Historic, archived document
Do not assume content reflects current
scientific knowledge, policies, or practices.
t, be
i:
pagel ie
~~)
Gos
~~) ~{C
PROCEEDINGS
18TH SOUTHERN FOREST TREE
IMPROVEMENT CONFERENCE
May 21-23 1985
LONG BEACH, MISSISSIPPI _
NE 3B | Bi bal “oe
¥
SPONSORED BY
THE SOUTHERN FOREST TREE IMPROVEMENT COMMITTEE
IN COOPERATION WITH
SOUTHERN FOREST EXPERIMENT STATION
USDA FOREST SERVICE
s
UNIVERSITY OF SOUTHERN MISSISSIPPI - GULF PARK
Schmidtling, R. C., and M. M. Griggs, eds.
Proceedings of the 18th Southern Forest Tree Improvement Conference.
1985 May 21-23; Long Beach, MS. 382 p.
45 papers are presented in four categories: Biotechnology, Conifer
Genetics, Seed Orchard
Management, and Hardwood
Genetics
Copies of this publication can be obtained
by writing to:
The National Technical Information Service
Springfield, VA 22161
PROCEEDINGS OF THE
EIGHTEENTH SOUTHERN FOREST TREE IMPROVEMENT CONFERENCE
May 21-23, 1985
Long Beach, Mississippi
Sponsored Publication No. 40 of the
Southern Forest Tree Improvement Committee
ny
>
a
a Ves aha
aint £41
‘ ite
' i.
CHOON ELE NS
Page
DEDICATION — PHILIP C. WAKELEY by O. O. Wells 1
KEYNOTE FOREWORD
Biotechnology and the Future of Forest Genetics Research
Stanley Ihe KLUGMAN se cvscceseccccccccscccsvescescecsscsesevesesece 2
BIOTECHNOLOGY I
Genetic Engineering in Forest Trees (invited paper)
1 Thomas Ledtg and Ronald ie SCA ETOlpiteleleyeicleleleieleieie ele'e creleie e's sie cieve 4
North Central Forest Experiment Station Biotechnology
Program—Application to Tree Improvement (invited paper)
Neil Dre INCIES OV Vetenciatstensholovorolelavokolanelole lois svelelere lero eielelelelielelovovevelevéeloleve eteve ele 14
Biotechnology and Forest Genetics: An Industry Perspective
(invited paper)
Inte Je DES 6 6 CBOSS SS BOO OO DECI CCE OIG COICO DIO OTR OCH RCECE Uo ee era 23
Testing and Deployment of Genetically Engineered Trees (invited paper)
(Chapter Outline for Bonga & Durzan, 2d Ed.)
W. elie IRDIZCP ISG OS OO COO OODODODODOD OD ODO ODOODOODOOOOCOODOOOOOOOOOG 30
BIOTECHNOLOGY II
DNA Transfer and Gene Expression in Loblolly Pine
Ronald R. Sederoff, Anne-Marie Stomp, W. Scott Chilton,
and Larry MOOR Slatercicleieletsielieisioloisleleleleisielcleisiocaleleleielel ore clevererelelcielevetelerecchelere 32
Use of Tissue Culture Techniques in a Hardwood Tree
Improvement Program
De We Einspahr and Sie Re WAN Vloterelelclevelevelicleleleleletelelolelelete el elerorelelelecelerelene 33
Tissue Culture of Sweetgum (Liquidambar styraciflua L.)
ike E. Sommer , He ILA Wetzstein and N. IGE Clevcvolioleveleleloiaiciel olereieneieretersievers 42
Vegetative Propagation of Scots Pine (Pinus sylvestris L.)
through Tissue Culture
H. Ges Sonta Tsat and He Hts HUANGececcccccccccccccccccccccccccccs 51
Micropropagation of Eucalyptus viminalis
M. W. Cunntnghan and R. Le MO Ciiieroteieioleleielevcieielerarslevoleleicleretelsheretcisieiorerete 56
Growth Changes in Loblolly Pine (Pinus taeda L.) Cell
“ Cultures in Response to Drought Stress
Hemel st NCWLON 15! 4 USCMRMANA ra Po (VAN BULICCNEN selereielelete s\e\s\s/e:vie wie eo O4
ata ts
BBQ
=f ae
‘irc ne
: } 7 7 oe es
2
fe
+ e#¢eaeee
La
Oe a
“¢
: we eS
ae
ee
eh
6o
i
i
oe nw e
aes 8
ey
pes
rire 3
2 eee
‘
‘ s =
a
a
cout eens ae a a 9
siacet aed Sats rae. ro:
aecall etanbal ee
aaesreseehdnamtaspeenemen he a4 6 aaeeiM ee *
: ae al
‘ ae “a
TPA? GUREG ae « ate
: : s)
sarvee nea 4-22 2 FE SAA ET Ot S © eee
a ee ; : *-e = Ai Pt i hes A = iota i
(at sits ois sant) ont +2698 Yo,
ria | ‘Hastert eget
«Re poo vanes ba qoliets: hs
ae et ee saved: ,. Sasser 4
pear eae eas ainie Sepak ede neds wap
i Alin}
eoast. ee 1y9na get 3 hTnaias sae io File ah
»
cn shee deed eh eenasseamednes emer eee aee es:
: {08 ‘BS. Po ee vi ‘pace ae]
re
»
ani giloldal ah udkeasnget-
+ dasa
J -_)
psx hoestbhy sh 2 ek 3 sipthdoa? »
$e eau Oe we i a” ae s=a8 viet rt fares A
jae - 2 i Lome ce =
(.7 apttioay +4 fetenbiny BEd) at
er
Anand ©. we
mie *
ou ts
{Le so ‘ re
es
Page
CONIFER GENETICS I
Techniques for Successful Artificial Regeration of Longleaf Pine
(invited paper)
Marc G. FROUIUS ADAM Crapetolovets tore ovekekaveWsieveleeve: eves telco eles ereveloie.e.evelelepele:e.enere 75
Longleaf Pine Tree Improvement in the Western Gulf Region
gE dD. Byram and We Je TE OW Clotolelexciololoretoletelalclelevelotelelcrovereleketorerevelevoeveielele 78
Polymorphic Isoenzymes from Megagamethophytes and Pollen of
Longleaf Pine: Characterization, Inheritance, and Use in
Analyses of Genetic Variation and Genotype Verification
Siete Em LED re of ei sVolelefelelereiololsisiclsielcle\ viele cles eleisiclele cioiele« csicecesiceces: OS
CONIFER GENETICS II
Genetic and Cultural Factors Affecting Growth Performance
of Slash Pine
G. L. Reighard, D. L. Rockwood, and C. W. Comeresesesececeeceee 100
Two-Stage Early Selection: A Method for Prioritization and
Weighting of Traits
Cheryl B. NDOT Cav averotercloreKe etevouche oleveverevetsicve ekekeieteieiorslelevelelere e-ere's sievents 107
Design Efficiencies with Planned and Unplanned Unbalance for
Estimating Heritability in Forestry
Barbara Gis MCCUE CH Ai aletatebetscecevetegaca\e.oveleveveleecetovedevs: eeveiexevelevevaveretetereteveceue 7,
Within-Tree Variation in Cortical Monoterpenes of Slash Pine
Susan We Kossuth and He David MUS Cletoieholctaveieteloneleioreiovenoleleerersicleiorcieleys 127
Field Performance of Loblolly Pine Tissue Culture Plantlets
L. John Frampton, Jr., Ralph L. Mott and Henry V. Amerson.....-.- 136
Parents Vs. Offspring Selection: A Case Study
Gary lige HOdgC.cccccccccccccccccccscvccccseceseceseeccccccccccce 145
Genetic Variation in Loblolly Pine Root Growth Potential
Le lie Dewalt, P. P. Feret, and R. Byte LOQA NGO GOOG OD OOOO OD OOOO CICLO E55
Response of Four Sources of Loblolly Pine to Soil Acidity Extremes
Michael Gé ORDO IMS SOCIO OO GO OCCO OOO OO OOCID OOOO CODD O COICO Doane 163
Dynamics of Improved Loblolly Pine Plantations and the
Implications for Modeling Growth of Improved Stands (winner of the
first Tony Squillace Award)
MELT LYM Aa Di fOVGmand sHArOLG sh au BUGKNGLEs elc\elsielo clei slcicielsicie clelcicisis) 10
iv
ms ‘ Py
, ; ®
an tS trad geen <o iheatitlan
Ros a 2. ;
a samt * ities se nuntesen ie mee
° rn
sotudt tiv ores ed aba
isan vee aa’ eewvpeno? ug eed mae eh ees
Sa eT, a ee h Se .
2o feliod ¢ is. wane cosh sen oer
Pit onl Jones ~eoopstiedatl ngage
Pike ar rsd ee 5 | ae (tated fis 9
saaywr a¢enst a® cieidet class enass ee
~ y
‘
2 Olt fwoud ert3oeiM stad
Par rekewse® soweve *\ ED ae = es it SERS. ow &
& frotys yea boduet A
Se
i : ene « . +m eale é eh eas 0 2s 20.8
oe Leet)
j rf
4 snitals aeoratqed tee bape’ CON >
rN yriketot gs erat . te
:
, ;
j one oe on ‘ . * cea nresennen ena A.
“4 SITS cies
. ‘ ” * chee a toe:
ga “oly a ht ak]
E F oe ‘: tomy RR 4
bute story f
e«teeteegeees hei owes vee eee
e ;
a [ntznoset twos soon ee
PPT 1: ae, @ ae A
La (eee cskees 864800 45 Pern —_. ely nw on
gouarted xtikioA: bie? 02 anh _shictage
= we oft ine soo tress
isis bestia te
Page
SEED ORCHARD MANAGEMENT
Estimating Pollen Contamination in Loblolly Pine Seed Orchards
by Pollen Trapping
MECHACLTMGReCenUOCd Gnd Terry “RUCKEY i. cee. clocccccecccccecicecccoes LAO
Supplemental Mass Pollination of Single Clone Orchards for the
Production of Southern Pine Hybrids
Donald R. Knezick, John E. Kuser, and Peter W. Garrett...eceeee 187
Theoretical Impact of Pollen Viability and Distribution of the Number
of Strobili to Use for Controlled Pollination in Loblolly Pine
D. L. Bramlett, F. E. Bridgewater, J. B. Jett, and
EN SMMMALICILCUGlalclelel ec lelele clclelele clcloicisieccwceicccceccocccsccoccesscoces 194
A Seven-Year-Old Ocala Sand Pine Seedling Seed Orchard
Ralph A. Lewts, Timothy LaFarge, and James L. McConnell......2. 204
Monitoring Coneworms with Pheromone Traps: A Valuable Pest
Detection Procedure for Use in Southern Pine Seed Orchards
JenGeaweatherby., G. Le DeBarr, and L. R. Barber. .sccccesccecese 208
Selection Potential for Coneworm and Seed Bug Resistance in
Loblolly Pine Seed Orchards
George R. Askew, Roy L. Hedden, and Gary DeBarr..cecccccseseeee 221
Predicting Loblolly Pine Seedling Performance from Seed Properties
Le late Shear and le (O}e- IqBG AO HOOOOOU OOO OUOO ODOC ODOODOOOUOOODOO 226
An Ocala Sand Pine Progeny Test Compared with a Seedling
Seed Orchard
Timothy LaFarge, Ralph Lewis, and James L. MCCoOnnelLesseeseceee 234
HARDWOOD GENETICS
Geographic Patterns of Variation in Growth of Sweetgum in East Texas
A. Be Stauder, Tks, and We de OW Cltotokeloreveteiotaieletetetetoleteleroteretotetereteters 240
Juvenile Growth Performance in A Provenance Test of Sweetgum
Kim C. Stetner, Bruce Bongarten, and Randall J. Roussedu....+-- 248
Genetic Parameters for Two Eastern Cottonwood Populations in the
Lower Mississippi Valley
Gs Sam HOS CReqeheieletelatatalevonaleiaketotenototeveleioielcielelelclekelisveretcleveletelelcicieielaicioveteie 258
Genetic Variation Among Open-Pollinated Families of Baldcypress
Seedlings Planted on Two Different Sites
Patricta Faulkner, Furcy Zeringue, and John Toliver..ecececceee 267
nen ee | sh Neh
RW CRy pe) a
ae an Atay aa
Be at oes a
a ya ake 2 ere ‘pine
iy ks Peay et wes ~ LE ry! a on
sept st wuint bh
4
‘i
P mH a ces - ey, ‘ ny ss
ERASE am Ace DA
Ni 4
hin ales ‘ucla 1 ce ern bit x yobs ane , Pe.
y ; <6 ate hele oe srneerye rye, aime Py
re) arr ine peeves kage al O:
ee el i
ent (hyed winiiw Bad pany awa ~“
ee" om ue ; \ wae bis he 7. : adaM ri
Ae len | ae ye «Ss al ne x — a ; = P %
a 7
’ eo | LURID & 3G twee = wat
ee ey <ean4 eae tev Ae
Saab vat i secs sey
eC 5 aan rsh a ie
e at X4,05 et oe
iz cd ate: va tae be vemos ” a
a “ rc cemsaliiine he iam aa iia ls
aH [rst
a seryatfan 39: oo
-_
ele
ae
vs et eh vie
(teas
Page
Biomass Characteristics of Sycamore Coppice Influenced by
Parentage and Type of Planting Stock
So B. Land , IBES 6 and [He B. RADE EE SO SOC OO ODE OO OSLO COCICOOCOTAE DUS
Predicted Genetic Gains Adjusted for Inbreeding for an
Eucalyptus grandis Seed Orchard
K. V. Reddy, D. L. Rockwood, C. W. Comer, and
GE HRMMESKUING Tsticlelelelelelelclelolalere’s\s eleleicicic ccc civicivciieccicicescocccesce, 203
Can Lateral Root Characteristics Be A Major Factor in
Assessing Seedling Quality
Paul P. KO VAITICLIUL Ce iaven eles aleleversietecioleelievereierarelellorelie rs Vererel 6\@ eveveroceveve lore. 6 ele love 290
CONIFER GENETICS III
Monoterpene Phenotypes in Loblolly Pine Populations: Natural
Selection Trends and Implications
A. E. Squillace, Harry R. Powers, Jr., and S. V. Kossuth...... 299
One-Quarter Century of Tree Improvement on National Forests
in the Southern Region
Robert N. IKGE CHETIS eratelarelaielanevcleiercteversland cis (havclevelele 610 oxclere 16! bia sie lalevelere 309
Third-Year Comparisons of Loblolly and Slash Pine Seed Sources for
Fusiform Rust Resistance and Growth Potential in North Central
Florida
John A. Pait, III, Lee Draper, Jr., and Robert A. Schmidt..... 314
Comparative Physiology of Loblolly Pine Seedlings from Seven
Geographic Sources as Related to Growth Rate
Bruce C. Bongarten, Robert O. Teskey, and Brett A. Boltze..... 321
Resistance to the Development of Pitch Canker in Open-Pollinated
Slash Pine Families
Gam OWCrCOR Me i LOCLD and Por NANG Idien EAUie «)elele/sielslelelelelcleleicicten OO
Growth Model Evaluation: South-Wide Loblolly Pine Seed Source Study
Fan H. KUNG occsccccvccccecvccscccccccevcceccececcescccccccccce 341
Evaluation of Slash Pine for Resistance to Pitch Canker
CramH em MCHAC Rw) sili pMAOCKWOOG.ANGNG Me) BLAKCSLEC si slelecicielcicieiiciere 51
Cold Tolerance Variation in Loblolly Pine Needles from Different
Branch Types, Families, and Environments
UL We Kolb and Ke GE IS HE es PROORCU PRCT TO CIO TO OEE Oe Te 358
vi
Sah
MP ae ee hee ic
1
aaa ae
ioe ae Re woo ae)
cA) ut
Page
POSTERS
Adventive Embryogenesis in Yellow-Poplar Tissue Cultures:
A Preliminary Report
See) MErKiLeninien HesSOnmer= <AanGiwGi.) li eBYOUM esis veisisicleclelcieieese ss 369
Differences in Seed Properties Among Reciprocal Crosses of
Loblolly Pine (A Preliminary Report)
MHOMAS Ol RELryUMANdan COAORCHH s | SNCAM wesle cisicleleicleslcloieicicicicicie cle ee ©6370
Variation in Fusiform Rust Stem Galls on Five- and Six-Year-Old
Slash Pines
Charles H. WAULIELTISHGID lalate cieloletclcletalcrciniete e cles cielecicleleis:eveletalercle « ble és Sy7/al
APPENDIX
SFTIC MEMDEES Ni pretetelelelelolelelelolels(ele) cleleleleis slelelolslclele/elelelclelelcielelcreielcleieis’s cle © « 373
List of Registrants, 18th SHH Chvetolevotelelcielelotevercvelelslevolcvetetolevelcletoieicie o cleicle 377
vii
4 a
i
+)
a
Dey .
2 +o %,%
ye @ w Mica é
, ”
be
¢
i
\ = :
‘ Vier chy a
weer een Cs tee oe 8.
oe
ie a
nt ca oer no y
Ae ono) 4 ewe 2a wee
Tele eer i
ae iy
: ‘h
f :
i 7 f Ne
These proceedings are dedicated to Philip C. Wakeley who died August 28, 1983, at
his home in Ithaca, New York. Phil was a charter member of the Southern Forest
Tree Improvement Committee and was best known to most of us for his work in
planning, establishing, and coordinating the Southwide Pine Seed Source Study. He
was very proud of the Southwide Study but actually, it was only a capstone on a
long career of silvicultural research with the Southern Forest Experiment Station.
He and his wife Christine came to New Orleans in 1924 and left 40 years later when
he retired. He worked part time for the Forest Service for about 5 years after
that, publishing the 30 year results of regeneration studies established by
himself, his Forest Service coworkers, and the Great Southern Lumber Company (now
Crown Zellerbach) near Bogalusa, Louisiana.
He was author or coauthor of well over a hundred publications and the most famous of
these, “Planting the Southern Pines" played an important part in the regeneration of
the southern pinery as we know it today.
It was my privilege to work with Phil for 2 years before his retirement. He was
truly dedicated to all aspects of his research from policy implications to the
finest detail, and was widely regarded as one of the best technical writers in the
profession. He received the USDA Superior Service Award and was a SAF Fellow. The
SAF honored him with the Barrington Moore Award for Biological Research in 1956.
Saves J
Cece
O. O. Wells
Southern Forest Experiment Station
1 USDA-Forest Service
Abstract: Biotechnology and the Future of Forest Genetics Research
1
Stanley L. Keuenane
Forest genetics and tree improvement activities have now become an integral
part of forest management in most parts of the U.S. In a sense our field of
science has become of age, and we are no longer immune from changing societal
pressures i.e., reduced research spending. In response to these changing
priorities at least in the Forest Service we are placing a greater emphasis on
basic genetics research. That is, research that would accelerate our under-
standing of the genetic variation found in trees and research that would
increase our ability to capture such variation in improved tree material.
These are among the major reasons for initiating a modest forest biotechnology
effort in forest genetics at this time. We hope to identify or develop an
appropriate array of techniques that will overcome or at least shorten the
time to identify, capture and incorporate those characteristics into Fueest
trees desired by our various user groups. Such a program to be successful and
useful must be supported by a broad based conventional forest genetics program.
There must be adequate known material to manipulate. We should continue our
ongoing evaluation of current programs as to usefulness; but priority research
in early selection, intraspecific and interspecific breedingydisease resistance
and elements of basic physiology should be continued. Care will need to be
taken to ensure that long-term conventional field evaluation studies are
adequately supported. We can expect the overall research program to be leaner
but with a sharper focus on the priority needs of the future.
