.8 for all of the early growth trai ts (Table 3-4). These across site genetic correlations were estimated with more precisi on using clonal replicates as evidenced by lower standard errors for the estimates from the clonal population. An additional genotype x environment inte raction between test and total genetic value can be estimated using clonal tests. The total genetic values of the clones were fairly stable when all of the data was cons idered from all sites, indicating that a good clone at one site is good at al l the sites (Table 3-4). Howe ver, based on analyses from pairs of trials (data not show n), there appears to be a propa gation effect relating to the season the cuttings were rooted. For exam ple, the worst genetic correlations were

PAGE 64

53 observed between the trial esta blished with rooted cuttings from the winter setting and any of the other trials, while the best genetic correlations were obtained from the field trials that contained rooted cuttings originat ing from the spring setting. Similar results Figure 3-2. Rank-rank plots showing type B genetic corre lations between clones and seedlings from Test B based on: a. Pare ntal BLUP values, b. Full-sib family BLUP values, where full-sib BLUP values are equal to the sum of the predicted general combining ability fo r each of the two parents plus the predicted specific combining ability of the cross. 0 5 10 15 20 25 30 35 05101520253035 Test B Parental Ranks ClonesTest B Parenatl Ranks Seedlings a. Parental ranks for 2nd year height 0 10 20 30 40 50 60 70 010203040506070 Test B Full-Sib Family Ranks ClonesTest B Full-Sib Family Ranks Seedlings b. Full-sib family ranks for 2nd year height 99 0 propGCABr 86 0 FS propBr 86 0 propFSBr

PAGE 65

54 were also observed with rooting with this same popul ation in that poor genetic correlations were observed between rooting ab ility in winter and spring (Chapter 2; Baltunis et al. 2005). Conclusion Several forest industries in the southeaste rn United States ar e deploying full-sib families of loblolly pine operationally. In addition, many of these companies are pursuing clonal forestry programs with loblolly pine. Genetic field trials established with clones and seedlings from the same full-sib families provide an opportunity for comparing both half-sib and full-sib family performances for both propagules. Based on the current study, several conclusions can be dr awn. First, clonally replicated seedling trials of loblolly pine provide genetic in formation with greater precision than zygotic seedlings. Second, genetic correlations between estimates of the genetic effects associated with these growth traits between these propagule types were highly favorable. These high genetic correlations between propagul e types reassure that parental and fullsib family rankings are stable regardless of propagule type. This implies that parental and full-sib family rankings based on existi ng seedling progeny trials could be used to select parents and families that perform well when they are deployed as rooted cuttings. Third, little genotype x envir onment interaction was observed across sites at the parental, family, and clonal level for all traits. Howeve r, there appears to be a carry-over effect relating to the season in which the cuttings we re rooted for the clonal material. Finally, randomization is essential at all stages in testing when estimating genetic parameters. The lack of randomization for the seedling popu lation apparently resulted in a problem with partitioning of the genetic variance, causing full-sib families to appear more different and inflating estimates of dominance genetic variation.

PAGE 66

55 CHAPTER 4 GENETIC GAIN FROM SELECTION FO R ROOTING ABILITY AND EARLY GROWTH IN VEGETATIVELY PROPAGAT ED CLONES OF LOBLOLLY PINE Introduction Loblolly pine ( Pinus taeda L.) is the most important co mmercial tree species in the southeastern United States (McKeand et al. 2003). Genetic improveme nt of loblolly pine has been occurring since the 1950s in seve ral tree improvement programs. These programs have aimed to increase the populati on mean breeding value of a few key traits, such as stem volume, disease resistance, and wood properties, through breeding and selection of superior genot ypes. Tree improvement pr ograms are based on recurrent selection for general combining ability and ca pture only a portion of the additive genetic variation with open-pollinated seedlings fo r deployment. However, the nonadditive portion of genetic variation, dominance and ep istasis, may be important components of variation for traits. For example, Stonec ypher and McCullough (1986) reported that the estimates of nonadditive variance were approx imately equal to those of additive variance for growth traits in Douglas-fir ( Pseudotsuga menziesii (Mirb.) Franco) through age six. Additionally, Paul et al. (1997) reported that both a dditive and dominance genetic variance increased from age one to five for he ight growth for loblolly pine clones. The only manner in which to capture the tota l genetic variation is through operational deployment of clonal propagules.

PAGE 67

56 Based on current technologies, several forest industries in the southeastern United States are pursuing clonal fore stry programs for loblolly pine using either somatic embryogenesis or rooted cuttings (Weber and Stelzer 2002). Two main criteria need to be met prior to operational deployment of loblolly pine clones. First, loblolly pine clones must perform well, e.g. meet the selection criteria for the desired traits. This involves the accumulation of reliable data for the cl ones from greenhouse screening, field trials, etc. Second, the selected clones have to be propagate d in large enough numbers for deployment. For a rooted-cutting-based clonal program, this involves bulking up the number of hedges (ramets) of a particular cl one or group of selected clones through serial propagation and then producing reforestation stock efficiently from the bulked-up clones. Only those tested clones that can be propa gated easily in sufficient numbers will be economically feasible for deployment. Loblolly pine is considered a difficult to root species (Wise an d Caldwell 1994). In populations that have not experi enced any selection pressure fo r rooting ability, loblolly pine has been reported to root near 50% (Foster 1990; Baltunis et al. 2005). Previous rooting studies with loblolly pine have demonstrated substantial family and clonal variation for rooting ability (Foster 1990; Baltunis et al. 2005), indicati ng the potential for increasing rooting efficiency for both clonal deployment and through recurrent selection and breeding. Sel ection for both rooting ability and field growth will be necessary for a successful loblolly pine cl onal forestry program ba sed on rooted cutting technology (Foster et al. 1985; Foster et al. 2000; Baltunis et al. 2005). The objectives of this study were to (i ) determine the genetic correlation between rooting ability and 2nd year height, (ii) predict the gene tic gain associated with selection

PAGE 68

57 for rooting ability, (iii) predict the genetic gain associated with selection for 2nd year height, and (iv) predict the ge netic gain from combined sele ction for rooting ability and 2nd year height using a Mont e Carlo selection index. Materials and Methods Population The parental population consisted of tw enty first-generation and ten secondgeneration selections from the larger Loblolly Pine Lower Gulf Elite Population. Two additional first-generation, slow-growing parent s were included. The parental selections represent the Atlantic Coastal Plain, Florid a, and Lower Gulf provenances of loblolly pine. These thirty-two loblolly pine parent s were mated in a partial diallel design and created 70 full-sib families from which approximately 2,200 seedling hedges were generated and given unique clonal identificati ons (Appendix A). On average, each parent was involved in about four crosses. Rooting and Field Trials The propagation of the rooted cuttings for this study has previously been described (Baltunis et al. 2005; Chapter 2). But briefly, th e seedling hedges were repeatedly sheared to slow down the effects of matu ration and increase the number of shoots available for collection. Cuttings were set in five rooting trials over two years in May 2001, July 2001, January 2002, April 2002, and June 2002 in trials Spring01, Summer01, Winter02, Spring02, and Summer02, respectively Four to nine ramet clonal row plots were set in four to six replications depe nding on trial (Chapter 2) At the time of collection, the hedges were 13, 15, 22, 25, and 27 months old, respectively. Cuttings were assessed for rooting nine to elev en weeks after they were set (Baltunis et al. 2005; Chapter 2).

PAGE 69

58 Six field trials (A, B, C, D, E, F) were established with rooted cuttings from the three latter rooting trials. Th ree field trials each were esta blished in Georgia and Florida between October 2002 to April 2003 (see Table 3-1). The clonal propagules for each field trial, however, came from a single rooti ng trial. Single-tree-p lots of a clone were established in four replications in each of two cult ural treatments (high and low intensity) in each test, except for Test E where there was only one cultural treatment and four replications. The goal for the high intensit y treatment was to push the trees to their utmost potential by reducing competition and providing a non-limiting supply of nutrients, while the low intensive culture provides insights into family and clonal performance under a less optimal cultural re gime (FBRC 2000). Previous analyses of growth for this population had shown that type B genetic correlations exceeded 0.85 indicating little cultural treatme nt x genetic effect interact ion, and therefore, cultural treatment effects were ignored for th e purposes of this study (Chapter 3). Statistical Analyses A bivariate parental linear mixed-effects model was used to obtain estimates of variance and covariance component s for rooting ability and 2nd year height using ASREML (Gilmour et al. 2002): [4-1] i i q i p i o i c i f i u i n i i ii i i i i i ie q Z p Z o Z c Z f Z u Z n Z b X y where iy is the vector of ob servations indexed ( i ) by rooting ability and 2nd year height, ib is the vector of fixed effects (i.e. mean, trials and replications within trials) and iX is the known incidence matrix re lating the observations in iy to the fixed effects in ib where height root height root i ib b X 0 0 X b X,