Liitreetore Timber Management Research, U.S. Forest Service, Washington, D.C.
BIOTECHNOLOGY |
MODERATED BY DR. BART THIELGES
University of Kentucky
cu
GENETIC ENGINEERING IN FOREST TREES
F. Thomas Ledig and Ronald R. sederorr!?
Abstract.--Gene transfer, using recombinant DNA technology, can
be used to engineer new, improved trees in a fraction of the time
required by traditional breeding methods. Genetic engineering
requires isolation of genes, their multiplication in bacteria, their
transfer to tree cells, and regeneration of the transformed cells into
new trees. Success has already been achieved in cloning conifer genes
and in developing a transfer system, and several genes of potential
value to forestry have been isolated from bacteria. The inability to
regenerate conifers from transformed cells is the major remaining
barrier to application of genetic engineering in tree improvement.
Additional keywords: Agrobacterium tumefaciens, Pinus lambertiana,
Pinus taeda, Cronartium ribicola, genetic transformation, isozymes,
heterozygosity, microinjection, recombinant DNA, biotechnology.
INTRODUCTION
The long life and large size of trees have always been major barriers to
progress in forestry, especially in forest genetics and tree breeding. To
surmount these barriers, forest biologists tried to develop techniques to
enable early evaluation of growth and disease resistance and to shorten the
reproductive cycle (e.g., Kinloch and Comstock 1980, Ledig 1974). However,
recent advances in molecular biology offer entirely new possibilities for tree
improvement. Instead of devising techniques for early evaluation, it is now
possible to direct genetic changes while bypassing the sexual cycle, at least
in particular instances (Sederoff and Ledig 1985). Using new biotechnologies,
improvements in forest trees can conceivably be made on the same time scale as
those in agricultural crops, and the large size of trees, which presently
restricts selection intensity, poses no difficulties for technologies that
operate on the cellular or molecular level.
The new capability for biological manipulation using such tools as genetic
transformation, parasexual hybridization by fusion of protoplasts, and
multiplication of high value materials by cloning, have captured the public
imagination like few other scientific developments. Our concepts of life are
being changed as surely as they were by the public announcement of Darwin and
Wallace's theory of evolution. If it is necessary to identify the beginning
of the current revolution, then 1953 is a good candidate, when Watson and Crick
published their classic paper on the structure of DNA. Since that time,
knowledge of the genetic material and the ability to use that knowledge have
been accelerating. The new genetic tools are much more powerful than the ones
provided by Mendelism and its rediscoverers.
Application of the new technologies in forestry will require a major
research effort. Our ignorance of the genetics, physiology, and biochemistry
Dreeciiute of Forest Genetics, Pacific Southwest Forest and Range Experiment
Station, U.S.D.A. Forest Service, Berkeley, California 94701
4
’
ok SOK ene yi Lis cM
m WGC t PE Tet ey he
A MESES es
) Pan oy
\ 1 i
it
& i
7 +.
‘ F e
P if
ip $!
hts
~ ¢ "Ke |
« ;
Zi
- wt
i ik
#>
i
% ne
re .
> y ‘ 4
. Sa
y j
’
‘ x
: i ae * f
pias eb ea
awe nena) Oa, TRS
be a : Ae rf
b
ii a 5
_ : ‘ ry
VoISLST A Wee
me j
; rh
* ; le
- Lay ti F
Fev S ’ Dae as
fo cbep Sergead aaa. eget
tc aannn ee
ky 7 " eR |
yaweasn —
"le , ‘
> _
cre) i. a j ,
‘Dat: ah ry
4 r if
a) AY 7 Hl !
va a 7 ia loae eel rT
of forest trees, their pests and pathogens is as deep as our opportunities are
broad. Until recently, fundamental research, such as studies of biosynthetic
pathways, photosynthesis, and stress metabolism, had little application because
they only explained how things worked, without providing ways to modify the
genetic controls. Now, basic studies have greater utility because they provide
information that can be used to modify processes to advantage. Many aspects of
biology, physiology, pathology, and biochemistry, have been integrated by the
new genetics. To narrow the subject, we concentrated on the possibilities of
direct genetic manipulation of forest trees (i.e., genetic engineering) and the
research needed to apply the technology.
Genetic engineering implies directed genetic change in individuals, and
subsequently, in populations. Directed change is not new. During prehistory,
early agriculturists brought about desirable changes in plants and animals
despite little formal knowledge of genetics. With the discovery of the
statistical laws of inheritance in the nineteenth and twentieth centuries,
breeders accelerated the rate of change in agriculturally important plants and
animals. However, genetic engineering now implies manipulations at the
cellular or molecular level, and one of the most powerful tools of genetic
engineering is transformation, the ability to insert new genes. Transformation
provides both an applied tool and a method of studying the nature of the gene.
TRANSFORMATION OF FOREST TREES
The Process
The insertion of a gene into a new host, thereby genetically "transforming"
the host, has four components: a DNA fragment consisting of a single gene ora
small block of genes must be identified and isolated; the block must be
inserted into a vector where it is multiplied; the foreign DNA must be
transferred to the host cell where it is incorporated and expressed; and the
transformed cells must be regenerated into a plant. The process of
DNA-directed transformation was first discovered in bacteria (Avery, MacLeod,
and McCarty 1944), but in recent years has been applied to cells of higher
plants (Goldsbrough et al. 1983, Murai et al. 1983), such as sunflower
(Helianthus annuus L.) and tobacco (Nicotiana tabacum L.). Application to
trees is not completely straightforward because of some unique aspects of their
biology and because of the general lack of past effort in basic forest
research.
The Genes
Single-gene traits in trees. Few simply inherited traits are known in
forest trees, and most of these are of little or no economic interest. They
can be divided into four major classes: isozymes, major visible aberrations,
terpenes and other volatiles, and disease resistance factors. As many as 60
isozyme loci are known in some species (Conkle et al. 1982), but there seems to
be no advantage in transferring them. Aberrations, such as albinism and
dwarfism (Franklin 1970), are of negative value except perhaps for the
narrow-crowned phenotype, considered to be inherited as a single gene in some
European conifers (Karki 1983). Many volatiles are simply inherited and may
have potential in conferring resistance to insects (Smith 1966), and in some
cases, as valuable extractives. Disease resistance is the class of genes of
obvious value for transfer among trees (Kinloch, Parks, and Fowler 1970).
5
Research at the Institute of Forest Genetics is directed toward transfer of
genes that could improve yield or value of forest trees. One of these
objectives is the eventual isolation and transfer of the major gene for white
pine blister rust (Cronartium ribicola J.C. Fisch. ex Rabenh.) resistance from
resistant sugar pine (Pinus lambertiana Dougl.) to susceptible individuals and
species.
Determining the mode of inheritance of genetic characteristics (single
genic or polygenic) is especially difficult in forest trees because it is
plagued with one of the traditional barriers confronting forest genetics; i.e.,
the long generation time. One or two generations of crosses must be made to
demonstrate Mendelian segregation, and even then the simultaneous segregation
of genes with pleiotrophic effects may make it difficult to draw definitive
conclusions. Trees are among the most heterozygous of organisms (Hamrick
1979), so the genetic background in most species is highly heterogeneous,
obscuring the effects of segregation at individual loci.
On the other hand, conifers have some advantages for genetics. Many genes
code for enzymes that are active in the megagametophyte, the nutritive tissue
or "endosperm" of the seed. The megagametophyte is a haploid tissue that is
derived from one of the four cells produced by meiosis (e.g. Allen and Owens
1972). Segregation can be detected as variation among seed (megagametophytes)
from the same cone or from different cones on the same tree. A sample of
several megagametophytes will show a 1:1 ratio of allelic types ina
heterozygous individual. In classic Mendelian genetics, a 1:1 ratio is usually
demonstrated by a "test cross", but use of the conifer megagametophyte
eliminates the need for test-crossing (Conkle 1974). Therefore, for allozyme
loci, conifers provide the advantages of haplogenetics, pioneered in fungi such
as the bread mold (Neurospora crassa; Barratt et al. 1954).
Isolating genes. To isolate a gene, the DNA is cleaved with restriction
enzymes and the fragments are spliced into the DNA of a self-replicating virus
or plasmid, called a "vector", that infects bacterial cells. When the vector
with its foreign DNA infects a bacteria, the fragment is multiplied, or
"cloned", along with the vector's DNA. The colon bacteria (Escherichia coli)
is a common organism used to clone DNA fragments, and a collection of bacterial
colonies, each incorporating a different fragment, forms a "library" of the
donor's DNA. There is often no way to tell which fragment carries the gene of
interest unless a similar gene, previously isolated from another species, is
available to "probe" for it with DNA-DNA hybridization techniques.
Isolation of genes in conifers would be difficult even if genes worthy of
transfer were known. The conifer genome is very large, apparently 34.7 pg for
2C content in sugar pine (Dhillon 1980). By comparison, the genome of corn
(Zea mays L.) is only about 11 pg (Bennett 1972), which itself is large
compared to many animal species. The human genome is only 7.3 pg (Bachmannn
1972), and many insects have 2C contents that are another order of magnitude
smaller, around 0.2 pg for fruit flies (Sparrow, Price, and Underbrink 1972).
a a - Knowing where a gene is located is important if it is to
be isolated. Linkage maps for conifers are very incomplete, and no genes have
been associated with individual chromosomes. If genes could be identified to
chromosome, it might be possible to rapidly sort out specifie chromosomes with
dual laser flow sorters (Dickson 1985). The task of constructing a fragment
6
library for a single chromosome would be less than one-tenth as difficult as
constructing a library for the entire genome. Inserting entire chromosomes in
plant cells, rather than fragments, is another possibility (Malmberg and
Griesbach 1983), although aneuploids are unstable and usually aberrant in
conifers (Mergen 1958, 1959). However, even isolating a chromosome would be
difficult in forest trees, given present knowledge. For example, the 12
chromosomes of the haploid set that characterize most of the family Pinaceae
are scarcely distinguishable with conventional stains (e.g., Saylor 1961). In
most of the pines, spruces (Picea spp.), and firs (Abies spp.) only the
smallest, heterobrachial chromosome can be identified with confidence. The
others are all homobrachial and similar in size. Newer radiological and
staining techniques employed in human cytogenetics may be fruitful. Recently,
Hizume, Ohgiku, and Tanaka (1983) claimed to distinguish all of the chromosomes
of Austrian pine (Pinus nigra Arnold) with fluorescent banding, but very little
present research effort is focused on the conifer karyotype.
There is a chance of finding linkage between allozyme loci and genes
controlling gther characteristics, such as disease resistance. M.T. Conkle and
B.B. Kinloch™’ (personal communication) have already demonstrated loose
linkage (27 map units) between the major gene for blister rust resistance in
sugar pine and a 6-phosphogluconate dehydrogenase locus. For isozyme loci to
be really useful for isolating genes with unknown products, such as the gene
for blister rust resistance, the two must be very tightly linked. Linkage maps
are being constructed for several species in the Pinaceae (e.g., Conkle 1981).
Because of the apparently high degree of conservatism in evolution of the
conifer karyotype, linkage maps in one conifer are likely to approximate those
in others. The same linkages are repeated in the pines, firs, and spruces
investigated so far (e.g., Conkle 1981, King and Dancik 1983, Neale and Adams
1981).
Restriction site mapping using enzymes that cut the DNA at specific base
sequences, combined with isozyme mapping, would provide an extensive map in a
short time. The development of isozyme technology in conifers provided a rapid
means for chromosome mapping, but its utility is limited; only about 60 isozyme
marker loci are available. While 60 is a considerable number, especially
compared to virtually none 10 years ago, restriction fragment mapping could
expand the number of markers to hundreds. Recombinant DNA techniques do not
depend on expression of a gene; fragments can be assayed at any time. By
contrast, genes coding for enzymes, such as alcohol dehydrogenase, may be
expressed only during a restricted period of development or in certain tissues
(Conkle 1971). Furthermore, fragments need not include functional genes in
order to be valuable markers. Any fragment can be used that can be recognized
by its banding pattern in molecular hybridization analysis.
Genes in heterozygous combination. Of special interest is the relation
between heterozygosity and growth. In trees, growth and fitness are closely
related, and they are correlated with heterozygosity. The notion that vigor
and heterozygosity are related is not new; explanations for hybrid vigor, or
heterosis, go back at least to the work of East and Shull over three-quarters
of a century ago (Shull 1952) and was the subject of Lerner's (1954) classic
book, "Genetic Homeostasis", However, the development of enzyme
TP a ee Se
M.T. Conkle and B.B. Kinloch, Institute of Forest Genetics, U.S.D.A. Forest
Service, Berkeley, California
electrophoresis revealed variation of such proportions that it was difficult to
explain it all as a result of balancing selection; i.e., selection favoring
heterozygotes (Lewontin 1974). Nevertheless, in a variety of organisms,
including trees, growth and heterozygosity for isozyme loci are positively
correlated in natural populations, a new finding (Ledig, Guries, and Bonefeld
1983). These results must be extended to additional species and to controlled
environments, to determine their generality.
The newly found correlation between growth and heterozygosity in forest
trees raises several questions regarding the conduct of tree improvement
programs. In the initial stage of tree improvement programs, trees are
selected one per stand and interplanted as grafted clones in seed orchards.
Are realized gains from seed orchards the result of crossing among unrelated
parents, which would favor heterozygosity? If so, will gains from a second
generation of selection be much lower than expected? When trees are selected
in natural populations based om growth, do heterozygotes have a greater
probability of selection? Would a better scheme be to select and cross trees
that differ at the maximum possible number of loci to produce highly
heterozygous progeny?
Before the proper tree improvement strategy can be identified, research is
needed to determine whether all the isozyme loci have an effect on growth or
just specific genes. And, do all the loci involved have equal effects or are
some more important than others? In fact, do the isozyme loci themselves
control growth rate or are they simply linked to other, more important genes?
Heterosis could actually be the result of inferiority of homozygotes at linked
deleterious loci. Inbreeding depression is the converse of heterosis.
Transformation offers a way to investigate some of these questions. It
would be relatively simply to isolate alternative alleles of isozyme loci and
introduce them into a homozygous background to investigate the effect of
heterozygosity at single loci. Torrey pines (Pinus torreyana Parry ex Carr.)
are a prime target, because they seem to be completely homozygous (Ledig and
Conkle 1983). Of course, regeneration of conifers from transformed cells is
still a barrier to completing such a critical experiment.
Research needs. If genes are to be isolated from forest trees, forest
biologists will need several types of knowledge: an understanding of the
physiological and biochemical mechanisms of traits of interest, their mode of
inheritance, and the gene products involved. Molecular geneticists will profit
from better linkage maps of the conifer genome. Tight linkage with genes for
which probes are available would facilitate isolation of the right fragment.
At first, genetic engineering in forestry will rely on genes from other
organisms because of the paucity of economically important, single-gene traits
identified in tree species. In fact, forestry will benefit from the much
larger research effort in agriculture and medicine. Genes for insertion in
conifers or hardwoods can come from any living system, bacterial, fungal,
plant, or animal. Some candidates for transfer are herbicide resistance and
salt tolerance (Chaleff and Ray 1984, Le Rudulier et al. 1984). Incorporation
and expression must be investigated in tree species if forestry is to make use
of genes from other organisms. At present we know little about the structure
of the genome in tree species -- why do conifers have so much highly repeated
DNA? Do conifer genes have introns? What are the promoters like? DNA content
8
may vary among populations and individuals: is DNA content itself adaptive,
perhaps related to drought or cold hardiness? These questions are researchable
and should be attacked early in any program of genetic engineering.
eC -V
Cloning conifer DNA. Few difficulties are anticipated in cloning conifer
DNA. R.R. Sederoff and P.D. Hodgskiss have inserted two copies from a highly
repeated fraction of the sugar pine genome into the bacterial virus M13 and
multiplied them in the colon bacteria. They have sequenced segments of about
400 base pairs in length and will extend this in the near future. These DNA
clones will. be useful probes to determine where the sequence occurs in the
sugar pine genome and its homology with the highly repeated fraction in other
pines and more distantly related conifers.
The cro b - Transformation is being approached from two
directions: through the use of the crown gall bacterium (Agrobacterium
tumefaciens) and by direct microinjection. Crown gall is the most widely-used
system for transformation in higher plants (Barton and Chilton 1983). It
carries a loop of DNA, the Ti plasmid. In an infected plant, part of the
plasmid DNA takes up residence in a linear chromosome of the host. The plasmid
genes are faithfully transcribed by the host, resulting in production of
substances necessary for growth and reproduction of the bacterium. The plasmid
DNA has been mapped, and can be modified to carry foreign genes, providing a
means to transform selected host-plants. However, crown gall was not known to
infect pines, although it had been reported on firs, incense cedar (Calocedrus
decurrens [Torr.] Florin), and other conifers (de Cleene and de Lay 1976).
Within the last year, R.R. Sederoff, A. Stomp, L. Moore, and W.S. Chilton have
found a strain that will transfer and express genes from the crown gall
bacterium in loblolly pine (Pinus taeda L.).
Microinjection. Microinjection is a direct way of introducing DNA into
target cells, and has been used successfully in animal systems (Lo 1983). Very
fine needles are guided into isolated, suspension-cultured cells, using
micromanipulators. DNA is moved from the needle into the cell by altering the
charge. Either vectors, such as the Ti plasmid, or "raw" DNA fragments can be
"injected", There are still many technical difficulties in applying the
procedure to conifer cells. Primary among these is the difficulty of
penetrating the thick cell wall. It may be simpler to inject naked protoplasts
(i.e., cells whose walls have been stripped by a cellulase enzyme). However,
for most conifers it has not been possible to regenerate viable cell suspension
cultures from protoplasts, although Teasdale and Rugini (1983) were successful
with loblolly pine. And it is not at all certain that injected DNA will move
into the nucleus, be incorporated in the conifer genome, or if incorporated, be
expressed. D.E. Harry and M. Freeling of the University of California at
Berkeley are working on these problems in cooperation with the Institute of
Forest Genetics.
Sel a fe) - Transformation usually happens with low
frequency, so transformed cells must be selected from among a larger population
of untransformed cells. A common way to accomplish this is to engineer a
vector that will permit easy identification of cells in which it has
incorporated. An example is the use of the kanamycin-resistance gene (neomycin
phosphotransferase, or NPT) from the colon bacterium, which has been spliced
9
into several plant vectors. When plant cells are plated onto agar with G418,
an aminoglycoside antibiotic that can be inactivated by NPT, only those that
have been transformed (i.e., those that have incorporated and expressed the
gene) survive.
For some genes, like the major gene for resistance to white pine blister
rust, direct selection may be possible. Fungal mycelia invade sugar pine cells
in callus culture, and resistance is expressed on the cellular level by a
hypersensitive reaction (Diner, Mott, and Amerson 1984). Following
microinjection, cell lines could be multiplied, subdivided, and one replicate
challenged by the fungus to identify transformed lines.
The Final Step: From T C Je
The inability to regenerate whole plants from transformed cells is the
greatest barrier to genetic engineering of conifers: there is no guarantee
that research efforts will be rewarded in the near future. On the other hand,
whole plantlets have been regenerated from cell and tissue culture in some
hardwoods (Karnosky 1981). Several laboratories in the United States, Canada,
and other countries are attempting to induce somatic embryogenesis in pines and
Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco), so far without success.
Perhaps, the problem should be sidestepped rather than met head-on.
For example, the megagametophyte of conifers has some features that might
be used to circumvent the problem of regenerating plantlets from cells in
vitro. During an extended period of time, the megagametophyte is in a free
nuclear state (i.e., the nuclei are not separated by cell walls). Mitotic
divisions result in over a thousand nuclei before cell wall formation begins,
and some of these nuclei differentiate into eggs (e.g. Allen and Owens 1972).
If DNA fragments or vectors could be injected into the megagametophyte during
the free-nuclear stage, they would be unimpeded by cell walls and, hopefully,
incorporate in the conifer DNA at a high rate. The target is large, apparently
up to 0.7 mm for an egg cell alone in sugar pine (Haupt 1941). Judging by the
size of the free-nuclear megagametophyte in Douglas-fir (Allen and Owens 1972),
the free-nuclear gametophyte in sugar pine may be several millimeters long.
Injury from injection should not cause irreversible damage to the
megagametophyte; e.g., seed bugs sometimes penetrate the megagametophyte
without destroying it (Krugman and Koerber 1969). After differentiation of the
egg and fertilization, the ovule could follow its normal course of development
and mature an embryo. It may be better to use the system in this way rather
than attempt to force conifer cells to do something they do not normally do
(i.e., undergo somatic embryogenesis). However, research is needed to develop
techniques for the direct injection of megagametophytes through the cone scales
in such a way that the cone can continue its normal development.
Regeneration of plants from cell and callus cultures remains the most
critical need in forest research. Other barriers to genetic engineering
already show signs of cracking, but there have been no major breakthroughs in
conifer regeneration. Without the capability of producing trees from cell
culture, the full benefits of transformation will not be realized. The
inability to regenerate trees from cells or callus is not only a block to the
use of genetic engineering, it prevents forestry from making full use of the
products of conventional selection and hybridization. There are several
interspecific hybrids and some desirable intraspecific crosses that cannot be
10
economically multiplied, and mass cloning would be an especially valuable
technique. One research approach would be the intensive study of embryogenesis
to chart the path of normal development, providing a guide to the necessary
steps in vitro.
CONCLUSION
Recombinant DNA technologies will make it possible to modify trees on a
time scale comparable to that of annual crops. Furthermore, manipulations at
the cellular level will result in greater gains than previously possible by
eliminating the barrier posed by the large size of trees; as long as whole
plants had to be evaluated in the field, selection intensity could never be as
great for space-consuming trees as for relatively smaller agricultural plants.
Already there are indications that these technologies can be applied to
conifers and hardwoods. Within the last year it has been possible to
demonstrate the insertion of the Ti-plasmid from crown gall in pine and prove
gene expression. DNA cloning and sequencing techniques have worked as well on
conifers as on other plants. While there are still only a few valuable,
single-gene traits known in forest trees, there are many markers, and linkage
maps are being constructed for conifer genomes. The massive research effort in
medical and agricultural sciences will provide valuable genes for the genetic
engineering of trees just as it has provided the tools. However, forest
biology cannot rely entirely on research in sister sciences.
Research should proceed on four parallel lines: 1) the genetic system of
forest trees; 2) transfer systems; 3) the physiological and biochemical basis
of valuable traits; and 4) the developmental path leading to regeneration from
cell culture. Of these, work on the genetic system and transfer systems shows
signs of progress.
With respect to research on physiological processes and gene products, it
is time to stop treating tree growth, form, disease resistance, etc. as black
boxes. While traditional breeding using the metrics of quantitative genetics
has proved quite successful, more effort is needed to identify underlying
mechanisms for important processes, their genetic control, and gene products if
forest genetics is to realize its full potential.
Regeneration of trees from cell and tissue culture remains the major
barrier to progress in forest genetics, and not because of lack of effort.
Yet, there is no reason that this barrier too cannot be overcome. When it is,
forestry will reap enormous benefits.
LITERATURE CITED
Allen, G.S. and J.N. Owens. 1972. The life history of Douglas-fir. Environ.
Can., For. Serv., Ottawa. 139 pp.
Avery, 0.T., C.M. MacLeod, and M. McCarty. 1944. Studies on the chemical
nature of the substance inducing transformation of pneumococcal types.
J. Exper. Med. 79: 137-158.
Bachmann, K. 1972. Genome size in mammals. Chromosoma 37: 85-93.
Barratt, R.W., D. Newmeyer, D.D. Perkins, and L. Garnjobst. 1954. Map
construction in Neurospora crassa. Adv. Genet. 6: 1-93.
11
Barton, K.A. and M.D. Chilton. 1983. Agrobacterium Ti plasmids as vectors for
plant genetic engineering. Pages 527-539 in: Methods in Enzymology, vol.
101. Academic Press, N.Y.
Bennett, M.D. 1972. Nuclear DNA content and minimum generation time in
herbaceous plants. Proc. Roy. Soc. Lond., B 181: 109-135.
Chaleff, R.S. and T.B. Ray. 1984. Herbicide-resistant mutants from tobacco
eell cultures. Science 223: 1148-1151.
Cleene, M. de and J. de Lay. 1976. The host range of crown gall. Bot. Rev.
42: 389-466.
Conkle, M.T. 1971. Isozyme specificity during germination and early growth
of knobeone pine. For. Sci. 17: 494-498.
Conkle, M.T. 1981. Isozyme variation and linkage in six conifer species.
Pages 11-18 in: M.T. Conkle (tech. coord.), Proc. Symp. Isozymes No. Am.
For. Trees and For. Insects, Berkeley, CA July 1979. USDA For. Serv. Gen.
Tech. Rep. PSW-48.
Conkle, M.T., P.D Hodgskiss, L.B. Nunally and S.C. Hunter. 1982. Starch gel
electrophoresis of conifer seeds: a laboratory manual. USDA For. Serv.
Gen. Tech. Rep. PSW-64. 18 pp.
Dhillon, S.S. 1980. Nuclear volume, chromosome size and DNA content in three
species of Pinus. Cytologia 45: 555-560.
Dickson, S. 1985. National laboratories establishing human gene libraries.
Genet. Eng. News 5 (1): 16, 35.
Diner, A.M., R.L. Mott, and H.V. Amerson. 1984. Cultured cells of white pine
show genetic resistance to axenic blister rust hypae. Science 224:
407-408.
Franklin, E.C. 1970. Survey of mutant forms and inbreeding depression in
species of the family Pinaceae. USDA For. Serv. Res. Pap. SE~61. 21 pp.
Goldsbrough, P.B., S.J. Karcher, S.B. Gelvin, and B.A. Larkins. 1983.
Introduction of a zein gene into the Ti plasmid of Agrobacterium
tumefaciens. Pages 35-43 in: Plant molecular biology. Alan R. Liss, Inc.,
N.Y.
Hamrick, J.L. 1979. Genetic variation and longevity. Pages 84-133 in:
0.T. Solbrig, S. Jain, G.B. Johnson, and P.H. Raven (eds.), Topics in plant
population biology. Columbia Univ. Press, N.Y.
Haupt, A.W. 1941. Oogenesis and fertilization in Pinus lambertiana and P.
monophylla. Bot. Gaz. 102: 482-498.
Hizume, M., A. Ohgiku, and A. Tanaka. 1983. Chromosome banding in the genus
Pinus. I. Identification of chromosomes in P. nigra by fluorescent
banding method. Bot. Mag. Tokyo 96: 273-276.
Karki, L. 1983. Crop tree idiotypes and harvest index should be the basis of
selection in cultivated trees. Pages 26-27 in: Metsanjalostussaatio, Ann.
Rep. Found. For. Tree Breeding, Helsinki, Finl.
Karnosky, D.F. 1981. Potential for forest tree improvement via tissue
culture. BioScience 31: 113-120.
King, J.N. and B.P. Dancik. 1983. Inheritance and linkage of isozymes in
white spruce (Picea glauca). Can. J. Genet. Cytol. 25: 430-436.
Kinloch B.B. and M. Comstock. 1980. Cotyledon test for major gene resistance
to white pine blister rust in sugar pine. Can. J. Bot. 58: 1912-1914.
Kinloch, B.B., G.K. Parks and C.W. Fowler. 1970. White pine blister rust:
simply inherited resistance in sugar pine. Science 167: 193-195.
Krugman, S.L. and T.W. Koerber. 1969. Effect of cone feeding by
occidentalis on ponderosa pine seed development. For. Sci. 15: 104-111.
12
Ledig, F.T. (ed.). 1974. Toward the future forest: applying physiology and
genetics to the domestication of trees. Bull. No. 85, Sch. For. Environ.
Studies, Yale Univ., New Haven, Connecticut. 80 pp.
Ledig, F.T. and M.T. Conkle. 1983. Gene diversity and genetic structure in a
narrow endemic, Torrey pine (Pinus torreyana Parry ex Carr.). Evolution
37: 79-85.
Ledig, F.T., R.P. Guries, and B.A. Bonefeld. 1983. The relation of growth to
heterozygosity in pitch pine. Evolution 37: 1227-1238.
Lerner, I.M. 1954. Genetic homeostasis. Oliver and Boyd, Edinburgh-Lond.
134 pp.
Le Rudulier, D., A.R. Strom, A.M. Dandekar, L.T. Smith, and R.C. Valentine.
1984. Molecular biology of osmoregulation. Science 224: 1064-1068.
Lewontin, R.C. 1974. The genetic basis of evolutionary change. Columbia
Univ. Press, N.Y. - Lond. 346 pp. |
Lo, C.W. 1983. Transformation by ionophoretic microinjection of DNA:
multiple integrations without tandem insertions. Molec. Cell Biol. 3:
Malmberg, R.L. and R.J. Griesbach. 1983. Chromosomes from
protoplasts--isolation, fractionation, and uptake. Pages 195-201 in: T.
Kosuge, C.P. Meredith, and A. Hollaender (eds.), Genetic engineering of
plants: an agricultural perspective. Plenum Press, N.Y. = Lond.
Mergen, F. 1958. Natural polyploidy in slash pine. For. Sci. 4: 283-295.
Mergen, F. 1959. Colchicine-induced p[olyploidy in pines. J. For.
57: 180-190.
Murai, N., D.W. Sutton, M.G. Murray, J.L. Slightom, D.J. Merlo, N.A. Reichert,
C. Sengupta-Gopalan, C.A. Stock, R.F. Barker, J.D. Kemp, and T.C. Hall.
1983. Phaseolin gene from bean is expressed after transfer to sunflower
via tumor-inducing plasmid vectors. Science 222: 476-482.
Neale, D.B. and W.T. Adams. 1981. Inheritance of isozyme variations in seed
tissues of balsam fir (Abies balsamea). Can. J. Bot. 59: 1285-1291.
Saylor, L.C. 1961. A karyotypic analysis of selected species of Pinus.
Silvae Genet. 10: 77-84.
Sederoff, R.R. and F.T. Ledig. 1985. Increasing forest productivity and
value through biotechnology. Jn Weyerhaeuser For. Potentials Symp.,
Tacoma, Wash. (In press)
Shull, G.H. 1952. Beginnings of the heterosis concept. Pages 14-48 in:
J.W. Gowen (ed.), Heterosis. Iowa State Coll. Press, Ames.
Smith, R.H. 1966. Resin quality as a factor in the resistance of pines to
bark beetles. Pages 189-196 in: H.D. Gerhold (ed.), Breeding
pest-resistant trees. Pergamon Press, N.Y.
Sparrow, A.H., H.J. Price, and A.G. Underbrink. 1972. A survey of DNA
content per cell and per chromosome of prokaryotic and eukaryotic
organisms: some evolutionaty considerations. Pages 451-495 in: H.H. Smith
(ed.), Evolution of genetic systems. Gordon and Breach, N.Y.=-Lond.=-Paris.
Teasdale, R.D. and E. Rugini. 1983. Preparation of viable protoplasts from
suspension-cultured loblolly pine (Pinus taeda) cells and subsequent
regeneration to callus. Plant Cell Tissue Organ Culture 2: 253-261.
Watson, J.D. and F.H.C. Crick. 1953. Molecular structure of nucleic acids.
Nature 171: 737-738.
13
NORTH CENTRAL FOREST EXPERIMENT STATION BIOTECHNOLOGY PROGRAM—
APPLICATION TO TREE IMPROVEMENT
Neil D. Nelson
Abstract.--In 1983 the USDA Forest Service initiated a new
research program on the genetic engineering of forest trees.
One-half of this initiative is the Biotechnology Multiproject
Research Program of the North Central Forest Experiment Station
(NC). The NC Biotechnology Program is centered at Rhinelander,
Wisconsin, and also has scientists at St. Paul, Minnesota, and
Madison, Wisconsin. The Program has nationwide responsibilities
and cooperators at another Forest Experiment Station, five
universities, and three biotechnology companies. The overall
Program purpose is genetic tree improvement, complementing
conventional breeding technologies. Program structure, studies,
and early results are described.
Additional keywords: Genetic engineering, somaclonal, protoplast, |,
recombinant DNA, herbicide resistance, disease resistance, Populus.
In 1983 the USDA Forest Service initiated a research program on the
genetic engineering of forest trees, probably the first such major program in
the world. One-half of this initiative is the Biotechnology Multiproject
Research Program of the North Central Forest Experiment Station (NC). The NC
Biotechnology Program, formally organized in 1984, has a nationwide
responsibility. Centered at the Forestry Sciences Laboratory in Rhinelander,
Wisconsin, the Program has scientists at Rhinelander, and Madison, Wisconsin,
and St. Paul, Minnesota, and research cooperators at another Forest Service
Experiment Station, five universities, and three firms in the biotechnology
industry.
Two previous papers have described the NC Biotechnology Program (Nelson
and Haissig 1984, Nelson et al. 1984). My purposes in this paper are to des-
cribe and update Program:
° structure
. studies
.- early results
1/
=~ Program Leader, Biotechnology Program, North Central Forest Experiment
Station, USDA Forest Service, Forestry Sciences Laboratory, Rhinelander,
WI.
14
STRUCTURE
Strategic Plan
We recognized early that strong strategic planning was necessary for
success in biotechnology. In the formative stages, we collaborated with an
internationally known biotechnology consulting firm, L. William Teweles & Co.,
on a strategic plan for forest biotechnology research (Kidd 1984). A unique
research program evolved from that collaboration as well as from extensive
further analysis by Program managers and scientists.
The Program's strategic plan is based on the following postulates about
conventional tree breeding:
~The new biotechnologies and conventional tree breeding are complementary,
rather than competing, technologies. Both are essential components of success-
ful tree improvement. For example, resistance to a specific stress imparted
through biotechnological techniques is of little value when the trait resides
in an otherwise maladapted genotype. Improved and elite genotypes and popula-
tions resulting from conventional breeding programs provide the most desirable
starting material for further specific biotechnological improvement.
-One of the most important benefits of the new biotechnologies is the poten-
tial ability to reduce the long time periods required for tree improvement
using conventional genetic technology alone. Three recent major studies of bio-
technology (Burg et al. 1983, National Academy of Sciences 1983, Skelsey 1984)
have identified woody perennial crops as prime targets for genetic engineering
A major conclusion was that combining the new biotechnologies with conventional
breeding may produce relatively greater payoffs in forestry than in any other
agricultural area (Skelsey 1984), largely because of the time-saving
potential of biotechnology.
~The new bioiotechnologies provide potential means for introducing rare
or foreign critical traits into otherwise desirable forest tree germplasm. In-
troducing such traits into tree genomes may be impractical or impossible through
conventional breeding alone.
-The new biotechnologies may aid in the development of practical means for
capturing non-additive, as well as additive, genetic variation in improved tree
populations. The established biotechnology of micropropagation provides the
main vehicle for capturing such improvement. The new biotechnologies of soma-
clonal selection, somatic hybridization, and perhaps, microinjection and recom-
binant DNA may improve the frequency and reliability of whole plant regenera-
tion in micropropagation systems.
The NC Biotechnology Program is modeled after a typical startup biotech-
nology company, a unique organizational plan for a public research effort. The
Program is based on four essential synergistic factors Gelge 1), also present
in new private biotechnology firms:
- entrepreneurship
. Locus
15
- integration
. flexibility
FOUNDATIONS OF PROGRAM STRUCTURE
Entrepreneurship
Be biotech company model oS
. uniqueness
. shared resources/synergism with cooperators
. seed $, outside $, "follow-on" §
. prototype systems are “products”
|
. limit objectives eG ae » Common objectives i
. Vimit spectes + COmmon germplasm
. emphasize strengths of staff « AInterdisciplinarity |
andiiorgantzation . interface with i
. use elite germplasm } conventional genetics |
|
|
. easfly shift research emphasis
. greater reliance on research assoctlates
Integration
Figure 1. Principles of strategic planning and organization for the North
Central Forest Experiment Station Biotechnology Program.
Entrepreneurship for the NC Program involves the forementioned emulation
of biotechnology industry strategies. These strategic characteristics include
ensuring uniqueness for the research program, sharing resources and developing
a synergistic relationship with cooperators, and using “seed money” to help
attract outside research contracts and grants and further “follow-on invest—
ment.” In contrast to a private biotechnology firm, the NC Program produces
prototype genetic transformation systems for forest trees, rather than com-
mercial products.
Focus in the research program includes limiting the number of cbjectives
and target species. It involves carefully selecting species based on both
biological flexibility and commercial importance. The NC Biotechnology Program
has optimized the use of our limited resources by emphasizing research on
biologically amenable model species, while still maintaining research on some
commercially important species at a lower but meaningful level. Focus in the NC
Program also encompasses an emphasis on the strengths of the staff and parent
organization in choosing research objectives. This analysis of endogenous
strengths includes scientific, logistic, and financial considerations. An
important part of research focus in the NC Program is the use of elite
foundation stock from conventional genetics programs as starting material for
biotechnological improvement, as mentioned above.
16
Integration within the NC Program means ensuring commonality across the
Program, including all Forest Service and cooperating scientists. The common-
ality includes common objectives and common germplasm. Integration in the NC
Program also includes interdisciplinary research planning and execution and a
strong interface with conventional genetics and breeding efforts. The latter
involves not only the selection of elite and well-defined germplasm as experi-
mental material for the biotechnology research, but also the joint planning of
how the biotechnologically improved genotypes will be delivered to users. In
some cases the latter consideration may involve incorporating the improved
trait in seed. Classical genetic analysis is an important part of analyzing
and verifying the genetic effects of the new biotechnologies and constitutes
another conventional genetics-biotechnology interface within the NC Program.
Flexibility in the NC Program includes building in the willingness and
ability to shift research emphasis to follow up promising results. Research
in biotechnology is "high risk" in that positive results often cannot be pre-
dicted. The history of research in this area also includes the common occur-
rence of unexpected results and important spinoff applications. A biotech-
nology research program must be flexible to capitalize on this situation.
Another component of flexibility is a greater reliance on research associates
and other employees on temporary appointments than has been common in Forest
Service research.
Objectives
The NC Biotechnology Program is using established biotechnologies (con-
ventional genetics and breeding, tissue culture) and new biotechnologies
(somaclonal/gametoclonal selection, somatic hybridization, recombinant DNA)
to accomplish three objectives:
- impart herbicide resistance to selected forest trees
- impart disease resistance to selected forest trees
- develop genetic guidelines for the regeneration of selected
forest trees in tissue culture (regeneration genes)
The rationale for the choice of these objectives is fully explained in
two previous papers (Nelson and Haissig 1984, Nelson et al. 1984). Current
research is on the technologies of somaclonal selection, tissue culture, and
breeding. We expect to gradually increase our commitment to somatic hybrid-
ization and recombinant DNA.