PAGE 70

59 in is the vector of random incomplete blocks nested within replication and test effects ~MVN 2 2 ,height rootINC height INC root I 0 0 I 0 iu is the vector of random parent (female and male) general combining ability effects ~MVNA G 0 where 2 2 height rootheight rootheight rootGCA GCA GCA GCA G and A= numerator relationship matrix, if is the vector of random specifi c combining ability effects ~MVNsI S 0, where 2 2 height rootheight rootheight rootSCA SCA SCA SCA S and sI is an identity matrix of size equal to the number of full-sib families, ic is the vector of random clones wi thin full-sib family effects ~MVNcI C 0, where 2 2 height rootheight rootheight rootCLONE CLONE CLONE CLONE C and cI is an identity matrix of size equal to the number of clones, io is the vector of random test by full-sib family effects ~MVN 2 2 ,height rootTESTxFAM height TESTxFAM root I 0 0 I 0, ip is the vector of random test by cl one within full-sib family effects ~MVN 2 2 ,height rootTESTxCLONE height TESTxCLONE root I 0 0 I 0,

PAGE 71

60 iq is the vector of random replication w ithin test by full-sib family effects ~MVN 2 2 ,height rootREPxFAM height REPxFAM root I 0 0 I 0, ie is the random vector of residual terms ~MVN 2 2 ,height rootERROR height ERROR root I 0 0 I 0, inZ, iuZ, ifZ, icZ, ioZ, ipZ, and iqZ are the known incidence matrices relating the observations in iy to effects in in, iu, if, ic, io, ip, and iq, respectively, and iI is the identity matrix of dimension equal to the num ber of observations for rooting ability or 2nd year height. Causal Components of Variance Genetic parameters were estimated and sta ndard errors were calculated for rooting ability and 2nd year height from the bivariate an alysis according to Foster and Shaw (1988). [4-2] AAA AA A GCA AV V V V 16 1 4 1 4 2 is the estimate of additive genetic variance. [4-3] DD AD AA D SCA DV V V V V 4 1 2 1 2 1 4 2 is the estimate of dominance genetic variance. [4-4] DD AD AA SCA GCA CLONE IV V V V 4 3 2 1 4 1 3 2 2 2 2 is the estimate of epistatic genetic variance. [4-5] 2 2 2 2 CLONE SCA GCA GV is the estimate of total genetic variance.

PAGE 72

61 [4-6] 2 2 2 2 2 2 2 ERROR REPxFAM TESTxCLONE TESTxFAM CLONE SCA GCA PV is the estimate of the phenotypic variance. [4-7] trfcn tr tfc tf fc f VERROR REPxFAM TESTxCLONE TESTxFAM CLONE SCA GCA PHS2 2 2 2 2 2 2 is the phenotypic variance of half-sib family means, f = harmonic mean number of fullsib families per half-sib family, c = harmonic mean number of clones per full-sib family, t = number of trials, r = harmonic mean number of replications per test, and n = harmonic mean number of ramets per cl one per replica tion per test ( n = 1 for field trials). [4-8] ctrn tr tc t c VERROR REPxFAM TESTxCLONE TESTxFAM CLONE SCA GCA PFS2 2 2 2 2 2 2 2 is the phenotypic variance of full-sib family means. [4-9] trn tr t t VERROR REPxFAM TESTxCLONE TESTxFAM CLONE SCA GCA PCL2 2 2 2 2 2 2 2 is the phenotypic variance of clonal means. Heritability Estimates Heritability estimates have previously been reported for this population for rooting ability (Baltunis et al. 2005; Chapter 2) and 2nd year height (Chapter 3). However, since rooting ability and 2nd year height were analyzed usi ng a bivariate model in the current study, heritabilities were estimated based on the genetic parameter estimates from the bivariate model and differed slightly from t hose reported previously. Standard errors were calculated using the Taylor series expansion method (Kendall and Stuart 1963; Namkoong 1979; Huber et al. 1992; Dieters 1994).

PAGE 73

62 [4-10] 2 2 2 2 2 2 2 2 2 2 4 ERROR REPxFAM TESTxCLONE TESTxFAM CLONE SCA GCA GCA P AV V h is the individual tree narrow-sense heritability. [4-11] 2 2 2 2 2 2 2 2 2 2 2 2 2 ERROR REPxFAM TESTxCLONE TESTxFAM CLONE SCA GCA CLONE SCA GCA P GV V H is the individual tree broad-sense heritability. [4-12] HSP GCA HSV H 2 2 is the half-sib family mean heritability. [4-13] FSP SCA GCA FSV H 2 2 2 2 is the full-sib family mean heritability. [4-14] CLP CLONE SCA GCA CLV H 2 2 2 2 2 is the clonal mean heritability. Type B Genetic Correlations Although the same genotypes, or clones, were tested in the rooting and field trials, these measurements were not necessarily taken on the same ramet. Therefore, type B genetic correlations between rooting ability and 2nd year height for general combining ability, full-sib family value, and the total ge netic value were calcula ted. Standard errors of these estimates were calculated using the Taylor series expansi on method (Kendall and Stuart 1963; Namkoong 1979; Huber et al. 1992; Dieters 1994). [4-15] 2 2 height root rootheight GCAGCA GCA GCA Br is the type B genetic correl ation between rooting ability and 2nd year height for additive effects, and rootheightGCA is the covariance between general combining ability effects for rooting ability and 2nd year height..

PAGE 74

63 [4-16] 2 2 2 2 2 2 2 height height root root rootheight rootheight FSSCA GCA SCA GCA SCA GCA Br is the type B genetic correlation between rooting ability and 2nd year height for full-sib families, and rootheight rootheightSCA GCA 2 is the covariance between the full -sib family effects for rooting ability and 2nd year height. [4-17] 2 2 2 2 2 2 2 2 height height height root root root rootheight rootheight rootheight TGCLONE SCA GCA CLONE SCA GCA CLONE SCA GCA Br is the type B genetic correlation between rooting ability and 2nd year height for the total genetic value of clones, and rootheight rootheight rootheightCLONE SCA GCA 2 is the covariance between the total clonal value effects for rooting ability and 2nd year height. Genetic Gain Genetic gain was estimated for a number of deployment options based on various selection scenarios using the BLUP values from the bivariate analysis. All deployed populations were assumed to be propagated as rooted cuttings. An additional assumption for gain calculations was that the seventy fu ll-sib families and all clones within families were available for deployment. The deploymen t strategies considered were 1) half-sib family deployment, 2) full-sib family deploymen t, and 3) clonal deployment. For a halfsib family deployment option, the predicted value for each parent (general combining ability) was used to select the best half-sib family for deployment as rooted cuttings. Full-sib family values were calculated by summing the predicted values for the female general combining ability, male general combining ability, and the full-sib family specific combining ability. The genetic gain over the trait mean was determined for deployment of the best full-sib family.

PAGE 75

64 The total genetic value of a clone was de termined by summing the predicted values for the female general combining ability, ma le general combining ability, the full-sib family specific combining ability, and clone within full-sib family. Several clonal deployment strategies were compared. Firs t, the genetic gain was determined for deployment of the single best clone for each tr ait. A second clonal deployment strategy was considered by selecting the top ten full-sib families and then deploying the single best clone from each of these ten families. Two additional clonal deployment strategies were compared by selecting the top 10% and 1% of clones (out of 2200 possible clones) using an unrestricted selection index for 2nd year height and roo ting ability. The total genetic values of the clones were weighted with the following weights: 1:0, 0.9:0.1, 0.8:0.2, ..., 0.2:0.8, 0.1:0.9, 0:1 for 2nd year height and rooting ab ility, respectively, in the Monte Carlo index (Cotterill and Dean 1990) All gains were expressed as the percentage gain over the mean of the trait: [4-18] % 100 % i i iy y x Gain, where ix is the average predicted value for trait i of the selected population, and iy is the population mean for eith er rooting ability or height. In addition genetic gain was calculated us ing theoretical gain formulae assuming an unrelated population (Falcone r and Mackay 1996), and these estimates were compared with the predicted genetic gain based on BLUP values for the half-sib family, full-sib family, and best clone deployment opti ons. The following formulae were used:

PAGE 76

65 [4-19] HSP HS hsV H i Gain 2 is the genetic gain associated with selection of the best halfsib family, and hsi is the selection intensity corresponding to selecting one out of thirtytwo half-sib families (i = 2.07) and assuming all half-sib families are unrelated, [4-20] FSP FS fsV H i Gain 2 is the genetic gain associated with selection of the best fullsib family, and fsi is the selection intensity correspo nding to selecting one out of seventy full-sib families (i = 2.38) and assuming all full-sib families are unrelated, [4-21] CLP CL cV H i Gain 2 is the genetic gain associated with clonal selection, and ci is the selection intensity co rresponding to selecting one out of 2,206 clones (i = 3.58) and assuming all individu als are unrelated. Results and Discussion Causal Components of Variance The genetic parameter estimates from the bi variate analysis of rooting ability and 2nd year height in loblolly pine clones we re consistent with the estimates from their respective univariate analyses and both traits showed gene tic variation (Table 4-1; Appendix D). Additive genetic variation accounted for the majority of the genetic variation associated with rooting ability and 2nd year height (Table 4-1). Dominance genetic variation contributed a minor portion to the total genetic varia tion for both traits. Epistasis appears to be more important for rooting ability than 2nd year height as evidenced by a negative estimate of epistasis for 2nd year height (Table 4-1). However, estimates of epistasis may be confounded with C effects since rooted cuttings of a clone originated from a single ortet.

PAGE 77

66 Type B Genetic Correlations The bivariate analysis of five rooting tria ls and six field trials also allowed for estimation of the genetic covariance between rooting ability and 2nd year height for parental effects, full-sib family effects, a nd the total genetic value of clones within fullsib family. There was a positive genetic re lationship between rooting ability and 2nd year height at all three genetic leve ls (Table 4-1). The genetic co rrelation at the parental level between rooting ability and 2nd year height (GCABr ) was 0.32. At the full-sib family level, the genetic correlation between traits (FSBr ) was 0.39. The correlation of total genetic values of clones for rooting ability and 2nd year height (TGBr ) was 0.29. Previous studies have also reported positiv e correlations between rooting traits and growth. For example, Paul et al. (1993) reported that 1st year and 5th year height growth had strong genetic correlations with rooti ng (0.61 and 0.69, respectively) in western hemlock clones (Tsuga heterophylla (Raf.) Sarg.). In another study, Foster et al. (1985) reported that the genetic correlation between rooting ability and growth of western hemlock clones in a growth chamber was 0.37. A weak, but positive, relationship between the number of roots on loblolly pine cuttings and subsequent field growth has been reported (Foster et al. 2000). On the other hand, Goldfarb et al. (1998) reported that the phenotypic correlation between the number of roots and first year field growth in loblolly pine rooted cu ttings was negligible. C effects relating to the season in which th e cuttings were set has previously been discussed in this population (Cha pter 3). Further evidence of the presence of C effects can be seen from the genetic correlations between the total genetic value of clones for rooting ability and 2nd year height based on analyses i nvolving a single rooting and field

PAGE 78

67 trial (Table 4-2). Rooted cu ttings planted in each field trial originated from a single sticking date, and the high est genetic correlations (TGBr ) occurred between traits from these trials. For example, rooted cuttings from the Winter02 rooting trial were planted only in the Field A trial. The total genetic correlation between rooting ability from Winter02 and 2nd year height from Field A was 0.24, while 2nd year height from Field A had much lower correlations with rooting ab ility from any of the other sticking dates Table 4-1. Means, variance component estimat es, heritabilities, and genetic correlations from the bivariate analysis of rooting ability and 2nd year height. Standard errors are given in parentheses. Rooting Ability 2nd Year Height x 42 % 210 cm AV 0.0135 (0.004) 399.3 (119.6) DV 0.0014 (0.002) 77.3 (39.6) IV 0.0085 (0.002) -94.9 (63.7) GV 0.0235 (0.002) 381.7 (59.9) PV 0.2299 (0.002) 1836 (61.0) HSPV 0.0038 (0.001) 108.1 (29.6) FSPV 0.0085 (0.002) 231.1 (59.1) CLPV 0.0286 (0.002) 434.1 (59.9) 2h 0.059 (0.02) 0.22 (0.06) 2 H 0.102 (0.01) 0.21 (0.03) 2HSH 0.88 (0.04) 0.92 (0.03) 2FSH 0.84 (0.04) 0.95 (0.01) 2CLH 0.82 (0.01) 0.88 (0.02) GCABr 0.32 (0.19) FSBr 0.39 (0.17) TGBr 0.29 (0.08)

PAGE 79

68 (Table 4-2). This indicates that the vigor of the hedge at the time of rooting may have influenced rooting and subsequent field growth. The biological significance of positive correlations between rooting and 2nd year height in the current stud y is unclear. Goldfarb et al. (1998) reported that root morphological traits were not meaningful for subsequent field growth after one year of growth for loblolly pine r ooted cuttings. However, it has generally been acknowledged that stock plant vigor is related to rooting. Perhaps the vigor of the stock plant is an indicator of the metabolic activity of cutt ings during root initiation. The more metabolically active genotypes may have rooted quicker and formed be tter root systems. In the current study, rooting was assessed at nine to eleven weeks, and selection may have indirectly been for rate of rooting as opposed to strictly rooti ng ability. Clones that rooted quickly during the roo ting period may have had increased metabolic activity. A positive genetic correlation between rooting a nd growth implies that some of the same genes are responsible for the expression of each trait. There may be genes in common for rooting and height relating to metabolic ac tivity. Genotypes that have a tendency for increased rates of metabolic activity may root at higher frequencies and grow larger than genotypes that do not ha ve these alleles. Genetic Gain The importance of selecting for rooting ab ility in a clonal fore stry program based on rooted cuttings has long been recognized (Foster et al. 1984; Foster 1990; Baltunis et al. 2005). Genetic gains in field traits will not be realized if the clones can not be propagated efficiently for deployment. Therefor e, increases in rooting ability could have broad economic impacts on a clona l forestry program (Foster et al. 1984). Positively correlated traits imply that se lection for one trait should also lead to improvement in the

PAGE 80

69 second trait. In the case of rooting ability and growth, positive genetic correlations can lead to substantial gains for both traits in a clonal forestry program based on rooted cuttings. The genetic gains in rooting ability and 2nd year height based on BLUP values were compared for a number of deployment strategies Selecting the top half-sib family for rooting ability would result in a gain of 36% in rooting abili ty (Figure 4-1). Deployment of the top half-sib family selected for roo ting ability would result in a genetic gain of 5.4% in 2nd year total height (Figure 4-2). Select ing the highest ranking half-sib family for 2nd year height would result in a gain of 14.8% and 8.1% in rooting ability and 2nd year height, respectively (Figur e 4-1 and Figure 4-2). The ge netic gain predictions using theoretical calculations were slightly lower than the gain predictions based on BLUP values. The gain in rooting ability associated with selecting the best half-sib family for rooting was 26.7% (using Equation 4-19). The gain in height by selecting the top growing half-sib family was 9.4% (using Equation 4-19). Slightly higher gains can be achieved by selecting the top full-sib family. Propagation of the top full-sib family selected for rooting ability results in genetic gains in rooting ability of about 43% over the population mean (Fig ure 4-1). Gains of nearly 9% in 2nd year total height could be expected by deploying the top rooting full-sib family (Figure 4-2). If the best gr owing full-sib family was propagated and deployed, then the genetic gains in rooting ab ility (Figure 4-1) and 2nd year height (Figure 4-2) would be 8.6% and 10.1%, respectively. The theore tical gain in rooting ability for full-sib selection was equivalent to the estimate base d on BLUP values. However, the theoretical

PAGE 81

70 gain for full-sib selection for 2nd year height was 16.3% which was higher than the estimate based on BLUPs. Table 4-2. The total genetic correlation (TGBr ) between rooting ability and 2nd year height from analyses of a single rooting trial a nd field trial. Shad ed values indicate the rooting trial in which cuttings originated from for their respective field trials. Standard errors are given in parentheses. Field A Field B Field C Field D Field E Field F Spring01 0.002 (0.07) 0.17 (0.07) 0.17 (0.07) 0.21 (0.07) 0.16 (0.07) 0.25 (0.07) Summer01 0.08 (0.06) 0.07 (0.07) 0.17 (0.07) 0.21 (0.07) 0.14 (0.07) 0.12 (0.07) Winter02 0.24 (0.07) 0.16 (0.07) 0.23 (0.07) 0.27 (0.08) 0.24 (0.08) 0.18 (0.08) Spring02 0.06 (0.07) 0.25 (0.07) 0.26 (0.08) 0.34 (0.07) 0.16 (0.07) 0.24 (0.08) Summer02 0.10 (0.07) 0.21 (0.07) 0.16 (0.08) 0.21 (0.08) 0.13 (0.08) 0.28 (0.08) Clearly, the most gain for either trait woul d be achieved by selecting the single best clone (Figure 4-1 and Figure 4-2). For inst ance, by selecting the top clone for rooting ability, the genetic gain based on clonal predicted values in rooting was 96% (Figure 4-1) indicating that approxim ately 82% of the ramets would r oot from this clone (42% mean rooting plus 0.96 x 42% = 82%). However, selecting the top roo ting clone from this population would result in a decreas e (genetic loss) in overall 2nd year height (Figure 42). On the other hand, se lecting the top clone for 2nd year height would result in a gain of nearly 27% in 2nd year height (Figure 4-2) and 43% in rooting ability (Figure 4-1) over the population mean for these traits. Genetic gain in rooting ability based on theoretical

PAGE 82

71 calculations was greater than those based on BLUP values for selection of the single best clone and was 118%, while the gain in 2nd year height for the top growing clone was 31.3% (using Equation 4-21). 0 10 20 30 40 50 60 70 80 90 100 Best Half-Sib Family Best Full-Sib Family Best Clone from Best 10 Full-Sib Families Best Clone% Gain in Rooting Ability % Gain in Rooting Ability when Selecting for 2nd Year Height % Gain in Rooting Ability when Selecting for Rooting Ability Figure 4-1. The genetic ga in in rooting ability (%) over the population mean for deployment of the best half-sib family, full-sib family, best clone from the best ten full-sib families, and the single best clone when selecting for rooting ability or 2nd year height. In order to address genetic diversity issues and the risk associated with deploying a single clone, a second clonal deployment opti on was considered by selecting the best clone from each of the ten highest ranking fullsib families. This strategy resulted in genetic gains nearly double th at of the full-sib family deployment strategy. When the trait selected is rooting ability, the best clone in the top ten full-sib families would result in gains of 77% in rooting abil ity (Figure 4-1) and 5.6% in 2nd year height (Figure 4-2).