Species
The NC Program is working with some species selected primarily for biol-
ogical reasons (models); some chosen primarily because they are commercially
important, and one species selected for both biological (model) characteristics
and potential commercial importance:
Model Model /Commercial Commercial
Populus spp. Pinus banksiana Pinus taeda
Larix spp. ; Pinus resinosa
17
About 75 percent of current Program research is on Populus.
Poplars were selected as our primary experimental species because they are
highly amenable to regeneration in a variety of tissue culture systems, can be
easily vegetatively propagated, are the subject of ongoing breeding programs,
are diploid, have a small genome size (Dhillon et al. 1984), are genetically
variable, and have ample background information in g genetics, physiology, and
plantation silviculture. Because of these characteristics, the USDA Forest
Service Genetic Engineering Workshop (F. T. Ledig, personal communication)
recommended poplars as model species to include in forest tree biotechnology
programs.
Staff
The NC Program currently includes three plant physiologists, a plant
geneticist, a plant anatomist, a tissue culturist, and two plant pathologists
(fig. 2). The Program Leader and Research Work Unit (RWU) NC-1403 are in
PROGRAM LEADER
IN. Nelson]
FPL-2202 NC-1403 NC-2205
K. Wolter-PL B. Haissig -PL D. Skilling
N. Nelson M.Ostry
D. Riemenschneider
R. Cecich
Tissue Culturist
Figure 2. Organizational structure and in-house scientific staffing of North
North Central Forest Experiment Station Biotechnology Program. Numbers are
Research Work Units (RWU). NC-1403 is the Program core unit. PL is Project
Leader of RWU. Arrow indicates research contribution to the core unit.
18
Rhinelander, Wisconsin; RWU FPL~2202 is in Madison, Wisconsin; and RWU NC~2205
is in St. Paul, Minnesota. NC-1403 is responsible for Program research on
herbicide resistance and regeneration genes. FPL-2202 contributes tissue cul-
ture research to the work of NC-1403. NC-2205 is responsible for Program re-
search on disease resistance.
Cooperators
Program cooperators, listed in table 1, work closely with the Forest Serv-
ice scientists in the Program, using germplasm that is common across the Program.
This planned research focus and coordination have created a synergistic
organization.
STUDIES
The current studies of the NC Biotechnology Program are listed in table 2.
Studies planned to begin soon are listed in table 3.
Table 1.--Research cooperators of the North Central Forest Experiment Station Biotechnology
Program.
Tie i eA eS
Government Universit Biotech. Industr
Institution Investigator Institution Investigator Institution Investigator
ee ee eee
Southern Forest Wisconsin B.McCown L.Williams ,
Exp. Station O.Wells Teweles & Co. G.Kidd
Minnesota P.Read
W.Hackett DNAP M.Sondahl
N.C. State S.Dhillon Calgene M.Moloney |
(J.Miksche) JFillatti
Michigan Tech. D.Karnosky
A.Diner
New Hampshire S.Minocha
a
EARLY RESULTS
Most studies of the NC Biotechnology Program were initiated less than 1
year ago. Nevertheless, Program scientists have obtained a number of signifi-
cant results. Some of the most noteworthy findings are listed below:
. A strategic plan for forest biotechnology was developed (Kidd 1984).
- A dozen Populus clones have been established in sterile shoot culture.
Most have been established in a proliferative state (McCown 1984).
Shoots of one clone were rooted and planted in a field test plot of
200 trees in northern Wisconsin (B. McCown and D. Riemenschneider,
personal communication).
- Callus and leaf dise cultures have been established for several
Populus clones. Nine clones have exhibited shoot and root formation
from callus through organogenesis. High frequency regeneration of
shoots from leaf discs can be readily obtained with two of these
clones (T. Ettinger, B. McCown, N. Nelson, M. Ostry, P. Read, un-
published data).
19
Table 2.--Current studies of
Program,
the North Central Forest Experiment Station Biotechnology
Study
Stratigic plan in biotegh-
nology for forest trees—
Genetics of whole plant
regeneration in vitro
Anther culture for
haploidy induction in
Populus
Tissue culture systems
for red pine
Genetics of hexazinone
resistance in jack pine
Imparting glyphosate and
sulfonylurea resistance
to elite poplar germplasm
Imparting Septoria
resistance to elite
poplar germplasm
Imparting scleroderris
and needlecast resistance
to selected Larix
Cotyledon and embryo
culture of red pine to
generate somaclonal
variation
Protoplast technology
for Populus
Nuclear DNA changes in
leaves of Populus and
Larix during the growing
season
Conferring glyphosate re-
sistance on selected
Populus through genetic
transformation with aro
A gene mee
Principal
investigators
G.Kidd
B.Haissig,B.McCown,
D.Riemenschneider,
R.Cecich
K.Wolter
M.Sondahl
D.Riemenschneider
N.Nelson,
B.Haissig
M.Ostry,P.Read,
W.Hackett
D.Skilling,
A.Diner,
O0.Karnosky
S.Minocha,
N.Nelson,
D.Riemenschneider
B.McCown
S.Dhillon,
J.Miksche,
R.Cecich
M.Moloney,J.Fillatti,
B.Haissig,
B.McCown
Program
Species objectivea/
all all
Populus rg
Populus = all
Pinus
resinosa
Pinus
Danksiana
Populus hr
Populus dr
dr
Larix
Pinus
resinosa
all
Populus all
Populus,
Larix all
Populus hr
Biotechnology
emphasis
general
genetics/breeding,
tissue culture
tissue culture
tissue culture
genetics /breeding
somaclonal selection
somaclonal selection
somaclonal selection
somaclonal selection
somatic hybridization
recombinant DNA
recombinant DNA
a/ All =
b/ Completed.
all objectives, rg
dr = disease resistance,
= regeneration genes, hr = herbicide resistance,
Table 3.--Future studies of the North Central Forest Experiment Station Biotechnology
Program, planned to begin in 1985.
Study
Genetic modulation of soma-
clonal variation in poplars
Genetics of regeneration
in vitro for loblolly and
jack pines
a/ hr =
herbicide resistance, rg =
Principal
investigators
B.Haissig,
N.Nelson
D.Riemenschneider
D.Riemenschneider,
O.Wells,
B.Haissig
Program
Species object ivesa/
Populus hr
Pinus taeda,
Pinus
banksiana rg
regeneration genes.
20
Biotechnology
emphasis
genetics /breeding,
tissue culture,
somaclonal selection
genetics/breeding,
tissue culture
An in vitro system has been developed for one Populus clone that allows
continued viability of individual protoplasts and division and growth
of these protoplasts to the large calli stage (> 1000 cells per callus)
(B. McCown, unpublished data).
An embryogenic cell suspension culture system has been developed for
one Populus clone, apparently the first report of embryogenesis from
cell suspension for the genus (B. McCown, unpublished data). This sys-
tem is potentially useful for somaclonal selection for chemical and di-
sease resistance as well as for genetic transformation through recom-
binant DNA.
Populus shoot cultures were found to exhibit the highest sensitivity to
cytokinin of any dicot deciduous tree species so far examined (Sellimer
et al. 1985). This cytokinin sensitivity differs markedly by clone
(Sellmer et al. 1985).
Callus was obtained from the anthers of three male Populus clones in our
research on haploidy induction and gametoclonal variation. Viable cal-
lus formed in 4 to 6 weeks, with spontaneous root formation after one
or two subcultures. Sporadic shoot formation was also observed. No
evidence of haploidy in any of these calli, roots, or shoots has yet
been found (K. Wolter, unpublished data). Several experimental param-
eters are being modified to increase the probability of achieving
haploidy.
Populus cells were found to have very small chromosomes and only 1.5 pg
of DNA per nucleus (Dhillon et al. 1984). At least 95 percent of the
higher plant species so far examined have more nuclear DNA than this
(J. Miksche, unpublished data). This surprisingly small genome size
should facilitate genomic analysis through recombinant DNA approaches.
Cytogenetic analysis, however, is made more difficult by the small
chromosome size (R. Cecich, unpublished data).
A leaf dise bioassay for Septoria susceptibility in Populus developed
at NC Station has been refined so that there is a high correlation be-
tween bioassay results and Septoria resistance in field plantations
(M. Ostry, unpublished data). This leaf disc bioassay is being used
in our somaclonal selection system for Septoria resistance.
Successful infection of Populus shoot and leaf disc cultures with
Agrobacterium tumefaciens has been achieved, with gall formation in
shoots in vitro (J. Fillatti and M. Moloney, unpublished data).
Agrobacterium is the transformation vector in our North Central
Station-Calgene-University of Wisconsin recombinant DNA herbicide
resistance work (table 2).
Large and highly significant differences were found between open pol-
linated families of Pinus banksiana in tolerance to the triazine herbi-
cide, hexazinone (D. Riemenschneider, unpublished data). Work will soon
begin on the mode of inheritance of this tolerance.
Cotyledon culture systems of high regenerative capacity have been de-
veloped for Larix decidua (A. Diner, D. Karnosky, D. Skilling, unpub-
lished data).
21
LITERATURE CITED
Burg, A. We, Sacco, G. R., and Wheat, D. 1983. Biotechnology: opportunities
in the biomedical, chemical, agricultural and food industries. 100 p.
Arthur D. Little Report. NY: Arthur D. Little & Co.
Dhillon, S. S., Miksche, J. P., and Cecich, R. A. 1984. DNA changes in sen-
escing leaves of Populus deltoides. Plant Physiol. (Suppl.) 75: 120.
Kidd, G. H. 1984. Strategic plan in biotechnology for forest trees. 19 p.
Final Report for USDA Forest Service, North Central Forest Experiment
Station, Forestry Sciences Laboratory, Rhinelander, WI. (Available from
N. D. Nelson, Forestry Sciences Laboratory, Rhinelander, WI).
McCown, B. He. 1984. From gene manipulation to forest establishment: shoot
cultures of woody plants can be a central tool. Proc. TAPPI 1984 Research
and Development Conf., p. 21-26. TAPPI, Atlanta, GA.
National Academy of Sciences. 1983. Report of the briefing panel on agri-
cultural research. 21 p. Washington, DC: National Academy Press.
Nelson, N. D., and Haissig, B. E. 1984. Biotechnology in the Forest Service's
North Central Forest Experiment Station. Proc. International Symp. of Re-
cent Advances in Forest Biotechnology, p. 139-154. Michigan Biotechnology
Institute, East Lansing, MI.
Nelson, N. D., Haissig, B. E., and Riemenschneider, D. E. 1984. Applying the
new somaclonal technology to forestry. Proc. TAPPI 1984 Research and De-
velopment Conf., p. 27-34. TAPPI, Atlanta, GA.
Selimer, Js Go, Russell, Jc As, Zeldin, E. le, and McCown, Be He 985 .eUrasice
ization of cytokinin response curves in tissue evaluation of Populus for
biotechnology research. HortScience (In press).
Skelsey, A. F. 1984. Biotechnology in agriculture--new tools for the oldest
science. In Reference document: needs assessment for the food and agri-
cultural sciences, p. 229-266. Washington, DC: Joint Council on Food
and Agricultural Sciences.
22
BIOTECHNOLOGY AND FOREST GENETICS:
AN INDUSTRY PERSPECTIVE
R. J. Dinus 1/
Abstract.--Biotechnology is not new to forestry or many
other industries. Man has been using biotechnology, in its
broadest sense, since he began domesticating crops. Recent
advances, however, have made available new genetic and molecular
techniques. The current excitement derives from their potential
application across the spectrum of activities in forest
industry--from producing wood through processing it to using
wastes. Tree improvement, as an example, can be expedited and
made less costly with techniques such as cell culture and gene
transfer. Shortening the time required for selecting, breeding,
and testing may allow forestry to benefit as much as or more
than other industries dependent on plant material. In addition,
forestry, for a change, may be on a par with other industries.
New discoveries, or even genes, can be captured and used
regardless of origin. Reaping dividends, however, requires that
the scientific and industrial communities collaborate in
selecting areas of work, choosing strategies, and planning
research. The most promising areas must be identified; i.e.,
those with the most economic leverage. Coordinated strategies
are likewise essential. Heavy spending on a narrow or applied
front could be harmful. Biotechnology cannot replace other
disciplines, rather it builds upon and provides tools for them.
Balance must be maintained between fundamental and
developmental work. Well-planned, far-sighted experimentation is
more important than ever. Modifying or transferring genetic
information provokes concern and questions. Precautions in
executing research and deploying products are needed to avoid
the perception that more problems are being created than solved.
Without effective safeguards and education, the public may
saddle the technology with unnecessary’ regulation. New
knowledge, as accumulated, should be applied toward betterment
of regulatory procedures.
Additional keywords Tree improvement, tissue culture, gene
transfer, research management, Agrobacterium tumefaciens,
Pinus taeda.
Biotechnology, in the broadest and oldest sense of the word, is not new
to forestry or a variety of other industries. Early man applied and
benefited from biotechnology when he began selecting desirable crop and
animal variants, and took advantage of genetic variation to increase
1/ Forest & Biological Research Manager, Corporate Research Center,
International Paper Company, Tuxedo Park, NY
23
productivity. In more recent times, the word has taken on a somewhat more
restrictive meaning in that new genetic end molecular tools have become
available. Such tools enable us to alter organisms of interest more
dramatically and precisely than ever before, and further allow us to effect
change much faster than by traditional means.
Such developments, not surprisingly, have provoked a wave of interest
and excitement. Prospects for application of the new techniques to a host
of industrial and commercial activities are manifold. In the forest
products industry, potential applications span the spectrum from reducing
the costs of producing raw material and increasing the efficiency of
processing and manufacturing to converting wastes into harmless residues,
salable products, or energy. Replacing even part of the energy and
chemicals needed to make and bleach pulp has considerable economic leverage.
Indeed, first applications in forestry may involve altered organisms or
enzymes for bleaching pulp and/or decolorizing effluents.
Also exciting is the rate at which biotechnology has been advancing.
Ten or so years transpired before restriction enzymes were understood or
became usable as something other than mere research tools. Developing
Agrobacterium tumefaciens into a workable vector for transferring genes
among dicotyledonous plants took roughly seven years. And now, we read
about new developments and products literally on a monthly basis.
Advances have occurred, and are now occurring quite rapidly, in
forestry as well as in other disciplines and industries. Witness that in
early 1983, an entire issue of Science was devoted to biotechnology and its
implications for science, industry, and society. The sole article on
forestry mentioned a number of techniques and applications, but did not
discuss isolation, modification, and transfer of genetic information. As
clearly demonstrated by other papers in these proceedings, involvement in
such research has since increased and progress has been substantial. Within
the last six or so months, evidence has been presented that Agrobacterium
tumefaciens can infect and transform loblolly pine (Pinus taeda). The pace
is thus rapid and undoubtedly will accelerate, as a variety of organizations
and persons have worked to increase funding and support.
As a result, a number of opportunities and problems lie ahead, and we
in research, education, and industry must prepare for them. My purpose then
is to highlight a few accomplishments and explore some of the challenges.
CELL AND TISSUE CULTURE
The first area of concern is the art (and hopefully soon the science)
of cell and tissue culture. Few coniferous species can be regenerated from
protoplasts or single cells, and regeneration from organ culture has not
proven an economical means for multiplying improved material. Even so, much
knowledge has been garnered from efforts to accomplish such goals. What can
now be gained from the experience? And, what direction should such research
take in the future? What role should workers in the public and private
sectors play?
Tree improvement has become an integral part of management in many
forest regions, and generally is recognized as a worthy enterprise.
24
Improvement of southern pines essentially has become a business--a highly
profitable one despite the considerable cost of entry and operation.
Progress and profitability nevertheless remain limited by the lengthy
reproductive cycle, the space and time required for testing, and the
considerable expense associated with testing and selection.
Can tree improvement be expedited and made less costly by imaginative
application of cell and tissue culture? The answer seems positive, provided
we work together and focus talent on important issues. Doing so seems
especially important in view of the economic situation facing us now and for
the foreseeable future. Demand for raw material is no longer rising as
rapidly as in earlier times, inflation has abated, and expectations have
changed. The key to sustained profitability (and I might add, continued
interest in research) may, therefore, rest on our ability to reduce costs of
production. Indeed, lessening the time and expense of selecting, breeding,
and testing may allow the forest industry to benefit from the new emphasis
on biotechnology as much as or more than other industries dependent upon
breeding and growing plants.
One example of such applications involves placing cell or callus
cultures under stress, such as that provoked by low temperatures, restricted
moisture availability, or toxins from pathogens. Testing and/or increasing
selection intensity in culture can hasten identification and isolation of
useful genetic variants. More entries can be evaluated in less time and
space than in conventional tests. Approaches, such as protoplast fusion,
can be used to increase variability. And, haploid material could be
generated for use in research.
Realizing full benefit from such approaches, however, requires recovery
of functional plants. Indeed, the utility of many new techniques will be
limited until efficient, reliable means of organogenesis and embryogenesis
are developed. Just how to effect that development most rapidly remains
controversial. Some argue for allocating more funding and workers to the
traditional empirical approach. Others hold that more emphasis should be
placed on fundamental studies of differentiation. The problem of balance is
serious, and need exists for work on both fronts.
What mechanisms control expression of the genes involved in
differentiation? How do growth regulators, environmental conditions, and
nutrients affect those mechanisms? What biochemical events occur in
developing embryos and can we learn to provoke them in culture? Much
remains to be learned, and answers to such questions will facilitate
progress on cell and tissue culture, and perhaps hasten the day when we can
generalize from an easily manipulated species to others of greater interest
but difficult to culture. Knowledge about the processes and mechanisms of
_differentiations will also improve our understanding of growth in intact
plants -- the components contributing to it, the underlying traits, and how
they can be manipulated more easily. Adequate justification exists for
continuing work on both approaches. We would be well advised, however, to
provide somewhat greater support for work on processes and mechanisms than
has been available in the past. Such a position, hopefully, would encourage
continued movement of public sector scientists back to fundamental issues.
25
GENE ISOLATION AND TRANSFER
Another area of concern involves a set of techniques that has
tremendous potential, that of isolating, cloning, modifying, and
transferring genetic information. Work is progressing rapidly as indicated
by another paper in these proceedings. Two specific strains of
Agrobacterium tumefaciens, have been shown to infect loblolly pine and some
bacterial genes were found to’ have been inserted into and expressed by the
pine genome. Work on other gene transfer systems, such as micro-injection
and liposomes, is also underway. Thus, we can expect within the near future
to have available the technology to transfer single genes into the genomes
of desirable trees. The ability to transfer the many genes presumed to
control the most important traits, however, will not be within reach for
some time.
Some other limitations should also be noted. Reaping dividends from
such research, after all, requires that the gene be expressed at the correct
time and place, that we can convert the transformed cell, callus, or organ
to an intact, functional plant, and that we can multiply that plant, by
sexual or asexual means, in a cost-effective manner.
Progress is also constrained by our understanding of only a very few
forest tree genes well enough to attempt isolation and transfer. To some
extent, this limitation can be overcome by borrowing genes for traits of
interest from other organisms (for example, herbicide resistance from
bacteria). This is one reason that biotechnology is so exciting for
forestry. Within limits, new discoveries and even genes can be captured and
utilized in forest research and development, regardless of origin. Thus,
biotechnology may place forestry, for a change, on a par with other
industries.