PAGE 83

72 When the trait selected is for 2nd year height, then the genetic gains would be 37.6% in rooting ability (Figure 4-1) and 18.3% for 2nd year height (Figure 4-2). -15 -10 -5 0 5 10 15 20 25 30 Best Half-Sib Family Best Full-Sib Family Best Clone from Best 10 Full-Sib Families Best Clone% Gain in 2nd Year Height % Gain in 2nd Year Height when Selecting for Rooting Ability % Gain in 2nd Year Height when Selecting for 2nd Year Height Figure 4-2. The genetic gain in 2nd year height (%) over the population mean for deployment of the best half-sib family, full-sib family, best clone from the best ten full-sib families, and the single best clone when selecting for rooting ability or 2nd year height. Deployment of well-tested clones can result in genetic gains in both rooting ability and 2nd year total height. Both tr aits should be considered fo r a successful loblolly pine clonal forestry program based on rooted cut ting technology. The re sponses to selection when considering both rooting ability and 2nd year height were compared for a number of selection indices (Figure 4-3). The gain in rooting ability associated with selecting the top 10% of clones ranged from 23.8% when only 2nd year height was considered to 57.6% when only rooting ability was consid ered (Figure 4-3). Higher gains can be

PAGE 84

73 achieved by increasing the selection intensity When only the top 1% of clones was selected, the gain in rooting abi lity ranged from 33.4% when only 2nd year height was considered to 80.6% when only rooting ability was considered (Figure 4-3). The genetic gain in 2nd year height associated with deployment of the best 10% of clones ranged from 4.8% when only rooting ability was considered to 12.6% gain when only 2nd year height was considered (Figure 4-3). Similarly, genetic gain in 2nd year height ranged from 3.8% to 18.5% when the best 1% of cl ones were selected (Figure 4-3). Clonal forestry programs that consider multiple traits need to optimize their selection strategies for deployment populations. Arbitrarily setting the optimum selection weights on 2nd year height and rooting ability to 90% of the maximum genetic gain obtainable for a single trait, then the optim um selection weights can be compared. For example, if the top 10% of clones are select ed and 90% of the maximum gain in rooting ability is considered, then the optimum weights on 2nd year height and rooting ability correspond to weights of 0.7 and 0.3, respectiv ely. These weights result in genetic gains of 53.6% in rooting ability and 8.7% in 2nd year height (Figure 4-3) If 1% of the clones are selected, then the optimu m weights are 0.6 and 0.4 on 2nd year height and rooting ability, respectively, and correspond to gains of 75.9% in rooting ability and 11% in 2nd year height (Figure 4-3). If 90% of the maximum gain obtainable for 2nd year height is the criterion, then the optimum weights for both deployment options are 0.9 and 0.1 on 2nd year height and rooting ability, respectively. The genetic gains associated with selecting the top 10% of clones utilizing these weights results in gains of 40.4% in rooting ability and 11.9% in 2nd year height (Figure 4-3). If only the top 1% of clones are deployed using these selection

PAGE 85

74 weights, then genetic gains of 54.2% in rooting ability and 17.4% in 2nd year height are obtainable (Figure 4-3). The ma ximum change in genetic gain of rooting ability results from increasing the weight on rooting ability from 0 to 0.1. In fact, an additional 17-21% gain in rooting ability can be obtained in the deployment population by increasing the weight on rooting ability from 0 to 0.1. When 10% of the clones were selected ( 220), then the number of full-sib families represented in the deployment population ra nged from 38 to 45 depending on selection index (Table 4-3). When only the top 1% of clones were selected ( 22), the number of full sib families represented in the deployment popul ation varied from 9 to 15 (Table 4-3). Maximum genetic gains can be achieved by ignoring relatedness among selections. However, if no constraints are placed on the relatedness of selections, then there is a tendency to make many selections from the better families. Although the approximate average number of clones selected per full-si b family was five when 10% of the clones were selected, nearly 45% of these selections came from only five families when only rooting ability was selected. Similar trends ca n be seen when selecting 1% of the clones. For example, when all of the selection wei ght is on rooting ability, then there was an average of 2.44 clones per full-sib family sel ected (Table 4-3). However, 15 of the 22 selections came from three families. On the other hand, when considering the number of half-sib families represented in the sele cted population, 26 to 30 out of 32 possible parents are represented when 10% of the clone s are selected. Thirteen to 17 parents are represented in the deployment population when 1% of the clones are selected (Table 4-3). Therefore, at least in this population, ther e may be sufficient gene tic diversity in the deployment population even with higher selection intensities. Isik et al. (2005) reported

PAGE 86

75 genetic gains in growth n ear 30% for loblolly clones from a different population irregardless of any restrictions on the relatedness of selections. Conclusion Rooting ability and 2nd year height are he ritable traits in lo blolly pine, and both traits showed substantial clonal variation. There was a positive genetic correlation between rooting and height at the parental, full-sib family and clonal levels. Genetic gains in rooting ability and 2nd year height are possible as demonstrated by a number of deployment strategies. For hard-to-root spec ies, like loblolly pi ne, a successful clonal 0% 10% 20% 30% 40% 50% 60% 70% 80% 90%1|0 0.9|0.1 0.8|0.2 0.7|0.3 0.6|0.4 0 .5|0.5 0 .4|0.6 0.3 |0.7 0.2 |0.8 0.1| 0.9 0|1Weight on 2nd Year Height | Weight on Rooting Ability% Gain % Gain in 2nd Year Height (10%) % Gain in 2nd Year Height (1%) % Gain in Rooting Ability (10%) % Gain in Rooting Ability (1%) Figure 4-3. Responses to sele ction in rooting ability and 2nd year height with various selection indices for two clonal deployme nt options: 10% of clones selected and 1% of clones selected.

PAGE 87

76 forestry program based on rooted cuttings must consider both rooti ng ability and growth when making selections. Because here these tr aits are positively correlated, selection for one trait should lead to positive gains in the other. Or, early culling of clones based on poor rooting will not negatively effect select ion of clones for growth when field data becomes available. Clonal forestry is not a breeding method to develop better genotypes. However, clonal forestry is a method to mass-produce we ll-tested genotypes. Short-term genetic gains may be maximized through deployment of well-tested clones, but long-term gains need to involve both clonal selection and recu rrent selection for additive genetic variation through repeated selection and breeding. Restrictions on th e relatedness of selections will be necessary when making selections fo r a breeding population or when deployment will be traditional zygotic seedlings from seed orchards in order to reduce detrimental effects of inbreeding depression.