The larger problem nevertheless will persist for some time. Neither we
nor other plant scientists know which genes are important or understand the
activity of those that have been identified. Identifying genes and
understanding gene action will not be easy, regardless of the plant or tree
species. This aspect of biotechnology may well prove the most challenging.
When considering plant genomes, one must contemplate which gene of perhaps
one or more million is of interest. Several thousands or tens of thousands
may be active in a particular organ or tissue at any given time. Which are
active, what activates them, and how we capture the one of interest remain
key questions. Such topics clearly deserve increased attention and seem
best addressed by scientists in the public sector.
One might regard this situation, regrettably, as but one symptom of
past neglect. The ebb and flow of research and education has been such that
sufficient attention was seldom given to the basics of how trees grow. One
danger inherent in the excitement about biotechnology, therefore, is that
qualified workers will all rush to get on the "genetic engineering"
bandwagon, reducing even further the magnitude of effort on fundamental
issues.
Despite the many difficulties, work will continue and advances will
OCCU: Before too long, useful genes will be moved into or among tree
26
species, and their expression will be confirmed. Ensuring continued
interest and support, however, will be as difficult as doing the research.
The economic climate of the present and foreseeable future is such that
first results must be winning ones -- preferably seen as shortcuts to
increased profitability. Careful selection of goals and areas of
investigation is, therefore, necessary. Important traits must be identified
and research strategies set such that early efforts will produce findings
and/or material that can be moved quickly from research through development
to commercialization. Continuing collaboration between the scientific and
industrial communities in selecting areas of work, choosing strategies, and
planning research is essential to ensure that investments in biotechnology
are worthwhile.
CHALLENGES FOR THE FUTURE
Balancing on the High Wire
The foregoing sections were intended to provide a sense of the
opportunities presented by forest biotechnology. They also should have
surfaced some problems that must be resolved before the promises can be
realized. Significant among the problems is the perennial tendency to
regard new activities as panaceas or bandwagons. Though the associated
dangers were mentioned earlier, the need to resist such tendencies must be
reemphasized.
Bending biotechnology to yield real accomplishments in forestry
requires a concerted effort of individuals from many disciplines and
organizations. Molecular biology is exciting, but significant challenges
also lie in the more traditional areas of tree breeding, physiology and
biochemistry. Thus, there is still need and perhaps even greater need for
increased research in such disciplines.
The new techniques are not a replacement for other disciplines, rather
they are new tools for all to use. Indeed, their most significant near-term
use may be in enhancing our understanding of tree growth and
development--differentiation in cell and tissue culture being but one
example. Molecular biology has as much to offer forest genetics as the
latter discipline has for the former. Never before have we seen greater
opportunities for collaboration among disciplines.
Coordinated and far-sighted strategies are thus’ essential to
maintaining reasonable balance between disciplines and approaches. Heavy
spending or plunging for headlines along narrow or applied fronts can do
more harm than good. Without continued emphasis on the traditional
disciplines and fundamental issues, progress will be as short-lived as it
has been dramatic. Achieving a balanced research agenda is also essential
if we are to attract the few brightest students to forestry, and train them
to investigate, develop, and implement this attractive, but complex
technology.
Maintaining an appropriate balance will not be easy. Economic
conditions have made funding for research harder to obtain. On the positive
side, attitudes about research have also changed, and so-called hard science
has become more popular. The research community may therefore find it
27
easier to work on fundamental topics in both new and traditional areas than
one might imagine at first glance.
The Sky May Fall
The emergence of forest biotechnology magnifies the traditional
challenges about plantation monocultures and clonal forestry. Modifying and
transferring genetic information naturally provokes concerns and questions
from the public. Adding such concerns to the usual ones will generate more
and harder questions.
Both our clients and the public will want to know more about what we
investigate, what we produce, how we deploy it, and how it will affect the
environment. Some actually may seek a role in determining what is done,
why, and how. Not providing them information will create doubt and possibly
fear. Imaccurate information or mere opinion will diminish credibility.
Thus, well-planned and far-sighted research is more important than ever.
Meaningful precautions must be taken in designing and executing research so
as to avoid any perception that more problems are being provoked than are
being resolved. We must also take the lead in educating our clients and the
public. They must be assured of safety. Unless we accept and meet this
challenge, all promises could be delayed or even forfeited. Without
effective safeguards and education, the public may insist upon regulations
that unnecessarily slow or complicate research, development, and
commercialization. As responsible scientists, we must further provide
accurate data to the agencies responsible for formulating and applying
regulations. Our goal should be a responsive and responsible system of
regulation that will satisfy public concerns and not inhibit sound research
and development.
United We are Funded, Divided We Fall
As mentioned earlier, attitudes toward and the outlook for forest
research have changed over the last decade. Funding, expressed in real
dollars, has declined during most years, regardless of the
organization--university, federal, or industrial. Some years have been
better than others, but the average trend has been down or flat. Yet
another trend has surfaced as well, that being the more careful choosing of
research directions, the justification of expenditures, and the evaluation
of payback. Just who does what research has also received more attention.
Thus, public sector organizations are moving away from shorter-term,
developmental activities to concentrate more on fundamental, longer-term
issues. The Industry, on the other hand, is tending to concentrate less on
hard science, and more on development and application. The outcomes have
been several.
While cause and effect cannot be proved, one certainly can argue that
such trends paved the way for increasing support of biotechnology. Most
such research, regardless of the organism, was once conducted in a few
universities, and largely supported by small federal grants. Now, many
universities are establishing so-called Institutes of Biotechnology, aided
by modest appropriations from their state legislatures. In addition, the U.
S. Forest Service has initiated a modest program. Much impetus has also been
lent by establishment of Competitive Grant Programs, first by the USDA, and
28
more recently, the Forest Service. Thus, increased activity on the
biotechnology front can be expected in universities, the Forest Service, and
eventually, industry.
Some other outcomes are also of interest. The several trends have led
to establishment of strong cooperative (and sometimes contractual)
relationships between universities, the Forest Service, and/or industry.
Witness the formation of herbicide, nursery, and pest management
cooperatives among others in addition to the long-standing tree improvement
variety. Moreover, the industry has become more involved in research
planning at the state, regional, and national levels. The National Forest
Products Association, for example, now has a National Research Committee and
five Regional Subcommittees. A special subcommittee monitors biotechnology
research. Industry representatives are prevalent on advisory boards, and
are frequent participants in formal and informal research reviews.
While such involvement is not new, the interactions are more intensive
and considerably more harmonious than in earlier times, and generally of
mutual benefit. That is, the several communities have learned much about
their individual strengths and weaknesses, and are acting to help one
another meet their respective needs. The research communities desire to
perform more and better research and need support. Industry desires to
promote the quality of research and to maintain the flow of research
information germane to its goals. With time and effort, such interactions
can be further strengthened, and used to secure a balanced research agenda
and make the promise of biotechnology become reality.
29
TESTING AND DEPLOYMENT OF GENETICALLY ENGINEERED TREES
W. J. Libbyl/
Chapter Outline for Bonga & Durzan, 2d Ed.
Ve INTRODUCTION
1.1 What is meant by “genetically engineered", “testing” and “deployment”?
1.1 Some preliminary questions
Dike TESTING
2.1 General considerations
Genetic variation within units
Environmental variation within and between test sites
= Cl
— —————_—_—_——_—_—_ ee a ee
Sie CLONAL TESTING
3.1 Sensitivity to broad-sense heritability
3.2 Sensitivity to selection intensity
3.3 Large contiguous plots vs non-contiguous plots
3.4 Typical vs atypical sites
3.5 Possibility of testing for competitive compatibility
4. SPECIAL CONSIDERATIONS FOR GENETICALLY ENGINEERED TREES
4.1 The relationship of genetic novelty and test stringency
4.1.1 Two true stories
4.2 The heritability of genetically engineered characteristics
Die GENERAL DEPLOYMENT STRATEGIES
1 Provenances
-2 Unusual sites
6. CLONAL DEPLOYMENT STRATEGIES
6.1 Maximum and minimum numbers of clones per locale
6.2 General-purpose vs. interactive clones
6.3 Wimps vs Moms
Tes SPECIAL STRATEGIES FOR GENETICALLY ENGINEERED TREES
7.1 Minority mixes
_1/University of California, Berkeley
30
BIOTECHNOLOGY Il
MODERATED BY DR. MIKE GREENWOOD
University of Maine
31
t
i eee
4)
ral ine
we,
DNA TRANSFER AND GENE EXPRESSION IN LOBLOLLY PINE
Ronal R. Sederoff’, Anne-Marie Stomp-, W. Scott Chilton?, and Larry
Moore .
We wish to report on a system for the transfer and expression of foreign genes
in pines. The purpose of these experiments is to explore the use of the crown
gall bacterium, Agrobacterium tumefaciens for the eventual goal of genetic
engineering in important forest species. Previous work has described pines as
resistant to infection by crown gall, however, we have found a strain of A.
tumefaciens that will produce galls on loblolly pine. The frequency of gall
formation is 3 percent. One of these galls has been removed and cultured on
pine tissue culture mediun. Cells from the resulting callus were extracted
with ethanol and tested for the presence of opines by high voltage paper
electrophoresis. The strain of crown gall that infects pine is known to
synthesize agropine and mannopine in galls that have been induced in sunflower.
Opines, particularly agropine, are found in abundance in callus derived from
the pine gall, but are not detected in extracts of uninfected plants or
uninfected loblolly pine callus. The presence of specific opines in infected
pine cells provides strong evidence for the transfer and expression of foreign
genes in pines. This system appears suitable for genetic engineering of
commercially important conifers including loblolly pine.
1. Institute of Forest Genetics, Pacific Southwest Forest and Range
Experiment Station, USDA Forest Service.
2. Department of Forestry, North Carolina State University.
3. Department of Botany, North Carolina State University.
4, Department of Botany and Plant Pathology, Oregon State University.
32
USE OF TISSUE CULTURE TECHNIQUES IN A HARDWOOD TREE
IMPROVEMENT PROGRAM
D. W. Einspahr and S. R. Wann*
ABSTRACT
Tissue culture per se is not a method of genetic improvement but is instead
a method of vegetatively propagating trees and plants. Tissue culture methods
available to forestry are micropropagation, organogenesis, and somatic embryo—
genesis. A long-term industry-state cooperative conventional tree improvement
program has used the techniques of selection, hybridization and polyploidy to
produce rapid-growing Populus species hybrids. Aspen hybrid seedling popula-
tions have been produced that at 18 to 20 years grow approximately twice as
fast, have 20 to 30 percent longer fiber length, and, 8 percent higher wood
density than widely used native aspen. Only modest improvement, has been made,
however, in producing hybrids that are resistant to the regions most serious
forest management problem, hypoxylon canker (Hypoxylon mammatum Wahl., Miller).
Recent Ph.D. and related research has resulted in tissue culture procedures
that allow us to readily screen seedling populations for seedlings that are highly
resistant at the cellular level to the canker toxin, the reported determining
factor in the disease. Also utilized are procedures for bioassaying field
tested parent trees, hybrids, and young seedling populations for resistance.
Planned is an expanded tree improvement program that will combine conventional
tree improvement techniques and tissue culture procedures to produce hypoxylon
resistant hybrid clones and seedling populations. Micropropagation and organo-
genesis methods are presently available for use in producing operational clonal
plantings. At present there appears to be adequate natural resistance in
existing seedling populations, so that the use of more sophisticated genetic
engineering techniques (transformation, protoplast fusion, etc.) may not be
required to solve this serious disease problem. Tissue culture techniques simi-
lar to those described could be expected to be useful in evaluating experimental
crosses and screening and generating disease resistant parent trees in other
hardwood tree improvement programs.
*The authors are, respectively, Senior Research Associate and Industrial
Research Fellow, Forest Biology Section, The Institute of Paper Chemistry,
Appleton, WI, 54912.
33
INTRODUCTION
Emphasis on the use of tissue and cell culture in the propagation of
forest trees has increased dramatically in the last five to six years. The
tissue culture methods available to forestry are micropropagation, organogenesis
and somatic embryogenesis. Micropropagation is the in vitro propagation of
plants using stem meristems, i.e., shoot tips and apical buds. Organogenesis is
the in vitro propagation of plants from explants or callus where the organs
(roots or shoots) are produced and then are manipulated to produce complete
plants. Somatic embryogenesis is the in vitro propagation of plants, from
single cells or small groups of vegetative cells, where the final stages of
development produce embryolike structures that are capable of developing into
intact plants. Although somatic embryogenesis appears to have the most promise
for use with forest trees, each of the other methods also has its place in
forest tree improvement work. The purpose of the discussion that follows is to
illustrate how tissue culture techniques have and are being used in a hardwood
tree improvement program in the Lake States. The hope is that in learning of
these results, you will see ways you may be able to use similar approaches to
solve problems associated with your tree improvement programs.
THE ASPEN GENETICS PROGRAM
The discovery of many rapid-growing, good-quality diploid and several
triploid quaking aspen (Populus tremuloides Michx.) clones resulted in the
establishment of an industry-sponsored aspen tree improvement program in 1955.
The objectives of the program were to use the techniques of selection, hybridi-
zation and polyploidy to produce rapid-growing trees with improved wood proper-
ties. Many of the early ideas were those of Dr. Philip Joranson and were
implemented by the first author and other researchers including Dr. Lawson
Winton and Dr. J. P. van Buijtenen. During the 20 years that followed, more than
700 full—- sib crosses were made, and the crosses included the hybridization of
canescens. Crosses were a5 A the EB TecanoUseeine the cut branch technique
of Wettstein!. The most often used procedure was to complete the crosses in
February and March, produce 300-400 1-0 seedlings from each cross the following
summer, and use these trees in one or two replicated field plantings. This
crossing and field testing program resulted in the development of several types
of crosses that appear to have potential for use by the paper industry. To
date, the most useful crosses are those between P. tremuloides and P. tremula.
Particularly promising have been a series of crosses using several Pe tremu-
loides females and a tetraploid (4n) P. tremula male developed by H. Johnsson in
Sweden2. As a result of this long-term research program, hybrid aspen seedling
populations have been produced that at 20 years grow about twice as fast as
native aspen, have 20-30% longer fiber length, and have about 8% higher wood
density>. Associated with the described improved wood properties were improved
paper properties (greater tearing and bursting strength and comparable tensile
strength). Presently the forest management approach being Sungeeted is to plant
the triploid hybrid seedling populations at wide spacing (80-100 ft 2/tree) on
medium to low quality northern hardwood sites, using conversion planting tech-
niques. The prolific suckering ability of most aspen and aspen hybrids will
then allow the management of the plantings using coppicing procedures for
several rotations.
34
THE HYPOXYLON CANKER PROBLEM
Hypoxylon canker (Hypoxylon mammatum Wahl., Miller) is the most serious
forest management problem influencing the use of aspen and aspen hybrids. Losses
in the Lake States have been estimated to be 3% annually’, an amount approximately
equal to the annual harvest in 1983 of 280 million cubic feet,
One of the original objectives of the aspen genetics program was to
select and breed for resistance to this serious canker disease. A modest amount
of progress has been made by selecting and using parent trees that were free
from hypoxylon. Progeny tests were evaluated at five-year intervals through age
25, and at present 20-year records exist for more than 60 full-sib aspen and
hybrid aspen crosses. Typically, crosses with low resistance to hypoxylon will
have infection levels of 46-68% at 20 years, whereas the best tremuloides x
tremuloides crosses had 20-year infection rates of 0 to 18%. Although progress
has been reasonable, there is an urgent need to increase the number of highly
resistant parent trees and to develop clones with a high degree of resistance
for use in a planned clonal forestry program.
CHARACTERISTICS OF THE DISEASE
Hypoxylon canker is caused by the fungus Hypoxylon mammatum (Wahl.,
Miller). The wide geographic range of quaking aspen and the seriousness of the
disease has resulted in the establishment of many research investigations into
the nature of the disease and factors associated with epEeed of the disease.
One of the most interesting was the discovery by Hubbes” of a diffusible
substance produced by the fungus that elicited symptoms characteristic of the
. disease. The necrotic response to “mammatoxin"™ was later shown to be host-
selective for species susceptible to hypoxylon canker and strongly suggests that
the toxin is a determinant in the disease’*%. The necrotic response to mam-
matoxin resulted in the development of leaf bioassay by Bruck and Manion? in
which the toxin was substituted for the pathogen in a procedure designed to
identify disease-resistant clones.
Many important diseases of agronomic crops have been shown to have
host-selective toxins as determinants in pathogenesis!9, Employing toxin in a
tissue culture system has resulted in the isolation of toxin-resistant cell
lines. In those cases where ote were regenerated from these cultures, toxin
resistance often persisted!!, 2, Further testing of the plants with the patho-
gen resulted in equating toxin resistance with disease resistance. Considering
the above success in agriculture and that mammatoxin appears to be a deter-
minant in hypoxylon canker, the use of mammatoxin in tissue culture systems to
screen for toxin resistant seedlings was a logical approach to attempt to pro-
duce cellular-level resistance to hypoxylon canker.
*% is the number of individuals lost or presently infected with hypoxylon canker
divided by the total number of field planted trees exposed to the disease.
35
TISSUE CULTURE SELECTION OF TOXIN RESISTANT ASPEN
During the summer of 1983 a student research program was initiated that
had the purpose of isolating and propagating mammatoxin-resistant quaking aspen.
Part of this Ph.D. research program was to develop a procedure that allowed the
regeneration of plantlets from cotyledon explants. The procedure that resulted
(Wann and Einspahr!3) involved induction of multiple adventitous buds on cotyle-
don and hypocotyl explants by culturing the explants on MS medium containing 0.1
mg/L NAA and 1.0 mg/L BA. Explants producing multiple buds were then transferred
to 1/2 MS (macro- and microelements) containing 0.3 mg/L BA for elongation. Root
formation was achieved by transfer to 1/3 MS containing 0.1 mg/L IBA. Normally
about 90 percent of the elongated shoots rooted, and transfer to soil was
accomplished with little difficulty by maintaining high humidity conditions for
one month after transfer.
The procedure used to screen and propagate mammatoxin-resistant quaking
aspen is illustrated in Fig. 1. Small hypocotyl and cotyledon explants were
placed on the “bud proliferation” medium in which mammatoxin was substituted for
part of the water. Following four weeks on this medium, organogenesis was eval-
uated, and surviving explants were rescued from the toxin-containing medium and
transferred to the toxin-free elongation medium. The surviving elongating
shoots were rooted and transferred to soil. The resulting plantlets were then
grown for 18 weeks and bioassayed using the previously cited bioassay method
gevetoped by Bruck and Manion? and described by Griffin, et al.! and Stermer,
eal ja The bioassay method consisted of removing three leaves from the
18-week-old aspen plantlets and placing the petiole of leaves in a small vial of
water. Small holes were made in the leaves with a minutin insect pin, and a 3 uL
drop of the properly diluted toxin was placed over the hole. Usually, three
holes were made in each half of the leaf blade. Following incubation in a humi-
dified chamber at 28°C for 48 hours, the response to the toxin was measured as
lesion diameter to the nearest 0.5 mn* .
An additional novel complementary study was run in which, for 120
seedlings, one cotyledon was removed and placed on the toxin containing screening
medium. The original seedlings (minus one cotyledon) were grown for 18 weeks for
use in leaf bioassay comparisons with those explants and resulting plants that
survived the toxin screening and tissue culture propagation procedure. Figure 2
illustrates the procedure used. In this way, the bioassay response of a toxin-
screened plantlet could be compared with the bioassay response of the donor plant.