PAGE 88

77 Table 4-3. The number of full-sib families (half-sib families) and average number of clones per full-sib family (half-sib family ) selected from selecting 10% or 1% of the top clones using the combined selection index. 10% Clones Selected 1% Clones Selected Weight on 2nd Year Height: Weight on Rooting Ability # FS Families (# HS families) Ave. # Clones per FS Family (HS Family) # FS Families (# HS families) Ave. # Clones per FS Family (HS Family) 1:0 38 (26) 5.8 (16.9) 14 (17) 1.6 (2.6) 0.9:0.1 41 (28) 5.3 (15.7) 15 (17) 1.5 (2.6) 0.8:0.2 43 (28) 5.1 (15.7) 12 (15) 1.8 (2.9) 0.7:0.3 43 (30) 5.1 (14.7) 13 (16) 1.7 (2.8) 0.6:0.4 45 (30) 4.9 (14.7) 10 (12) 2.2 (3.7) 0.5:0.5 45 (29) 4.9 (15.2) 12 (15) 1.8 (2.9) 0.4:0.6 45 (29) 4.9 (15.2) 11 (15) 2.0 (2.9) 0.3:0.7 45 (29) 4.9 (15.2) 11 (15) 2.0 (2.9) 0.2:0.8 45 (29) 4.9 (15.2) 10 (14) 2.2 (3.1) 0.1:0.9 45 (29) 4.9 (15.2) 9 (13) 2.4 (3.4) 0:1 45 (29) 4.9 (15.2) 9 (13) 2.4 (3.4)

PAGE 89

78 CHAPTER 5 CONCLUSION Loblolly pine is the most important commercial tree species in the southeastern United States. Several forest industries in the southeastern United States are deploying full-sib families of loblolly pine operationally. In addition, many of these companies are pursuing clonal forestry progr ams with loblolly pine. Clones need to be well-tested before they can be deployed. This invol ves the accumulation of reliable data from propagation, growth traits, dis ease resistance, etc. Those well-tested clones that meet the selection criteria can lead to substantial genetic gain. Several key results can be concluded from this research. With rooting data from 2,200 clones from 70 full-sib families, the curr ent study gives better es timates of genetic components of variance for rooting than severa l previous studies. These results show a great deal of genetic variation for rooting among families and clones of loblolly pine. Combined with moderate to high estimates of familyand clonal-mean heritabilities and type B correlations for rooting ability in different rooting trials, these results indicate the potential for increasing rooti ng efficiency by selecting good rooting families and clones or culling poor rooters. Field testing of clones is a necessary com ponent to clonal forestry programs based on rooted cuttings. In addition, genetic field trials establis hed with clones and seedlings from the same full-sib families provide an opportunity for comparing both half-sib and full-sib family performances for both propagul es. Based on the results in Chapter 3, several conclusions can be drawn. First, cl onally replicated seedli ng trials of loblolly

PAGE 90

79 pine provide genetic information with greater precision than zygotic seedlings. Second, genetic correlations between propagule types for the growth traits were highly favorable. These high genetic correlations between propagul e types reassure that parental and fullsib family rankings are stable regardless of propagule type. This implies that parental and full-sib family rankings based on existi ng seedling progeny trials could be used to select parents and families that perform well when they are deployed as rooted cuttings. Third, little genotype x envir onment interaction was observed across sites at the parental, family, and clonal level for all traits. Howeve r, there appears to be a carry-over effect relating to the season in which the cuttings were rooted for the clonal material. Finally, randomization is essential at all stages in testing when estimating genetic parameters. The lack of randomization for the seedling popu lation apparently resulted in a problem with partitioning of the genetic variance, causing full-sib families to appear more different and inflating estimates of dominance genetic variation. Rooting ability and 2nd year height are he ritable traits in lo blolly pine, and both traits showed substantial clonal variation. There was a positive genetic correlation between rooting and height at the parental, full-sib family and clonal levels. Genetic gains in rooting ability and 2nd year height are possible as demonstrated by a number of deployment strategies. For difficult-to-root species, like loblolly pine, a successful clonal forestry program based on rooted cuttings must consider both rooti ng ability and growth when making selections. Because here these tr aits are positively correlated, selection for one trait should lead to positive gains in the other. Or, early culling of clones based on poor rooting will not negatively affect select ion of clones for growth when field data become available.

PAGE 91

80 Clonal forestry is not a breeding method to develop better genotypes. However, clonal forestry is a method to mass-produce we ll-tested genotypes. Short-term genetic gains may be maximized through deployment of well-tested clones, but long-term gains need to involve both clonal selection and recu rrent selection for additive genetic variation through repeated selection and breeding. Restrictions on th e relatedness of selections will be necessary when making selections fo r a breeding population or when deployment will be traditional zygotic seedlings from seed orchards in order to reduce detrimental effects of inbreeding depression.

PAGE 92

APPENDIX A LOBLOLLY PINE PARTIAL DIALLEL MATING DESIGN. THIRTY-TWO PARENTS WERE CROSSED TO GE NERATE 70 FULL-SIB FAMILIES.

PAGE 93

82 1234567891011121314151617181920212223242526272829303132 1XXXX 2XXX XX 3XX X 4XX X 5XXXX 6XXX 7XX 8XXX 9XX 10X 11XX 12XXX 13 XX 14 XXXXX 15 XX 16 XX 17 X 18 X 19 20 XXX 21 XX 22 XXX 23 XX 24 X 25 XX 26 XX 27 X 28 XX 29 X 30XX 31 X 32 Parent 2Parent 1

PAGE 94

83 APPENDIX B VARIANCE COMPONENT ESTIMATES FOR ROOTING ABILITY Table B-1. Observed variance component es timates for rooting of loblolly pine stem cuttings from single-trial analyses. Spring01 Summer01 Wint er02 Spring02 Summer02 2TRAY 0.003977 0.005357 0.002891 0.010084 0.004796 2GCA 0.005018 0.004629 0.005383 0.004451 0.003817 2SCA 0.002683 0.000771 0.001363 0.001868 0.00187 2CLONE 0.041427 0.039236 0.024317 0.026364 0.023738 2REPxFAM 0.000836 0.000559 0.000637 0.000661 0.001043 2ERROR 0.190089 0.177991 0.204829 0.200698 0.142

PAGE 95

84Table B-2. Observed variance component estimates for rooting of lobl olly pine stem cuttings fr om pair-wise-trial analyses. Spring01 Summer01 Spring01 Winter02 Spring01 Spring02 Spring01 Summer02 Summer01 Winter02 Summer01 Spring02 Summer01 Summer02 Winter02 Spring02 Winter02 Summer02 Spring02 Summer02 2TRAY 0.004751 0.003637 0.007712 0.004474 0.004115 0.007929 0.005059 0.006905 0.003774 0.007824 2GCA 0.003434 0.002923 0.00433 0.003929 0.003191 0.002844 0.002557 0.003065 0.002347 0.003915 2SCA 0.001409 0.000417 0.001262 0.001876 0.00067 0.000654 0.001027 0.001107 0 0.001147 2CLONE 0.021605 0.011304 0.017015 0.014235 0.012645 0.014791 0.014151 0.011209 0.010850 0.013135 2TESTxGCA 0.001464 0.002034 0.000422 0.000688 0.001846 0.001380 0.001472 0.001609 0.002123 0.0004 2TESTxFAM 0.000302 0.001758 0.001299 0.000539 0.000396 0.000572 0.000029 0.000423 0.001480 0.000485 2TESTxCLONE 0.018862 0.021637 0.017991 0.022116 0.019564 0.018869 0.020504 0.014919 0.014083 0.012163 2REPxFAM 0.000674 0.000764 0.000703 0.000956 0.000598 0.000613 0.000797 0.000646 0.000825 0.000767 2ERROR 0.183321 0.198958 0.197306 0.162528 0.192458 0.192031 0.159335 0.202419 0.174684 0.177164

PAGE 96

85 Table B-3. Observed variance components for rooting ability from the across-trial analysis using all five rooting trials. Variance Component Estimate 2TRAY 0.005785 2GCA 0.002921 2SCA 0.001062 2CLONE 0.01638 2TESTxGCA 0.001364 2TESTxFAM 0.000691 2TESTxCLONE 0.017349 2REPxFAM 0.000713 2ERROR 0.184907

PAGE 97

86 APPENDIX C VARIANCE COMPONENT ESTIMATES FO R EARLY GROWTH TRAITS OF LOBLOLLY PINE CLONES AND SEED LINGS FROM THE SAME FULL-SIB FAMILIES Table C-1. Observed variance components fo r loblolly pine clones from the across-trial analyses of 1st year height, 2nd year height, height incr ement, and crown width. A separate error variance was modeled for each trial. 1st Year Height 2nd Year Height Height Increment Crown Width 2INC 53.4293 300.188 123.143 105.75 2GCA 17.4525 100.363 38.141 27.5703 2SCA 2.9992 21.601 7.5808 4.93 2CLONE 32.3599 157.931 59.3416 56.4437 2TESTxGCA 4.1225 13.3866 7.9053 5.9029 2TESTxFAM 1.2003 3.7766 1.2451 1.3288 2TESTxCLONE 22.4253 82.7545 29.0673 23.4036 2TxCxGCA 0.4069 4.5692 2.942 2.2437 2TxCxFAM 0.0000008 0.3873 0.0000005 1.1495 2TxCxCLONE 0.0000008 17.1395 14.179 10.7818 2REPxGCA 0.3275 1.0281 1.5638 0.6583 2REPxFAM 1.9242 5.511 1.5806 0.8574 2AERROR 512.142 1700.04 692.262 610.815 2BERROR 279.24 853.321 427.247 306.51 2CERROR 527.513 2061.14 928.903 843.627 2DERROR 264.114 1110.76 864.549 329.026 2EERROR 313.481 1124.11 682.649 423.413 2FERROR 123.318 804.457 527.064 296.22

PAGE 98

87 Table C-2. Observed variance components fo r loblolly pine seedlings from the acrosstrial analyses of 1st year height, 2nd year height, height increment, and crown width. A separate error variance was modeled for each trial. 1st Year Height 2nd Year Height Height Increment Crown Width 2GCA 3.6682 29.9688 12.9402 13.7566 2SCA 4.3654 20.8173 10.4732 5.0053 2TESTxGCA 1.5235 8.4371 8.6503 3.8956 2TESTxFAM 8.5693 25.1585 8.5047 7.9082 2TxCxGCA 0.0000007 0.0000004 0.0000002 0.5247 2TxCxFAM 0.6036 0.0000001 1.375 1.0372 2REPxGCA 0.0000001 0.0000001 0.0000001 1.2412 2REPxFAM 6.3223 55.867 26.1737 15.4417 2AERROR 599.229 1621.95 713.814 590.37 2BERROR 361.489 1104.24 567.689 325.987 2CERROR 508.95 2524.33 1246.59 1110.51 2DERROR 294.534 1356.8 1059.94 501.614 2EERROR 379.556 1766.73 1109.52 610.975 2FERROR 101.715 755.187 585.714 341.353

PAGE 99

88 APPENDIX D VARIANCE COMPONENT ESTIMATES FROM THE BIVARIATE ANALYSES OF ROOTING AND 2ND YEAR HEIGHT Table D-1. Observed variance component estimates from the bivariate analyses of rooting ability from Spring01 and 2nd year height from each of the field trials. Spring01 Field A Field B Field C Field D Field E Field F 2rootINC 0.0039 0.0039 0.004 0.0039 0.004 0.004 2heightINC 219.85 232.03 801.24 390.46 105.53 89.22 2rootGCA 0.0051 0.0052 0.0052 0.0051 0.0052 0.0051 rootheightGCA 0.0778 0.219 0.2515 0.3664 0.3268 0.1727 2heightGCA 93.85 77.71 134.98 134.81 144.59 98.15 2rootSCA 0.0027 0.0026 0.0026 0.0027 0.0026 0.0027 rootheightSCA 0.0654 -0.1556 -0.0148 0.0203 0.0815 0.1294 2heightSCA 28.34 3.7 8.13 28.18 43.77 30.08 2rootCLONE 0.0415 0.0415 0.0414 0.0415 0.0414 0.0414 rootheightCLONE -0.21 0.3412 0.4084 0.3997 0.2865 0.7351 2heightCLONE 386.08 191.46 250.52 230.24 426.68 207.67 2rootREPxFAM 0.0008 0.0008 0.0008 0.0008 0.0008 0.0008 2heightREPxFAM 0.0996 5.31 23.68 15.47 24.33 15.44 2rootERROR 0.1901 0.1901 0.1901 0.1901 0.1901 0.1901 2heightERROR 1641.17 879.86 2003.72 1127.51 1072.05 834.69

PAGE 100

89 Table D-2. Observed variance component estimates from the bivariate analyses of rooting ability fr om Summer01 and 2nd year height from each of the field trials. Summer01 Field A Field B Fiel d C Field D Field E Field F 2rootINC 0.0053 0.0053 0.0053 0.0053 0.0053 0.0053 2heightINC 220.06 232.4 801.8 389.37 105.55 89.37 2rootGCA 0.005 0.005 0.005 0.005 0.005 0.0049 rootheightGCA 0.243 0.1731 0.2933 0.3457 0.2579 0.0796 2heightGCA 98.15 79.12 135.89 138.56 147.16 98.3 2rootSCA 0.0007 0.0007 0.0007 0.0007 0.0007 0.0008 rootheightSCA 0.0397 0.0141 -0.0125 0.0078 0.0616 0.1556 2heightSCA 26.74 3.33 6.33 24.69 40.85 26.5 2rootCLONE 0.0393 0.0392 0.0393 0.0393 0.0392 0.0393 rootheightCLONE -0.0944 -0.0803 0.3137 0.4131 0.2729 0.2437 2heightCLONE 385.69 190.78 250.56 230.58 426.65 206.64 2rootREPxFAM 0.0006 0.0006 0.0006 0.0006 0.0006 0.0006 2heightREPxFAM 0.0757 5.3 23.92 15.69 24.4 15.71 2rootERROR 0.1779 0.1779 0.1779 0.1779 0.1779 0.1779 2heightERROR 1641.1 879.8 2003.35 1127.44 1072.03 834.0

PAGE 101

90 Table D-3. Observed variance component estimates from the bivariate analyses of rooting ability fr om Winter02 and 2nd year height from each of the field trials. Winter02 Field A Field B Fiel d C Field D Field E Field F 2rootINC 0.003 0.0029 0.0029 0.0029 0.0029 0.0029 2heightINC 220.76 232.19 803.25 391.48 106.31 89.66 2rootGCA 0.0052 0.0053 0.0053 0.0052 0.0052 0.0052 rootheightGCA 0.2043 0.1833 0.271 0.3 0.2017 0.0927 2heightGCA 102.17 78.17 134.82 136.55 149.43 101.99 2rootSCA 0.0014 0.0014 0.0014 0.0014 0.0014 0.0014 rootheightSCA 0.1209 0.0611 0.1015 0.1848 0.1924 0.1857 2heightSCA 28.81 4.03 7.75 27.65 44.14 26.48 2rootCLONE 0.0243 0.0243 0.0243 0.0243 0.0243 0.0243 rootheightCLONE 0.6345 0.1392 0.3584 0.3934 0.6888 0.3461 2heightCLONE 392.91 190.82 251.6 232.29 430.86 206.31 2rootREPxFAM 0.0006 0.0006 0.0006 0.0006 0.0006 0.0006 2heightREPxFAM 0.3872 5.3 23.62 15.68 24.36 15.49 2rootERROR 0.2048 0.2048 0.2048 0.2048 0.2048 0.2048 2heightERROR 1639.72 879.85 2003.09 1126.58 1071.65 834.4

PAGE 102

91 Table D-4. Observed variance component estimates from the bivariate analyses of rooting ability from Spring02 and 2nd year height from each of the field trials. Spring02 Field A Field B Field C Field D Field E Field F 2rootINC 0.0101 0.0997 0.01 0.01 0.01 0.01 2heightINC 219.97 232.63 803.37 391.27 106.4 88.85 2rootGCA 0.004 0.0041 0.004 0.004 0.004 0.0039 rootheightGCA 0.1176 0.1904 0.1485 0.2154 0.1198 0.0669 2heightGCA 97.42 78.53 136.8 140.76 148.65 97.39 2rootSCA 0.0018 0.0018 0.0018 0.0018 0.0019 0.0019 rootheightSCA 0.1254 0.1025 0.1448 0.242 0.3037 0.278 2heightSCA 28.64 3.41 6.99 26.41 41.76 32.59 2rootCLONE 0.0264 0.0265 0.0264 0.0265 0.0264 0.0264 rootheightCLONE -0.0583 0.4215 0.7019 0.8602 0.2998 0.5378 2heightCLONE 385.33 192.98 256.21 23.29 427.8 205.39 2rootREPxFAM 0.0007 0.0007 0.0007 0.0007 0.0007 0.0007 2heightREPxFAM 0 5.33 23.42 15.61 24.23 15.59 2rootERROR 0.2007 0.2007 0.2007 0.2007 0.2007 0.2007 2heightERROR 1641.16 879.54 2002.81 1127.07 1071.69 834.82

PAGE 103

92 Table D-5. Observed variance component estimates from the bivariate analyses of rooting ability fr om Summer02 and 2nd year height from each of the field trials. Summer02 Field A Field B Fiel d C Field D Field E Field F 2rootINC 0.0048 0.0048 0.0048 0.0048 0.0048 0.0048 2heightINC 219.68 232.5 802.51 390.45 105.24 88.7 2rootGCA 0.0037 0.0037 0.0037 0.0037 0.0037 0.0037 rootheightGCA 0.1006 0.2194 0.1656 0.2104 0.0913 0.1656 2heightGCA 96.96 78.41 135.51 138.45 145.59 99.21 2rootSCA 0.0017 0.0017 0.0017 0.0018 0.0017 0.0017 rootheightSCA 0.0439 0.0693 0.0831 0.101 0.2114 0.1497 2heightSCA 27.33 3.68 7.906 26.24 41.0 29.59 2rootCLONE 0.0238 0.0238 0.0238 0.0237 0.0238 0.0237 rootheightCLONE 0.2082 0.1956 0.2738 0.3435 0.2526 0.5838 2heightCLONE 386.1 190.66 249.78 229.44 426.75 206.54 2rootREPxFAM 0.001 0.001 0.001 0.001 0.001 2heightREPxFAM 0.1938 5.25 23.88 15.48 24.3 2rootERROR 0.142 0.142 0.142 0.142 0.142 2heightERROR 1641.0 879.82 2003.21 1127.52 1072.29