The results of this toxin screening procedure and the leaf bioassay of
the resulting plantlets turned out to be very interesting. When plants sur-
viving the screening procedure were tested in the leaf puncture bioassay, this
resistance was still maintained in the tissue-culture propagated ramets. For
example, 5 clones comprising a total of 23 individuals were obtained from a
full-sib cross, and all responded in a manner analogous to cottonwood, a species
*additional details on producing the toxin and running the leaf bioassay are
available in a paper by Wann and Einspahr!®,
36
that is resistant to hypoxylon canker and reacts negatively (lesion diameter < 1]
mm) to the leaf bioassay. Equally interesting is that when the donor plants
were compared with 22 clones of toxin-screened, cotyledon-derived plantlets (an
ortet/ramets comparison), both the donor plants and corresponding toxin-derived
plantlets tested resistant in the leaf bioassay. Figure 3 illustrates the
results of in vitro toxin screening of two full-sib seedling populations.
As a result of these investigations it appears that mammatoxin can be
used to rogue organ cultures of aspen seedlings for cellular level resistance to
mammatoxin, and the plants propagated from these cultures retain this trait. The
ability of an explant to survive and produce shoots on a toxin-containing medium
is apparently inherent in the seedling from which it was derived, and not induced
by the tissue culture system. Highly resistant individuals occurred to a signifi-
cant extent in all crosses examined, indicating the resistance was of natural
origin. This indicates that despite the current interest in genetic engineering
(transformation, protoplast fusion, etc.), it appears the full genetic potential
of many commercially important forest species is not now being fully utilized. =
PLANNED EXPANSION OF THE ASPEN PROGRAM
Aspen is the most important pulpwood species in the Lake States Region.
An attempt is being made to obtain an appropriate level of funding that will allow
us to (1) determine the ability of the leaf puncture bioassay procedure to identify
canker-resistant trees, (2) determine the heritability of mammatoxin resistance
in quaking aspen, (3) verify the usefulness of tissue culture procedures in the
production of a highly resistant parent tree for use in future breeding work, and
(4) develop procedures for screening 15 to 25-year-old hybrid aspen populations
for resistant individuals that can be used in a planned clonal forestry program.
The existence of appropriate parent trees and of 20-year field data on
full-sib crosses will allow us to repeat crosses of interest and to compare the
tissue culture seedling screening results and leaf bioassay data with the
20-year field results. The existence of parent trees and the ability to produce
and evaluate progeny will also allow heritability estimates to be made. The
existence of thousands of 15 to 25-year-old aspen hybrids in Northern Wisconsin
plantings will serve as a source of rapid growing clones, a percentage of which
may also turn out to be resistant to hypoxylon canker.
APPLICATION OF THE APPROACH TO OTHER HARDWOOD PROGRAMS
The tissue culture work by Wann described above represents the first
instance where, for a commercially important forest tree species, a tissue
culture system has been used to select resistant cultures and regenerate plants
that maintain the resistance. This approach and similar procedures appear to be
particularly appropriate for use with forest tree disease problems where the
disease is of a toxin-determinate nature (Fusaria and Alternaria species, for
example). This work also emphasizes the need for the development of several
alternative tissue culture systems for the important U.S. tree species so that
we can efficiently cope with future insect and disease problems.
37
ll.
Die
13.
14.
Pi
16.
LITERATURE CITED
Wettstein, W. V. 1933. Die Kreuzungsmethod und die Beschreibung von
Fl Bastarden bei Populus. Zeitschrift fur Ziichtiing, A. Pflanzenztichtung
18:579-626.
Einspahr, D. W. 1984. Production and utilization of triploid hybrid
aspen. lowa State J. of Research 58(4):401-409.
Einspahr, D. We; Wyckoff, Ge. W. 1975. Aspen hybrids promise future source
of Lake States fiber. Pulp Paper 49(12):118-119.
Marty, Re 1972. The economic impact of hypoxylon canker on the Lake
States resource. U.S. Dept. of Agric. For. Serv. Gen. Tech. Rept. NC-l.
Blyth, J. E.; Smith, W. Be. 1984. Pulpwood production in the Lake States
by County. The Timber Producer, Sept., 1983:20-22.
Hubbes, M. 1964. New facts on host-parasite relationships in hypoxylon
canker of aspen. Can. J. Bot. 42:1489-1494.
Schipper, A. L., Jr. 1975. Hypoxylon pathotoxin necessary to the infec-
tion by Hypoxylon mammatum. Proc. Am. Phytopath. Soc. 2:46-47.
Schipper, A. L., Jr. 1978. A Hypoxylon mammatum pathotoxin responsible
for canker formation on quaking aspen. Phytopath. 68:866-872.
Bruck, R. I.; Manion, P. D. 1980. Mammatoxin assay of genetic and
environmental predisposition of aspen by Hypoxylon mammatum. Plant Disease
64: 306-308.
Scheffer, R. P.; Livingston, R. S. 1984. Host selective toxins and their
role in plant diseases. Science 223:1/7-21.
Gegenbach, R. G.; Green, C. E.; Donovan, C. M. 1977. Inheritance of
selected pathotoxin resistance in maize plants regenerated from cell
culttunses. Proc.) Nat. Acad. Sci. USA. 7435113=5117.
Hartman, C. Le; McCoy, T. J.3;)Knovs, I. R. 1984. Selection of alfalfa
cell lines and regeneration of plants resistant to the toxin(s) produced by
Fusarium oxysporum f.sp. Medicaginis. Plant Sci. Lett. 34:183-194.
Wann, S. R.; Einspahr, D. W. Reliable plantlet formation from seedling
explants of Populus tremuloides (Michx.). Accepted Silvae Genetica. 1985.
Griffin, D. H.; Ehrenshaft, M.; Manion, P. D. 1980. Host-selective toxins
of Hypoxylon mammatum for Populus tremuloides. Pages 209-214 in: Resistance
Mechanisms in Poplar Diseases. Ed. by M. Ride. FAO International Poplar
Commission, 2lst Conference.
Seermmer-) Be Aes) Schetfer, KR. Pos Hart, Jie He L984. Lsolation of Toxins: of
Hypoxylon mammatum and demonstration of some toxin effects on selected
clones of Populus tremuloides. Phytopath. 74:654-658.
Wann, S. R.; Einspahr, D. W. 1985. In Vitro isolation and propagation of
Mammatoxin-resistant quaking aspen. Submitted to Forest Sci.
38
*aany[No ensst}
quetd BFA set} JWUEISTSeT UTXO} JO UOTIDeETES oY} JOJ pesn poyyew eyL_ “*] e1NnsTY
uedse }UC}SISO/ UXO} JO UOIJDOJOS 104 WI9}SAg 81N3jND eNssi|
ONSSi} 1US}SISO1-UIXO} JO
e6e1U99100 jjeUIs UO pednpoid s}00US
pe1001 pue pesidxe sJOOYUS vA i
uixoy Buiuie,Uuod eipew uO
pesejd shuljpees pei2essig
a4 uojAxodAy 0}
@9U21SISOJ UXO}
10} pe}Se} SABE} J10S 29ue}sises BuiAuen YUM
0} Pa110jSUes} S}O]}Ue}, uo1jzejndod Buljpees
. UIXO} —<«—gj——— sn6uun}
91N}jND enssi} }UeId LIA S99J1} JULJSISAI BSLASIP JO UOI}I3Bj9S
39
seedling
establish shoot culture
from single cotyledon
(+ and - toxin)
grow |~18 weeks
propagate
and establish |~26 weeks
in soil
compare bioassay of ortet (seedling)
vs. ramet (tissue culture propagule)
Figure 2. The methods used to propagate plantlets for the ortet vs. ramet
bioassay comparison.
40
XT—20—84
Z2
©
a=
6
axel
=
a
©
oO.
x2
SS
S)
Zz
Lid
>
‘e)
Lid
oc
a
No Lesions Lesions >imm
open bars = screened population; closed bars = control
Figure 3.
Illustrated by the open bars are percentages of toxin-resistant indi-
viduals (no lesions) vs. susceptible plants (lesions > 1 mm) from two
aspen full-sib populations that were screened using the in vitro
screening procedure. The closed bars indicate the inherent variation
in toxin resistance in the original populations.
41
TISSUE CULTURE OF SWEETGUM (LIQUIDAMBAR STYRACIFLUA L. )
H. E. Sommer, H. Y. Wetzstein and N. ace
Abstract.--An improved method for the tissue culture
propagation of sweetgum (Liquidambar styraciflua L.) using a
liquid culture stage is under development. This method produces
more and larger shoots per culture than previous agar based
methods. Plantlets from these shoots have been hardened off
and grown in a nursery bed. The root collar diameters and
heights of several clones after one season in the nursery are
reported. Poor root form is the current problem limiting the
use of these plantlets for field establishment. Photosynthesis,
anatomy and alternate rooting methods have been studied for the
evaluation of the efficiency and predictability of plantlets.
Liquidambar styraciflua is one of the major hardwood species in the
Southeast United States. Once superior selections have been made, a method
of propagation will be needed. Several alternatives include use of ¥4 sib
families, conventional vegetative propagation and tissue culture. The
latter two methods have the advantage of immediate genetic gain equal to
the genotype of the selection through clonal replication. However, with
Liquidambar, conventional vegetative propagation methods are inefficient.
Thus alternative propagation methods using tissue culture are being in-
vestigated for this species. This report describes some of the refinements
in tissue culture methods for Liquidambar.
MATERIALS AND METHODS
Agar Culture Methods
Seed was collected by the U.S. Forest Service from the Oconee National
Forest and kept as half-sib lots. Seeds were surface sterilized and germi-
nated under aseptic conditions on a modified Risser and White's basal
medium (1,4,7). Hypocotyl sections were placed on a modified Risser and
Whites medium with 1.0 ppm IAA and 5.0 ppm 2ip. After excision, the shoots
were rooted on a modified Risser and White's rooting medium (4,7).
Liquid Culture Methods
Seed were prepared as described under agar culture methods. Shoots
were initiated from hypocotyl sections on a modified Risser and White's
medium with 0.1 ppm NAA and 0.5 ppm 2BA (4,7), multiplied
1/
— Associate Professor, School of Forest Resources, Assistant Professor and
Graduate Student, Department of Horticulture, University of Georgia, Athens,
Ga. i
42
on a modified Blaydes' medium with 0.01 ppm NAA and 0.5 ppm BA, and then
placed on a modified Risser and White's basal medium for growth and rooting
as previously described (6,7,8). Cultures were maintained at 25 + 2°C with
a 15 hr photoperiod, under cool white fluorescent lamps.
Nursery Bed Evaluations
Plantlets were removed from the agar rooting medium, and planted in
Can Am pine tubes filled with a potting mix of vermiculite and sand (1:1
v/v). Plantlets were hardened off by gradually lowering the relative
humidity (10,11), then maintained in a greenhouse or lathhouse prior to
planting.
About 800 plantlets were planted on 4" centers in a cement block
nursery bed between 30 May 1983 and 20 June 1983. Plantlets were lifted on
5 March 84. Stem length and root collar diameter were measured; root
quality ratings were made.
Photosynthesis and Anatomy
Plantlets and seedlings were placed in a growth room maintained at 25
+ 2°C with a 16 hr photoperiod and placed under one of three quantum flux
densities: _,50,+ 5 (low light), 155 + 10 (medium light), 315 + 15 (high
light) wEm “s (2). Net photosynthesis of seedlings and plantlets was
determined using an infra-red CO, analyzer (9). Tissues were prepared for
light and scanning electron microscopy as previously described (10,11).
RESULTS
Agar Culture
The results of the culture of sweetgum hypocotyl sections on agar have
been reported (6). The yield of plantlets using the agar system is shown
in Table 1. No net increase in plant number was obtained when considering
initial seedling numbers. Even in terms of plants responding to culture,
multiplication rates were low. Results were highly variable among seedlots.
This method was the optimum of an investigation involving a 2x5x5 factoral
experiment with 2400 cultures (7). It was thus felt that the limiting
factor was not nutritional, but some other factor in the system. Agar in
the medium may cause water stress, thus a liquid medium step was incor-
porated into the culture protocol.
Table 1.--Yield of plantlets from agar cultures
‘% Shoots
# of % Seedlings Giving Average # Yielding
Seedlot Seedlings Successful Cultures Shoots Plantlets
76-1B 80 28 Des) 36
76-5B 93 56 216 39
76-7B 50 80 3.9 28
76-10B 513} 40 Be il 23
78-1B 23 26 Dee 23
Hypocotyl sections from seedlings were cultured on Risser and White's
medium. After excision, shoots were rooted in medium without hormones.
43
Liquid Culture
For liquid culture experiments, buds were first initiated from
hypocotyl sections on a revised Risser and White's medium, solidified with
agar (4,7,8). Eight to 12 weeks following initiation, hypocotyls were
transferred to a liquid Blayde's medium (8,12). After 8 weeks, some of the
cultures had proliferated and started to produce shoots of varying sizes.
The yield of shoots using liquid culture proliferation is shown in Table 2.
Shoot number varied with clone. There exists the potential of harvesting
large numbers of shoots per culture. In addition, several harvests and
subcultures can be taken from each flask, which would greatly amplify the
multiplication potential of this system. Table 3 shows the rooting of
shoots obtained from liquid culture, when placed on basal modified Risser
and White's medium. Most clones exhibited well over 50% rooting of shoots.
Table DE of shoots from liquid culture
Seedlot
Clone Origin # of Shoots
Spy) 80-5B 28
5587 81-8U 17
5595) 81-8U 32
5635 81-2U 52
5637 81-2U 45
5646 81-2U 0
5659 81-2U 77
5688 81-14B 70
5718 81-4U 15
5727 81-4U 10
Si: 81-4U 40
5780 81-76-8U 16
Hypocotyl sections were placed on modified Risser and White's
medium 26 August 1983, removed 27 February 1984 and transferred
to modified Blaydes liquid medium. The above data is from the
harvest of 10 June 1984.
Growth of Plantlets in Nursery Beds
Growth of the six major clones in nursery beds is given in Table 4 and
in Figures 1 and 2. From 74 to 86% of the plantlets were recovered; average
height and root collar diameters varied among clones (Table 4). Variation
within a clone was large (Figs. 1, 2). Most of the plantlets were less
than 85 cm high (Fig. 1), which is considered the maximum height desired
for planting. The root collar diameter of most of the plantlets exceeded
6.25 mm (minimum diameter for planting), and few exceeded 9.5 mm (preferred
minimum for planting) (Fig. 2).
44
Table 3.--Rooting of shoots from liquid culture
Seedlot Total #
Clone Origin of Shoots % Rooted
5529 80-5B 27 63
55a 80-5B 4 50
5587 81-8U 12 67
5635 81-2U 38 55
5637 81-2U Si7/ VS
5659 81-2U 78 73
5670 81-14B 22 92
5688 81-14B il 90
57.24 81-4U 20 60
5757 81-76-8U 40 as
5780 81-76-8U 16 44
Shoots excised from liquid multiplication culture 5-10 May 1984 and
placed on rooting medium. Observations were made 2-21 August 84.
When the plantlets were lifted from the nursery beds in March 1984, it
was surprising to find a high incidence of plantlets with girdling roots
(Table 5). The girdling was traced back to the placement of the plantlets
into the Can Am plugs. Manipulation during transplanting pushed the roots
into a position promoting girdling root growth. This could similarly have
occurred during transplant into the nursery beds. Plantlets were found
with bent, "S"-shaped stems under the surface, proposed to result from
transplant of the extremely flexible cultured shoots.
Table 4.-- Growth of sweetgum plantlets in cement block nursery beds
Root
Seed Lot Number ie Height Collar Diameter
Clone # Origin Planted Recovered (cm) (mm )
3070 79-6B 96 84 66.3 Tso
3083 79-6B 138 86 Doe 8.0
3090 79-6B 106 83 63.7 Shin Z4
3201 78-1B Sil 78 43.7 6.3
3346 77-10U 34 74 564 Sr
Plantlets were planted between 30 May 83 and 20 June 83, and lifted
on the week of 5 March 84.
Table Se och lant lets pwith girdling roots in the nursery
Total #
Clone Plantlets % Girdling
3070 81 68
3083 119 87
3090 88 93
3201 4O 88
3346 25 80
Evaluation of roots of plantlets lifted the week of 5 March 1984 for
girdling root systems.
45
% IN SIZE GROUP % IN SIZE GROUP.
% IN SIZE GROUP
Figure 1.
rs]
ig
19
s
95 i—} 23 3s 43
STEM LENGTH
23
= CLONE 3090
iE}
10
Ss
°%5 13 23 3 4s t-)
STEM LENGTH
STEM LENGTH
(cm)
73
(cm)
ie)
(em)
20 CLONE 3083
% (IN SIZE GROUP
0
# cy 13 23 c) Er) s [c) 73 «3 Ee)
STEM LENGTH
(cm)
CLONE 3201
% IN SIZE GROUP
= s is oo 3S 6s s « 7s «= os
STEM LENGTH (cm)
2 CLONE 3513
% IN SIZE GROUP
3 is 2 36 2 6 L) am @& 2
STEM LENGTH. (cm)
Stem length of 6 major clones grown in oursery beds.
46
CLONE 3070
% IN SIZE GROUP
1S 3S 8S 7S OS NS 8S
ROOT COLLAR DIAMETER (mm) ROOT COLLAR DIAMETER (mm)
30
25
a. CLONE 3090 CLONE 3201!
Oo 20
S
Ww 18
N
w
Zz 10
#
5
Vs 35 55 75 985 NS 135 os 35 55 75 95 US 135
ROOT COLLAR DIAMETER (mm) ROOT COLLAR DIAMETER (mm)
Qa
>
ro
«
(G)
J
N
wo
2
#
45 35 35 75 #498 WS 135 q\
ROOT COLLAR OIAMETER (mm)
5s 3s ws rs as Ws is
ROGT COLLAR OJAMETER (mm)
Figure 2. Root collar diameter of 6 major clones grown in nursery
beds. Dotted lines indicate the minimum root collar
diameter for planting and the preferred minimum root
collar diameter for planting.
47
Photosynthesis and Anatomy
Net photosynthesis of seedlings and plantlets is shown in Figure 3.
Seedlings showed typical light saturation curves when grown at the different
light levels. High light seedlings had the highest maximum rate of photosyn-
thesis. Photosynthesis of in vitro grown plantlets saturated at higher
levels, with medium light plantlets attaining the highest photosynthetic
rates. Plantlets exhibited much higher photosynthetic rates than seedling
under all conditions.
—— PLANTLET
SEEDLING
HIGH LIGHT
MEDIUM LIGHT
Net photosynthesis (mgco,dm‘ hn" )
LOW LIGHT
o 100 200) 1 300 400
Photosynthetically active radiation (ue m*s"')
Figure 3. Net Photosynthesis of seedlings and plantlets grown under
3 quantum flux densities: _30 + 5 (low), 155 + 10 (medium)
or 315 + 15 (high) pEM “5
Anatomical observations of plantlet leaves showed that chlorplast
structure, leaf thickness, and mesophyll development were affected by
quantum flux differences in culture (2). The higher stomatal densities and
sizes found in cultured leaves were not affected by light levels. Factors
other than light are responsible for the atypical stomatal configurations
which contribute to water loss and wilting of cultured plants.
DISCUSSION
The inclusion of the liquid medium step has greatly increased the
number of shoots obtainable in culture. However, more research is needed
to increase the percentage of cultures that respond in the liquid step.
48
The other obvious problem that needs solution is that of root girdling.
McKeand and Wisniewski (3) reported a similar problem with pines, which was
solved by using shorter roots at planting and a ridged tube instead of a |
pot. This may not be feasible for Liquidambar, however, in that cuticular
development and stomatal functioning are less developed in cultured sweetgum
than pine (8). A well developed root system is expected to be important in
maintaining good water relations (10,11). Rooting in a stationary medium
such as foam or peat plugs may be a possible solution.
We have determined that plantlets developed in vitro are capable of
significant levels of photosynthesis. However, it is unknown if culture
conditions (i.e. CO, and light levels) promote photosynthesis. The efficiency
of our culture systém could be improved if this photosynthetic capability =
were utilized. Further research is needed in this area as are anatomical
observations for evaluating and predicting plantlet growth and culture
efficiency.
ACKNOWLEDGEMENTS
This research was supported, in part, by DOE contract #7860-X02, and
by State and Hatch funds allocated to the Georgia Agricultural Experiment
Station.
REFERENCES
1. Birchem, R., H. E. Sommer, and C. L. Brown. 1981. Scanning electron
microscopy of shoot and root development in sweetgum callus
tissue culture. Forest Sci. 27:206-212.
2. Lee, N. 1984. MS Thesis, University of Georgia.
3. McKeand, S. E., and L. A. Wisniewski. 1982. Root morphology of
loblolly pine tissue culture plantlets. North American Forest
Biology Workshop 7:214.
4. Risser, P. G., and P. R. White. 1964. Nutritional requirements of
Spruce tumor cells in vitro. Physiol. Plant. 17:620-635.
5. Sommer, H. E., and C. L. Brown. 1980. Embryogenesis in tissue cultures
of sweetgum. Forest Sci. 26:257-260.
6. Sommer, H. E. 1981. Propagation of sweetgum by tissue culture.
pp. 184-188. IN Proc., 16th Southern Forest Tree Improvement
Conference, Blacksburg, June 1981.
7. Sommer, H. E. 1983. Organogenesis in woody Angiosperms: Application
to vegetative propagation. Bull. Soc: bot. Fr., Actual. bot.,
130(2):79-85.
8. Sommer, H. E., H. Y. Wetzstein, M. Stine, and N. Lee. 1984.
Differentiation in tissue culture of sweetgum and southern pine.
TAPPI 1984 Research and Develop. Conference. p. 35-37.
49
10.
1h
We
Vanesmuhietta Ae wArLMLcage. vor. kK. Chen). Z. Po Tus and C. iC. Black.
1982. A transient burst of CO, from geranium leaves during
illumination at various light intensities as a measure of photo-
respiration. Plant Physiol. 70:629-631.
Wetzstein, H. Y., and H. E. Sommer. 1982. Leaf anatomy of
tissue-cultured Liquidambar styraciflua (Hamamelidaceae) during
acclimatization. Amer. J. Bot. 69:1579-1586.
Wetzstein, H. Y., and H. E. Sommer. 1983. Scanning electron microscopy
of in vitro-cultured Liquidambar styraciflua plantlets during
acclimatization. J. Amer. Soc. Hort. Sci. 108(3):475-480.
Witham, F. H., D. F. Blaydes, and R. M. Devlin. 1971. Experiments
in plant physiology. p. 195. N.Y.: Van Nostrand Reinhold Co.
50
VEGETATIVE PROPAGATION OF SCOTS PINE (PINUS SYLVESTRIS L.)
THROUGH TISSUE CULTURE
HG. "Sonia Lsaiand bah. Huang!/
Abstract.--A pulse treatment with a high-concentration NAA
solution (125 mg/L) not only enhanced rooting up to 33% but also
increased the number (up to 10 roots/propagule) and size of roots
(2mm in diameter) in cultured Scots pine adventitious shoots. This
induced multiple roots system should increase the vigor of regenerated
plantlets, and hence, shorten the adaptation period while being trans-
ferred to soil.
Additional keywords: Adventitious buds, adventitious roots, pulse
treatment, seedling, embryonic cotyledons.
Tissue culture methods have been evaluated for Scots pine by several
research groups. Tranvan (8) studied the formation, localization of ad-
ventitious buds on seedlings, and the initiation of cotyledon adventitious
buds. Bornman and Jansson (2) attempted to increase the rooting percentage
of four types of explants by applying the growth-active compound, coumarin,
alone or in combination with auxin. Shen and Arnold (7) completed the
culture sequences to regenerate plantlets from cultured embryonic tissue
with an overall regeneration rate of 10% over a 10-month period.
This experiment was directed to improve the survival rate at elongating
stage of adventitious buds and then to promote the rooting percentage of those
elongated adventitious shoots to provide mass, clonal propagules for improve-
ment of a Christmas tree program.
METHODS
Scots pine seeds of Central Massif were chosen from germination tests
as experimental material from among twelve varieties purchased from F.W.
Schumacher Co.
Two types of explants were established from seeds by embryo culture.
Seeds were pretreated with 1% H»05 for 1 week to facilitate the removal of
seed coats and to stimulate germination (6). After the removal of seed
coats, seeds were surface-disinfected with 1/6-strength Clorox solution
for 15 min and then rinsed with autoclaved, distilled water. Embryos were
aseptically separated from endosperms and planted on culture media. Seed-
lings were produced by growing the embryos on 1/3-strength M.S. minimal
organic medium (5), and cotyledonous adventitious buds were initiated on
1/3-strength M.S."B" medium with supplements of 1.0 mg/L kinetin and 30
g/L sucrose. The whorl of cotyledon with a stub of subtended hypocotyl
was excised from 2- to 4-week-old seedlings and planted on the M.S."B"
1/Mrs. Tsai is a research assistant and Dr. F.H. Huang is an associate
professor in the Department of Horticulture and Forestry. University
of Arkansas, Fayetteville, AR 72/701.
51
medium described above (Fig. 1A).
Four adventitious shoots, in addition to the apical shoot, were formed
on the basal area of the cotyledon whorl (Fig. 1B). Those shoot bundles were
separated from the mother explant at 2- to 3-month intervals and subcultured
on M.S."B" medium for continuous multiplication. Cotyledonous adventitious
buds were transferred to 1/6-strength M.S."B" medium to encourage the elonga-
tion of adventitious buds into shoots. Elongated shoots with a visible stem
region were suitable to be rooted (4). Each of two chemical (adenine sul-
fate at 66 mg/L concentration and coumarin at 1.46 mg/L concentration) was
incorporated into the basal medium (1/6-strength M.S."B" medium) separately
to test for effect on rooting. Finally, a pulse treatment with a high con-
centration NAA solution (125 mg/L) was applied for 24 h before transferring
the shoots to medium free of growth regulator (1, 3).
RESULTS AND DISCUSSION
Adventitious shoots were produced in two ways: 1) The cotyledon whorl
excised from 2- to 4-week-old seedling readily produced four adventitious
shoots on the basal area when it was planted on 1/3-strength M.S."B" medium
with supplements of 1.0 mg/L kinetin and 30 g/L sucrose; 2) The adventitious
buds initiated on embryonic cotyledons were transferred to 1/6-strength M.S."B"
medium to encourage the elongation of buds into shoots. Those adventitious
shoots developed on the basal area of the cotyledon whorl can be excised from
mother explant at 2- to 3-month intervals and cultured on the same medium for
continuous multiplication, or they are ready to be rooted.
The adventitious buds initiated on embryonic cotyledons were numerous,
but the elongation of those adventitious buds was sporadic. Efforts to
stimulate the growth of adventitious buds have been without much success
so far. Increase in boron concentration to minimize the phenolic compound
synthesis in order to prevent the stunting of the adventitious buds might
be tried in further experiments.
Adenine sulfate and coumarin incorporation in culture medium did not
show any significant effect over the control. But, those pretreated
adventitious shoots responded quickly to pulse treatment with high-concen-
tration NAA solution. In 3 months, ten adventitious shoots (33%) produced
root primordium with four of them having actual root protrusion and root
growth. The NAA solution pulse treatment not only increased the number of
rooted propagules, but also increased the number and size of root formed
in the individual propagule (Fig. 2A, 2B). A single root is typical in cultured
pine tissue (6), but the above 8-month culture sequences resulted in an increase
in number (up to 10 roots/propagule) and size (2 mm in diameter) of rooting
which should increase the vigor of the regenerated plantlets, and hence,
shorten the adaptation period when being transferred to soil. One regenerated
plantlet from previous culture without the pulse treatment has been trans-
ferred to vermiculite for one year (Fig. 1D). A single rooting of 7 cm long
was observed when being transferred to soil recently (Fig. 1C).
52 -
CONCLUSIONS
Before the survival rate at the elongation stage of cotyledonous adventi-
tious buds can be increased or the embryogenesis method of Scots pine callus
can be developed, the most practical way for vegetative propagation of Scots pine
is through induction of adventitious shoots on the cotyledon whorl and rooting
of those adventitious shoots with NAA solution pulse treatment.
LITERATURE CITED
Amerson, H.V. and R.L. Mott. 1982. Improved rooting of Western White pine
shoots tissue cultures. Forest Sci. 28(4):822-825.
Bornman, C.H. and E. Jansson. 1980. Organogenesis in cultured Pinus
sylvestris tissue. Z. Pflanzenphysiol. Bd., 96(5):1-6. 2
Bornman, C.H. 1984. Application of in vitro culture technology in clonal
forestry. 6 P. 78-194 Mich. "Biotechnolosy, inst... Proc... Int eS vmpriots
Recent Advances in Forest Biotechnology.
Huang, F.H. and S. Tsai. 1984. Regeneration of plantlets from in vitro
cultured tissue of Scots pine (Pinus syiJivestris L.)) Po d54—155. . Utah
State University, 8th North American Forest Biology Workshop.
Huang, L.C. and T. Murashige. 1976. Plant tissue culture media:major
constituents, their preparation and some applications. TCA Manual 3(1):539-548.
Mehra-Palta A., R.H. Smeltzer and R.L. Mott. 1978. Hormonal control of
induced organogenesis experiments with excised plant parts of Loblolly
pine. appa, ol G)ss7—40)
Shen, X.B. and.S. Ve. Arnold. 982. In vitro formation of, adventatrous
plantlets from embryos of Pinus sylvestris L., Scientia Silvae Sinicae,
18(4) :405-408.
Tranvan, H. 1979. In vitro adventitious bud formation on isolated
seedlings of ‘Pinus (sylvestris (i. Biol. Planta. e201 (G)r2o0—23aor
53
Fig. 1 (A) 4-week-old seedling from which apical slices were
excised. Apical slices included the base of the
cotyledonary whorl subtended by a stub of 2mm
hypocotyl.
(B) Adventitious buds induced from apical slices after
4-week culture
(C) Plantlets regenerated in vitro after one month in
the rooting medium.
(D) Plantlets potted in vermiculite.
54
2 (A) 10 roots per plantlet
Fig
iameter root
ah9y ial
(B) 2 mm
55
MICROPROPAGATION OF Eucalyptus viminalis
M. W. Cunningham and R. L. Mot tt!
Abstract.--Cooperators of the North Carolina State University
Hardwood Research Program have selected over 50 Eucalyptus viminalis
trees that demonstrated superior growth rates and tolerance to frost
in test plantings in southern Georgia and northern Florida. Twenty-
eight of these trees have been vegetatively propagated by grafting
or rooting at North Carolina State University. These stock plants
will be micropropagated as in vitro methods are developed, to estab-
lish a seed orchard on lands of Cartén de Colombia.
Results are presented for in vitro multiple shoot production,
using single node explants. Genotype, node selection, and stock
plant vigor influenced axillary shoot production of the original
explants. The induction of multiple shoots on excised axillary
shoots was not affected by the cytokinin concentrations or durations
of treatments tested. When excised axillary shoots were cut into
single nodes, there was a slight enhancement of multiple shoot pro-
duction.
Additional Keywords: tissue culture, vegetative propagation
The need for a hardwood fiber source that would occupy upland sites and
therefore be more readily available during wet seasons led members of the
North Carolina State University Hardwood Research Program to begin screening
seed sources of several species of Eucalyptus in 1971. The objective of
this effort was to select species, seed sources, and trees within seed sources
that would be fast-growing, of good form, and tolerant to both drought and
frost. Over the years more than 100 species and 5/7 seed sources were tested.
Initially, plantings ranged from the coastal plain of North Carolina to north-
ern Alabama. As the program progressed, the plantings were restricted to
southern Georgia and northern Florida because of harsh winter temperatures
north of that area. The best suited species to this area were determined
to be E. viminalis, E. macarthurii, E. nova-anglica and E. camphora.
The failure of any one seed source to consistently produce fast-growing
trees that were also frost-tolerant emphasized the need to develop a land
race of eucalyptus suitable for planting in the southeastern United States.
The seedling seed orchard approach proposed by Purnell and Kellison (1983)
for the genetic improvement of other hardwood species could not be used with
the eucalypts because the species flowers during the winter months when
freezing temperatures destroy the seed-bearing potential of the trees. The
decision was thus made to establish a clonal seed orchard farther south,
where there would be no danger of freezing temperatures. Container Corpor-
ation of America, Fernandina Beach, Florida, in conjunction with one of its
South American subsidiaries, Cartén de Colombia, provided the funding for the
orchard which is to be established near Popoydn, Colombia.
1/
— Graduate Research Assistant, Department of Forestry, and Professor of Botany,
N. C. State University, Raleigh, N. C. (Funding for this project provided
by Container Corporation of America, Fernandina Beach, Florida)
56
Eucalyptus viminalis, the species showing the most promise of the four
mentioned previously, has proven difficult to propagate vegetatively. At-
tempts to root cuttings from induced epicormic shoots from mature trees,
as is done with many other species of eucalypts, have met with minimal success.
Grafting, while more successful than rooting cuttings, is also an unreliable
method of vegetatively propagating the species. Micropropagation under controlled
in vitro conditions may offer a viable alternative as a means to vegetatively
propagate selected trees of E. viminalis.
Vegetative propagules have been produced by micropropagation for a number
of species of eucalyptus (Hartney, 1983). Franclet and Boulay (1982),
Boulay (1983), and Depommier (1981) described methods used by AFOCEL in France
to successfully regenerate plantlets of E. gunnii and E. dalrympleana, using
nodal explants derived from grafted or rooted cutting stock plants. Ages
of the original ortets from which stock plants were derived ranged from 2
to 20 years. The general procedure followed in each of these papers included
(1) induction of axillary shoot formation on single node explants, (2) multiple
shoot formation from excised axillary shoots, (3) shoot elongation, (4) root
initiation of excised shoots, and (5) root elongation and transfer to soil.
Franclet and Boulay (1982) anticipated producing 25,000 plants per month,
using this system.
The objectives of the Hardwood Research Program project are to employ
and modify where necessary the techniques outlined by AFOCEL workers to vege-
tatively propagate 28 clones of E. viminalis. These clones will be shipped
to Popoyan, Colombia, S. A., where they will be used to establish a seed
orchard. The objectives of this paper are to report on the successes in
inducing nodes to produce axillary shoots and the methods tested for multiple
shoot formation from excised shoots.
ESTABLISHMENT AND MAINTENANCE OF CLONES
In 1983, cooperators of the Hardwood Research Program selected a total of
41 E. viminalis, E. macarthurii and E. nova-anglica trees that were phenotyp-
ically superior in growth rate and frost tolerance. Selections were made
from seed source trials and operational test plantings three or more years
old. In addition, 14 surviving E. viminalis clones from a previously estab-
lished seed orchard near Ft. Green Springs, Florida were included, making
a total of 55 selections. Twenty-eight of these selections have been vege-
tatively propagated by rooted cuttings or grafting and are being used as
stock plants for the micropropagation work.
Cuttings were rooted following slightly modified techniques of Campinhos
and Ikemori (1980). Basal epicormic shoots were induced on mature trees
by girdling at a height of one meter. Shoots were cut into four-—leaved, two-
node cuttings, treated with a rooting powder consisting of 0.8% indolebutyric
acid, and placed under intermittent mist. Rooting success was low for these
basal sprouts, averaging less than 15% for all clones. Twelve of the 28
clones were propagated in this way. The remaining 16 clones were grafted,
using scion material from the upper branches.
57
The stock plants were maintained in 4.5-1 pots in the greenhouse with
an extended photoperiod of 18 hours. They were fertilized weekly with a
0.5 g/1 solution of 20-19-18 (N-P-K) fertilizer and sprayed twice weekly with
Benlate, a systemic fungicide. The plants were maintained at a height of
30 - 60 cm by cutting them back at eight-week intervals.
INITIATION OF CULTURES
Initial explants were derived by cutting shoots into single-node sections
and immediately sealing both ends in paraffin. The leaves were then trimmed
to approximately four-mm squares. The explants were surface-sterilized for
10 minutes in a 10% solution of commercial bleach with two drops of surfactant
added. The wax ends were then cut off, leaving an explant of 8 to 14 mm
in length, with 3 to 4 mm above the node. The explants were then placed
vertically into the medium, which was contained in sterilized petri plates.
They were placed in the dark for one week and then moved to a continuous
light environment supplied by two 40-watt, cool white fluorescent bulbs,
approximately 15 cm above the top of the plates. The temperature in the
culture room was 22° + 2° C.
The basic medium used for all experiments consisted of half-strength
salts and vitamins (Murashige and-Skoog, 1962),3.0% sucrose, and 0.7% agar.
The pH of the media was adjusted to 5.6. For the initial nodal explant studies,
0.5 mg/1 of benzylaminopurine (BAP) and 0.01 mg/1 of napthaleneacetic acid.
(NAA) were added.
By the end of the first week in culture, axillary shoots began to emerge
from many of the explants. In most cases callusing occurred at the base of the
explant and on the leaves where they were touching the media. Callusing occa-
sionally occurred in the nodal region but this did not appear to influence
the emergence of axillary shoots. By the third week on the medium, many
of the axillary shoots had begun to elongate, some over one cm in length,
and were ready to be excised.
As with most vegetative propagation methods, the genotype substantially
influenced the response achieved. The responses of three clones that were
used in a number of experiments are summarized in Table 1. Growth regulator
concentrations were varied slightly from trial to trial but all clones were
treated the same within a particular trial, and within a trial the explants
for all clones came from stock plants at the same stage after pruning. Clones
B16 and B6 consistently had response rates of greater than 50%, while explants
from Clone 2594 never responded at rates above 50%.
Table 1.--Percentage of single-node explants producing axillary shoots in
3 - 4 weeks in four different trials
Trial
Clone 1 2 3 4
SS PCR CCT
2594 3355 8.6 8.3 -
B15 50.0 68.6 - 13.0
B6 84.6 60.0 50.0 66.7
For a given clone, the percentage of explants producing axillary shoots
seemed to be influenced by stock plant vigor and node selection. In an experi-
ment to determine which nodes were most useful, explants were grouped according
to their distance from the growing point. The majority of stock plant shoots
were no longer than 6 nodes in length, so explants were divided into three
groups consisting of two nodes each (Nodes 1-2, 3-4, and 5-6, with Nodes 1-2
being closest to the growing point). After three weeks on the initiation
medium, data were recorded for the number of axillary buds and shoots (shoots
were distinguished from buds by the visible presence of at least a 1-mm inter-
node) and the length of the longest shoot. The percentage of explants with
buds or shoots was calculated. Percent and count data were transformed,
using the arc-sine square root and the square root methods, respectively,
for the analysis of variance.