PAGE 104

93 Table D-6. Observed variance component estimates from the bivariate analysis of rooting ability using all fi ve rooting trials and 2nd year height using all six field trials. Variance Component Estimate 2rootINC 0.0058 2heightINC 331.45 2rootGCA 0.0034 rootheightGCA 0.1858 2heightGCA 99.83 2rootSCA 0.0004 rootheightSCA 0.1155 2heightSCA 19.33 2rootCLONE 0.0164 rootheightCLONE 0.3988 2heightCLONE 162.72 2rootTESTxFAM 0.0035 2heightTESTxFAM 30.74 2rootTESTxCLONE 0.0173 2heightTESTxCLONE 103.55 2rootREPxFAM 0.0007 2heightREPxFAM 12.1 2rootERROR 0.1849 2heightERROR 1308.39

PAGE 105

94 LIST OF REFERENCES Anderson, A.B., Frampton, L.J., and Weir, R.J. 1999. Shoot production and rooting ability of cuttings from juvenile greenhouse loblolly pine hedges. Transactions of the Illinois State Academy of Science. Volume 92(1 and 2): 1-14. Baltunis, B.S., Huber, D.A., White, T.L., Gold farb, B., and Stelzer, H.E. 2005. Genetic effects of rooting loblolly pine stem cutt ings from a partial diallel mating design. Can. J. For. Res. 35: 1098-1108. Banks, B.D., Mao, I.L., and Walter, J.P. 1985. Robustness of the restricted maximum likelihood estimator derived under normality as applied to data with skewed distributions. J. Dairy Sci. 68: 1785-1792. Borralho, N.M.G. and Kanowski, P.J. 1995. Correspondence of performance between genetically related clones and seedlings. Can. J. For. Res. 25: 500-506. Burdon, R.D. 1977. Genetic correla tion as a concept for studying genotypeenvironmentinteraction in forest tree breeding. Silvae Genet. 26: 168-175. Burdon, R.D. and Shelbourne, C.J.A. 1974. The use of vegetative propagules for obtaining genetic information. N.Z.J. For. Sci. 4: 418-425. Byram, T.D., Miller, L.G., and Raley, E.M. 2004. 52nd Progress Report of the Cooperative Forest Tree Improvement Progr am. Texas Forest Service, College Station, TX. 24 p. Cech, F.C. 1958. The vegetative propagation of Pinus taeda L. (loblolly pine). Ph.D. Dissertation. Texas A. & M., College Station, TX. Cockerham, C.C. 1954. An extension of the concept of partitioning hereditary variance for analysis of covariance among relatives when epistasis is present. Genetics 39: 859-882. Comstock, R.E. and Moll, R.H. 1963. Genotype-environment interactions. In: Statistical Genetics and Plant Breeding. Editors R.E. Hanson and H.F. Robinson. NAS-NRC Pub. 982, Washi ngton, DC. pp 53-93. Cooney, G. and Goldfarb, B. 1999. Effects of shearing height, pr uning intensity, shoot origin and family on shoot morphology and th eir effects on rooting of loblolly pine stem cuttings. In: Proceedings of the 25th Biennial Southern Forest Tree Improvement Conference. July 11-14 1999, New Orleans, LA. pp 52-53.

PAGE 106

95 Costa e Silva, J., Borralho, N.M.G., and Potts, B.M. 2004. Additive and non-additive genetic parameters from clonally re plicated and seedling progenies of Eucalyptus globules. Theor. Appl. Genet. 108: 1113-1119. Cotterill, P.P. and Dean, C.A. 1990. Succe ssful Tree Breeding with Index Selection. Center for Scientific and Industrial Re search Organization (CSIRO), Collingwood, Victoria, Australia. p. 80. Cunningham, M.W. 1986. Genetic variation in rooting ability of American sycamore cuttings. In: Proceedings of Research and Development Conference, Sept. 1986, Atlanta. Technical Associat ion of the Pulp and Paper Industry, Norcross, GA. pp. 1-6. Dempster, E.R. and Lerner, I.M. 1950. Her itability of threshold characters. Genetics 35: 212-236. De Souza, S.M., White, T.L., Hodge, G.R., a nd Schmidt, R.A. 1991. Genetic parameter estimates for greenhouse traits of slash pine artificially inoculated with fusiform rust fungus. For. Sci. 37: 836-848. Dickerson, G.E. 1962. Implications of genetic-environmental interaction in animal breeding. Animal Production 4: 47-64. Dickerson, G.E. 1969. Techniques for re search in quantitati ve animal genetics In: Techniques and Procedures in Animal Science Research. Amer. Soc. Anim. Sci., Albany, NY, pp. 36-79. Dieters, M.J. 1994. Inheritance of volume and rust resistance in slash pine. Ph.D. Dissertation,Univ. Florida, Gainesville, FL, 105 p. Dieters, M.J., Hodge, G.R., and White, T.L. 1996. Genetic parameter estimates for resistance to rust (Cronaritum quercuum) infection from full-sib tests of slash pine (Pinus elliottii), modelled as functions of rust incidence. Silvae Genet. 45: 235242. Falconer, D.S. and Mackay, T.F.C. 1996. Introduction to Quantitative Genetics. 4th Edition. Longman Group Ltd., Essex, England. 464 p. Farmer, R.E., Durst, J.T., Deng Shaotang, a nd Yang Jun-Tao. 1992. Effects of clones, primary ramets, and age of stock plants on tamarack rooting. Silvae Genet. 41: 2224. Farmer, R.E., Freitag, M., and Garlick, K. 1989. Genetic variance and C effects in balsam poplar rooting. Silvae Genet. 38: 62-65. Forest Biology Research Cooperative (F BRC) 2000. Study B: clonal biology and performance of elite genotypes of loblo lly and slash pine. FBRC Report # 13. Forest Biology Research Cooperative, Un iv. Florida, Gainesville, FL, 31 p.

PAGE 107

96 Forest Biology Research Cooperative (FBRC) 2003. Forest Biology Research Cooperative 7th Annual Report. FBRC Report # 23. Forest Biology Research Cooperative, Univ. Florida, Gainesville, FL, 100 p. Foster, G.S. 1978. Genetic variation in rooti ng stem cuttings from f our year old loblolly pine. Weyerhaeuser Co., Hot Springs AR. Tech. Rep. No. 042-3204/78/97. Foster, G.S. 1990. Genetic control of rooting ab ility of stem cuttings from loblolly pine. Can. J. For. Res. 20: 1361-1368. Foster, G.S. and Shaw, D.V. 1987. A tree improvement program to develop clones of loblolly pine for reforestation. In: Proceedings of 19th Southern Forest Tree Improvement Conference, College Station, TX. pp 17-21. Foster, G.S. and Shaw, D.V. 1988. Using cl onal replicates to expl ore genetic variation in a perennial plant species. Theor. Appl. Genet. 76: 788-794. Foster, G.S., Campbell, R.K., and Adams, W.T. 1984. Heritability, gain, and C effects in rooting of western hemlock cuttings. Can. J. For. Res. 14: 628-638. Foster, G.S., Campbell, R.K., and Adams, W. T. 1985. Clonal selection prospects in western hemlock combining rooting traits with juvenile height growth. Can. J. For. Res. 15: 488-493. Foster, G.S., Lambeth, C.C., and Greenwood, M.S. 1987. Growth of loblolly pine rooted cuttings compared with seedlings. Can. J. For. Res. 17: 157-164. Foster, G.S., Stelzer, H.E., and McRae, J.B. 2000. Loblolly pine cutting morphological traits: effects on rooting and field performance. New Forests 19: 291-306. Frampton, L.J., Jr. and Foster, G.S. 1993. Field testing vegetative propagules. In: Clonal Forestry I. Genetic and Biotechnology. Editors M.H. Ahuja and W.J. Libby. Springer-Verlag, Berlin, Germany. pp. 110-134. Frampton, L.J., Jr. and Huber, D.A. 1995. Clonal Variation in f our-year-old loblolly pine in coastal North Carolina. In: Proceedings of 23rd Southern Forest Tree Improvement Conference, Asheville, NC. June 20-22, 1995. pp. 254-264. Frampton, L.J., Jr., Goldfarb, B., and Surles, S.E. 1999. Nursery ro oting and growth of loblolly pine cuttings: effects of rooti ng solution and full-sib family. South. J. Appl. For. 23(2): 108-116. Frampton, L.J., Jr., Li, B., and Goldfarb, B. 2000. Early field growth of loblolly pine rooted cuttings and seedlings. S outh. J. Appl. For. 23: 108-115. Gezan, S., Huber, D., Medina, A., Parisi, L., and Powell, G. 2004. 46th Annual Progress Report. Cooperative Forest Genetics Res earch Program. Gainesville, FL. 30 p.

PAGE 108

97 Gilmour, A.R., Gogel, B.J., Cullis, B.R., Welham, S.J., and Thompson, R. 2002. ASReml User Guide Release 1.0. VSN International Ltd., Hemel Hempstead, HP1 1ES, UK. 267 p. Goldfarb, B., Surles, S.E., Thetford, M., and Blazich, F.A. 1998. Effects of root morphology on nursery and firstyear field growth of rooted cuttings of loblolly pine. South. J. Appl. For. 22(4): 231-234. Greenwood, M.S. and Weir, R.J. 1995. Genetic variation in rooting ability of loblolly pine cuttings: effects of auxin and fa mily on rooting by hypocotyl cuttings. Tree Physiology 15: 41-45. Grigsby, H.C. 1962. Propagation of loblolly pine by cuttings. Comb. Proc. Int. Plant Propag. Soc. 11: 33-35. Huber, D.A. 1993. Optimal mating designs and optimal techniques for analysis of quantitative traits in forest genetics. Ph .D. Dissertation, For. Res. Conserv., Univ. Florida, Gainesville, FL, 151 p. Huber, D.A., White, T.L., and Hodge, G.R. 1992. The efficiency of half-sib, half-diallel and circular mating designs in the estimati on of genetic parameters in forestry: a simulation. For. Sci. 38(4): 757-776. Huber, D.A., White, T.L., and Hodge, G.R. 1994. Variance component estimation techniques compared for two mating desi gns with forest genetic architecture through computer simulation. Theor. Appl. Genet. 88: 236-242. Isik, F., Li, B., and Frampton, L.J. 2003. Additive, dominance, and epistatic variance estimates from a replicated clonal test of loblolly pine. For. Sci. 49: 77-88. Isik, F., Li, B., Frampton, J., and Goldfarb, B. 2004. Efficiency of seedlings and rooted cuttings for testing and selection in Pinus taeda. For. Sci. 50: 44-53. Isik, F., Goldfarb, B., LeBude, A., Li, B., and McKeand, S. 2005. Predicted genetic gains and testing efficiency from two loblolly pine clonal trials. Can. J. For. Res. 35: 1754-1766. Jayawickrama, K.J.S. and Carson, M.J. 2000. A breeding strategy for the New Zealand radiata pine breeding cooperati ve. Silvae Genet. 49: 82-90. Kendall, M.G. and Stuart, A. 1963. The Advan ced Theory of Statistics. Vol. 1. Hafner, New York, 433 p. LeBude, A.V., Goldfarb, B., Blazich, F.A., Wi se, F.C., and Frampton, Jr., L.J. 2004. Mist, medium water potential, and cutting wa ter potential influen ce rooting of stem cuttings of loblolly pine. Tree Physiology 24: 823-831.

PAGE 109

98 Libby, W.J., and Jund, E. 1962. Variance a ssociated with cloning. Heredity 17: 533540. Lopes, U.V., Huber, D.A., and White, T.L. 2000. Comparison of methods for prediction of genetic gain from mass selection on bina ry threshold traits. Silvae Genet. 49: 50-56. Lush, J.L., Lamoreux, W.F., and Hazel, L.N. 1948. The heritability of resistance to death in the fowl. Poultr y Science 27(4): 375-388. Marino, T.M. 1982. Propagation of southern pines by cuttings. Comb. Proc. Int. Plant Propag. Soc. 31: 518-524. McKeand, S.E. and Bridgewater, F.W. 1998. A strategy for the th ird breeding cycle of loblolly pine in the south-eastern U.S. Silvae Genet. 47: 223-234. McKeand, S., Mullin, T., Byram, T., and Wh ite, T. 2003. Deployment of genetically improved loblolly and slash pines in the s outh. Journal of Forestry 101: 32-37. McRae, J.B., Stelzer, H.E., Foster, G.S., and Caldwell, T. 1993. Genetic results from a tree improvement program to develop clones of loblolly pine for reforestation. In: Proceedings of 22nd Southern Forest Tree Improveme nt Conference, Atlanta, GA. June 14-17, 1993. pp. 424-433. Mullin, T.J. and Park, Y.S. 1994. Estimating genetic gains from alternative breeding strategies for clonal forestry. Can. J. For. Res. 22: 14-23. Murthy, R. and Goldfarb, B. 2001. Effect of handling and water stress on water status and rooting of loblolly pine stem cuttings. New Forests 21: 217-230. Namkoong, G. 1979. Introduction to Quantitati ve Genetics in Forestry. USDA For. Serv. Tech. Bull. No. 1588. 342 p. Paul, A.D., Foster, G.S., and Le ster, D.T. 1993. Field perfor mance, C effects, and their relationship to initial rooting ability for western hemlock clones. Can. J. For. Res. 23: 1947-1952. Paul, A.D., Foster, G.S., Caldwell, T., and McRae, J. 1997. Trends in genetic and environmental parameters for height, diameter, and volume in a multilocation clonal study with loblolly pi ne. For. Sci. 43: 87-98. Reines, M. and Bamping, J.H. 1960. Seasona l rooting responses of slash and loblolly pine cuttings. Journal of Forestry 58: 646-647. Rockwood, D.L. and Goddard, R.E. 1973. Pred icted gains for fusiform rust resistance in slash pine. In: Proc. 12th Southern Forest Tree Improvement Conference, June 1213. Baton Rouge, LA. pp. 31-37.

PAGE 110

99 Roff, D.A. 1997. Evolutionary Quantitative Genetics. Chapman and Hall. New York, NY. 493 p. Rowe, D.B., Blazich, F.A., and Raper, C.D. 2002a. Nitrogen nutrition of hedged stock plants of loblolly pine. I. Tissue ni trogen concentrations and carbohydrate status. New Forests 24: 39-51. Rowe, D.B., Blazich, F.A., Goldfarb, B., a nd Wise, F.C. 2002b. Nitrogen nutrition of hedged stock plants of loblolly pine. II. Influence of carboh ydrate and nitrogen status on adventitious ro oting of stem cuttings. New Forests 24: 53-65. Shaw, D.V. and Hood, J.V. 1985. Maximizing gain per effort by us ing clonal replicates in genetic tests. Theor. Appl. Genet. 71: 392-399. Sohn, S.I. and Goddard, R.E. 1979. Influe nce of infection per cent on improvement of fusiform rust resistance in slash pine. Silvae Genet. 28: 173-180. Sorensen, F.C. and Campbell, R.K. 1980. Ge netic variation in r ootability of cuttings from one-year-old western hemlock seedli ngs. USDA For. Serv. Res. Note PNW352. Stelzer, H.E., Foster, G.S., Shaw, D.V., a nd McRae, J.B. 1998. Ten-year growth comparison between rooted cuttings and seed lings of loblolly pine. Can. J. For. Res. 28: 69-73. Stonecypher, R.W. and McCullough, R.B. 1986. Estimates of Additive and Nonadditive genetic variances from a clonal diallel of Douglas-fir Pseudotsuga mensiesii (Mirb.) Franco. In: Proceedings of the IUFRO Joint Meeting of the Working Parties for Breeding Theory, Pr ogeny Testing, and Seed Orchards, 13-17 Oct. 1986, Williamsburg, VA. North Carolina State University Industry Cooperative Tree Improvement Progr am, Raleigh, NC. pp. 211-227. Van Vleck, L.D. 1972. Estimation of heritabili ty of threshold charac ters. J. Dairy Sci. 55: 218-225. Weber, J. and Stelzer, H. 2002. Operati onal rooted cuttings in southern pines. In: National Proceedings: Forest and Conserva tion Nursery Associations 1999, 2000, and 2001. Technical coordinators R.K. Dumroese, L.E. Riley, and T.D. Landis. Proceedings RMRS-P-24. USDA Forest Service, Rocky Mountain Research Station, Ogden, Utah. pp. 91-92. Westfall, P.H. 1987. A comparison of va riance component estimates for arbitrary underlying distributions. J. Amer Stat. Assoc. 82(399): 866-874. Williams, E.R., Matheson, A.C., and Harwood, C.E. 2002. Experimental Design and Analysis for Tree Improvement. 2nd Edition. CSIRO Publishing, Collingwood, Victoria, Australia. 214 p.

PAGE 111

100 Wise, F.C. and Caldwell, T.D. 1994. Macr opropagation of conifers by stem cuttings. In: Applications of Vegetative Propagation in Forestry. Editors G.S. Foster and A.M. Diner. USDA Forest Service, S outhern Forest Experiment Station, New Orleans. Gen. Tech. Rep. SO-108. pp. 51-73. Yamada, Y. 1962. Genotype by environment interaction and genetic correlation of the same trait under different environmen ts. Jap. J. Genet. 37: 498-509.

PAGE 112

101 BIOGRAPHICAL SKETCH Brian Baltunis graduated from Southern Il linois University at Carbondale with a Bachelor of Science degree in forest resour ces management in 1995. He began graduate work at the University of Maine in August 1995 where he concentrated on forest genetics and tree improvement. He received a Master of Science degree in forestry from the University of Maine in December of 1997. For the next two and a half years, Brian worked at Boise Cascade Corporation in Loui siana where he served as the Breeding and Testing Coordinator in Boises tree improveme nt program for loblolly and slash pines. Brian began his doctoral resear ch at the University of Florida in August 2000.