The mean response of three clones after three weeks in culture showed
that explants derived from Nodes 3 and 4 gave the highest response rate and
the greatest number of shoots per plant (Table 2). While fewer explants from
Nodes 1 and 2 responded than those from Nodes 5 and 6, those explants that
did respond had more and longer shoots. This implies that explants from
Nodes 5-6 are slower-growing and probably have many explants with buds that
lave yet to elongate. There were significant clone effects for all variables
but the "clone by node" interaction was not significant, indicating that the
influence of node number was consistent for all three clones tested.
Table 2.--Effect of node number on the response of single-node explants
after four weeks in culture
Percent with Average Shoots Average Length
Node Number Buds or Shoots per Explant Longest Shoot (mm)
V/
1-2 36.1 a— ko 2era'b 3oue
3-4 59/35) D aia Sama
5-6 “7 2 ap 0.6 b Zi le a
ines within a column followed by the same letter were not significantly
different (P = 0.05) using Duncan's Multiple Range Test.
Stock plant. vigor was also found to influence the response of nodal
explants. In two separate trials, explants were collected from the same set
of nodes and placed on the same medium. In one trial, explants were derived
from stock plants that had been repeatedly pruned and fertilized weekly;
while in the other experiment, stock plants were slower-growing, had not been
fertilized on a weekly schedule, and had not been pruned for several months.
For the two clones used in the trials, contamination and mortality rates
were considerably higher and the number of explants producing axillary buds
or shoots was much lower for the less vigorous stock plants (Table 3). These
results emphasize the importance of maintaining pruned and vigorous stock
plants as a source of explant material.
59
Table 3.--Effects of stock plant vigor on the response of nodal explants
after 3 weeks in culture
Stock Percent Percent Percent with
Clone Vigor Contaminated Dead Buds or Shoots
2604 Poor 50.0 25.0 250
Good Ae. AV: 66.7
B15 Poor OS 2 5),0 16.7
Good 8.3 0 66.7
MULTIPLE SHOOT PRODUCTION
Buds and shoots continued to grow when left attached to the original
node, about 3 mm by 3 weeks (Table 2) but reaching 7 to 20 mm by 6 weeks.
However, the explants did not continue to make additional buds. Published
methods require that shoots be excised both for additional bud production
and for growth of shoots in preparation for the rooting stage.
At 3 to 4 weeks, some axillary shoots were sufficiently elongated to
be excised and transferred to a new medium. In preliminary trials, excised
shoots were placed on the same medium as the original explant (0.5 M.S. +
0.5 mg/1 BAP +0.01 mg/1 NAA) under the same environmental conditions. After
2 to 3 weeks, explants began to turn red, older leaves senesced, and the
explant died basipetally. Multiple buds were sometimes formed at the lowest
node near the base of the explant. To determine if cytokinin concentrations
influence this response, excised shoots from two clones were tested on the
original medium, but with either 0.2, 0.5 or 1.0 mg/1 BAP.
Neither clone nor BAP concentrations significantly (P= 0.05) influenced
the number of buds, number of shoots, or shoot length (Table 4). Explants
from all treatments responded the same as those from preliminary studies.
Some of the explants produced clumps of buds at their base, but none of the
buds elongated to more than 5 mm in length during the four weeks. Even at
seven weeks, the average shoot length was 2.7 mm.
Table 4.--Effects of BAP concentration on the response
of excised axillary shoots
Ly
BAP— Average Buds Average Shoots Average Length
(mg/1 ) per Explant per Explant Longest Shoot (mm)
OryZ Dial Ih 72 Ar] |
Ove) Ase) 2 oe)
1.0 29, ORS Desi
iL //
— Analyses of variance showed no significant treatment effects (P = 0.05)
for any of the variables measured.
60
In a second experiment, cytokinin pulses were tested to see if the length
of time the explants were treated with BAP would influence their response.
Shoots were excised from nodal explants of three clones and placed on the
original medium with either 1.0 or 2.0 mg/1 BAP for one, two or three weeks.
The explants were then transferred to a medium (Franclet and Boulay, 1982)
used for shoot elongation. This medium was the same as the original medium
except for a reduced concentration of BAP (0.1 mg/1) and the addition of
1.5% activated charcoal.
Neither the duration nor the concentration of the BAP pulse significantly
influenced the explant responses (Tables 5 and 6). The average number of buds
and shoots was very similar to the results obtained from the BAP concentration
study. Again, explants would respond with red foliage, callusing and senes-
cing from the apex down and the formation of multiple bud clusters at the
base of some of the explants. The use of charcoal appeared to make a slight
improvement in the response. The average shoot lengths of explants in the
pulse experiment (Table 5) were a little longer than those from the BAP concen-
tration study (Table 4). It was noted in previous experiments that when
multiple bud explants were transferred to a medium containing charcoal, the
buds would begin to expand and elongate for two to three weeks. The explants
would then begin to turn pale yellow and callus, probably as a result of the
charcoal tying up needed growth regulators. Shoots seldom reached a length
suitable for rooting.
Table 5.--Effects of length of BAP pulse on the response
of excised axillary shoots
1/
Pulse— Average Buds Average Shoots Average Length
(wks) per Explant per Explant Longest Shoot (mm)
1 3.4 GS 4.3
Zz “oS a) 4.9
3 S29 a2 35)
Wf
—Analyses of variance showed no significant treatment effects (P= 0.05)
for any of the variables measured.
Table 6.--Effects of BAP pulse concentration on the response
of excised axillary shoots
1
pap —/ Average Buds Average Shoots Average Length
(mg/1) per Explant per Explant Longest Shoot (mm)
1.0 S02 eS 4.3
20 308) Ios) 4.3
Uy)
— Analyses of variance showed no significant treatment effects (P= 0.05)
for any of the variables measured.
61
The tendency of excised shoots to die back and produce multiple buds
at the basal node led to a test of using excised shoots that had been further
cut into single-node explants for multiple bud production. The explants
were on a medium with 0.2 mg/1 BAP and 0.01 mg/1 NAA for three weeks. The
group of single-node explants from one shoot tended to produce more buds
and shoots than did comparable entire excised shoots (Table 7). This increase
was the result of more explants producing bud clusters rather than an increase
in the number of buds per cluster. The increased bud number and increased
growth of the buds produced was probably a function of removing the apex
from the explant, thus relieving the apical dominance; however, this hypoth-
esis has yet to be tested.
Table 7.--Comparison of the response of single-node versus entire axillary
shoots as an explant source for multiple shoot production
Average Buds Average Shoots Average Length
Explant Type per Explant per Explant Longest Shoot (mm)
Single Node 4.1 page) 4.6
Entire Shoot Droid. ez 75) 5)
CONCLUSIONS
Nodal explants of E. viminalis could be induced to produce axillary shoots
in vitro. The response varied from clone to clone and was improved by main-
taining vigorously growing stock plants and by selecting the proper node.
Proper greenhouse care to ensure rapidly growing, disease- and insect-free
stock plants seems to be the most critical step for acceptable contamination
rates and for the induction of buds and shoots. Axillary shoots, when left
intact, continued to elongate through six weeks in culture. However, no
additional shoots were initiated on the origianl explant after week four.
Therefore, in order to multiply the number of plants for rooting, it is necessary
to excise axillary shoots and induce multiple shoot formation.
Excised axillary shoots grew poorly, producing few buds, which failed
to elongate. The response of the excised shoots was not influenced by cytokinin
concentrations or the duration of these treatments. The addition of charcoal
to the media was also ineffective in enhancing shoot elongation.
Boulay (1983) reported similar problems with several frost-tolerant
eucalyptus species. He stated that for many clones, frequent subculturing
was necessary before the explants would begin to grow vigorously and produce
multiple shoots. We believe that this enhanced response after many subcultures
probably resulted from better control of shoot vigor and/or node development
stage of the in vitro stock plants used for serial subcultures. The influence
of in vivo stock vigor and node number on the response of original explants
found in our studies supports this hypothesis. From our preliminary data
it appears that we can enhance multiple shoot production by selecting single
nodes from in vitro-produced axillary shoots rather than using the entire
shoot as an explant. However, the buds produced still did not elongate suf-
ficiently for rooting purposes. We will be further investigating the effects
of node selection of in vitro- gpreduec’ BSL EY shoots to enhance multiple
shoot production.
62
Multiple shoot production and shoot elongation cannot occur under the
same in vitro conditions for E. viminalis. Future efforts will also be
directed toward development of a separate medium for shoot elongation of
multiple shoot explants and eventually the rooting of elongated shoots.
LITERATURE CITED
Boulay, M. 1983. Micropropagation of frost-resistant eucalypts. Proc.,
Workshop on Eucalyptus in California. USDA For. Serv. Gen. Tech. Rept.
PSW-69. pp. 102-107.
Campinhos, E. and Y. K. Ikemori. 1980. Mass-production of Eucalyptus species
by rooting cuttings. IUFRO Symp. on Genetic Improvement and Productivity
of Fast-Growing Trees. 16 pp.
Depommier, D. 1981. Micropropagation d' Eucalyptus resistant au froid.
IN Colloque International Sur La Culture In Vitro Des Essences Forestiers.
Pe LAT asy2 &
Franclet, A. and M. Boulay. 1982. Micropropagation of frost-resistant
eucalypt clones. Aust. For. Res. 13:83-89.
Hartney, V. J. 1983. Tissue-culture of Eucalyptus. In Proc., Intl. Plant
REOp. 2 SOC. 152 195-1098
Murashige, T. and F. Skoog. 1962. A revised medium for rapid bioassays
with tobacco tissue cultures. Physiol. Plant. 15:473-497.
Purnell, R. C. and R. C. Kellison. 1983. A tree improvement program for
southern hardwoods. IN 1/th South. For. Tree Imp. Conf., Athens, Ga.
pp- 90-98.
63
GROWTH CHANGES IN LOBLOLLY PINE (PINUS TAEDA L.)
CELL CULTURES IN RESPONSE TO DROUGHT STRESS
R. J. Newton, S. Sen! , and Ji.) Pe van Buijtenen2
Abstract.-- Two in vitro systems; callus proliferation from
(1) cotyledons and (2) pre-existing callus, were used to evaluate
growth responses to low tissue water potential among two loblolly pine
sources including 6 families. Both systems showed significant
differences in growth response to low tissue water potential between
sources from Louisiana and Texas. Both systems also showed significant
differences between a fast-growing family from Louisiana (A-1-14) and a
slow growing family from Texas (GR1-8). Louisiana sources sustained
more growth under in vitro drought stress than did Texas sources. The
minimum tissue water potential below which callus growth was halted was
-1.0 MPa. These preliminary results suggest that in vitro drought
stress techniques may have applicability for predicting growth
potential in the field under stressed and nonstressed conditions and
investigating drought tolerance mechanisms which appear to be distinct
and separate from drought avoidance mechanisms. Additional research is
in progress to confirm these results.
Additional keywords: Tissue culture, polyethylene glycol, callus,
water potential, drought tolerance.
Several investigators have shown that there are differences in loblolly
pine (Pinus taeda L.) source responses to drought. Most of these studies
have been concerned with drought avoidance whereby plants tolerate drought
by maintaining a high tissue water potential. Whole plant studies have
shown that increased root growth (Youngman 1965, van Buijtenen et al. 1976,
Bilan et al. 1978, Cannell et al. 1978), reduction in epidermal conductance
(Thames 1963, Knauf and Bilan 1974, van Buijtenen et al. 1976, Bilan 1977)
and reduction in evaporative surfaces (Wells and Wakeley 1966, Wright and
Bull 1968, Woessner 1972a,b; Venator 1976) are mechanisms by which high
tissue water potentials are maintained in loblolly pine resulting in drought
avoidance. Only a few studies have investigated the mechanisms whereby
drought is tolerated at low tissue water potential. These whole plant
lAssociate Professor and Research Associate, Department of Forest
Science and the Texas Agricultural Experiment Station, Texas A&M
University, College Station, TX 77843.
2Professor, Department of Forest Science and the Texas Forest Service,
Texas A&M University, College Station, TX 7/7843.
The authors wish to acknowledge the assistance of Drs. R. H. Smith and
S. Bhaskaran, Department of Soil and Crop Sciences, Texas A&M
University, in preparing this manuscript.
64
studies have shown that solutes accumulated (Hodges and Lorio 1969) and that
growth potential was maintained through osmotic adjustment (Hennessey and
Dougherty 1984) as low tissue water potentials were experienced. Other
workers have shown differential tolerance to low tissue water potential
among loblolly pine families (Kaloyereas 1958, van Buijtenen et al. 1976,
Newton and van Buijtenen 1984) when whole plants were subjected to
desiccation.
A recently developed technique for investigating drought tolerance
at low tissue water potential is in vitro tissue culture. Tissue culture
has several advantages over whole plants: (1) drought avoidance mechanisms
such as increased -rooting, decreased stomatal conductance and leaf drop can
be dismissed, (2) cellular mechanisms of drought tolerance at low water
potential can be investigated, and (3) it could be used to screen a large
number of different sources of germplasm for drought tolerance in a short
period of time and in a small space. Therefore, if the suitability of
tissue culture as a selection tool for drought tolerance could be
demonstrated, it could aid existing tree improvement programs, particularly
those in the western edge of the loblolly pine region.
The first objective of this investigation was to evaluate two tissue
culture systems for their suitability in differentiating between loblolly
pine sources in their growth response to low tissue water potential. To
accomplish this objective, seed sources from Louisiana and Texas were
compared. Based on previous, percent-survival data in response to drought,
the Louisiana sources have been designated as drought-susceptible and the
Texas sources as drought-hardy (Newton and van Buijtenen 1984).
Drought—-hardy and drought-susceptible families have been selected for
avoidance and maintenance of high tissue water potential rather than
tolerance of low tissue water potential. We wanted to evaluate them
for drought tolerance at low water potential by exploiting the unique system
of cell culture.
The second objective was to determine the tolerance limit of callus
growth in loblolly pine. That is, we wanted to determine minimum water
potentials whereby growth could be sustained when callus was subjected to
drought stress. This information would be important for comparison with
other callus systems, particularly crop plants and other woody species.
Furthermore, drought responses of cell cultures have not been previously
reported for loblolly pine or for any other woody plant species. The data
presented here show that in vitro cultures can be used for determining
differences between loblolly pine source tolerance to low water potential
and that cell growth is not sustained at water potentials lower than -].0
MPa.
MATERIALS AND METHODS
Six half-sib loblolly pine families from known sources were used.
Three of the families (BA3R13-41, BA3L11-1, GR1-8) were taken from the
western edge of the loblolly pine range in Texas (TX) and three families
65
(A-1-4, A-1-7, A-1-14) were obtained from Crown Zellerbach Corporation and
originated from Louisiana (LA). Explant and media procedures were modified
from Mott and Amerson (1982). Seeds were placed in sterile flasks
containing 1% H207 at room temperature (27°C) to stimulate germination
(Ching and Parker 1958). After 4-5 days, seeds were further surface
sterilized in 15% clorox solution for five minutes followed by three rinses
of sterile water. Embryos were then aseptically removed and the cotyledons
were excised as explants. In one in vitro system, the cotyledons were
inoculated onto the experimental liquid media and allowed to produce callus
(cot->cal) and in the second system, excised cotyledons were first placed on
a complete agar media, allowed to proliferate callus for four weeks and then
callus was inoculated onto the experimental media (cal->cal). Both the agar
and the experimental liquid media were amended with napthalene acetic acid
(NAA) and benzylamino purine (BAP) (see RESULTS). The cotyledons or callus
were inoculated on Heller supports in test tubes containing Gresshoff—Doy
(GD) (1972) liquid media with polyethylene glycol (PEG, Mol. Wt. 8000,
Sigma) added to provide varying water potential (jw). The media water
potential was determined with a microvoltmeter (Wescor HR 33T) and
thermocouple psychrometer chamber (Wescor C52). The standard media without
PEG had a water potential of -0.4 MPa. Six sets of PEG media were prepared
with a water potential of -0.6, -1.0, -1.3, -1.8, -2.3 and -2.7 MPa. [In the
cot->cal system the initial cotyledon explant weights were between 1.1
and 1.5 mg, and there were no significant differences between families
in regard to their initial weights. In the cal->cal system, 80 to 90
mg of callus tissue were placed into each tube. The inoculated tissues
were subjected to drought stress for eight weeks in a controlled environment
chamber with a temperature of 21°C and continuous light (fluorescent and
incandescent) with an intensity of yEm2sec7!. Each treatment was
replicated ten times. Test tubes were placed in a randomized block design
and the position of each rack of tubes was randomly arranged every 24 hours
to ensure equal treatment. After eight weeks the resulting callus was
frozen in liquid nitrogen, lyophilized, and the dry weights recorded.
All data were analyzed with Analysis of Variance and Duncan's Multiple
Range Test.
RESULTS
It was important first to determine which combination of napthalene
acetic acid (NAA) and benzyl-amino-purine (BAP) would provide the most
callus growth. One mg/1NAA and 3mg/1BAP resulted in maximum growth of
callus after 8 weeks on a GD-agar medium. This modified medium was used
thereafter as the standard medium for all subsequent experiments.
In vitro fresh weight growth of callus from cotyledons (cot->cal)
and from pre-existing callus (cal->cal) over an eight week period in
normal media with no PEG added are shown in Table 1. The water potential
was -—0.4 MPa. In both systems, family A-1-14 increased in fresh weight more
than the other five families and family GR-1-8 had the least growth. Family
rankings based on mean fresh weight were also similar in both systems with
66
the exception of BA3R13-41 and A-l- 4 which were reversed.
TABLE 1. In vitro fresh weight growth of 6 loblolly
pine families at a water potential of -0.4
MPa.
Mean Fresh Weight (mg)
Family Source Cot->Call Cal->Cal!
A-1-1 LA 58.0 a2 263.8 a2
Nei les]/ LA 54.9 a 237.8 ab
A-1-4 LA 43.3 ab W/SoS) Zio)
BA3R13-41 TX 42.8 ab 184.8 ab
BA3L11-1 TX 29.5 ab 128.4 ab
GR-1-8 TX hou’ fp 10829) sib
1 10 replicates per family
means sharing a common letter are not significantly
different at an alpha level of 0.05
Cultures from each family were next tested for their capacity to
grow when PEG was added to the media. Cot->cal and cal->cal systems
were both used with water potentials ranging from -0.6 to -2./7 MPa and
-0.6 to -2.3 MPa for the two systems, respectively. The overall mean
cot->cal fresh weight at 6 different water potentials was the largest
for family A-1-14 and was smallest for family GR1-8 (Table 2). The overall
mean cal->cal fresh weight at 5 different water potentials was also larger
for A-1-14 and smaller for GRI-8 (Table 2). Growth of GRI-8 was 60 and 40%
less than growth of A-1-14 in the cot->cal and cal->cal systems,
respectively. Fresh weight growth under stressed and nonstressed conditions
was significantly different between these two families (Table 1,2).
Analysis of Variance showed significant differences at the 6% level between
sources in response to drought stress (Table 2). Louisiana sources
sustained more growth under drought stress than did Texas sources.
67
TABLE 2. In vitro fresh weight growth of 6 loblolly
pine families averaged over all water potentials.
Mean Fresh Weight (mg)
Family Source Cot—>Call 54 Cal->Cal2>4
A-1-14 LA ie a> 93.7 a3
A-1-7 LA 10.9 a 84.3 ab
A-1-4 LA 9.4 ab 66.9 be
BA3R13-41 TX 8.5 ab 67.3 be
BA3L11-1 TX 8.2 ab D079 Cc
GR-1-8 TX 4.5 b 55 0me
| 6 different water potentials with 10 replicates per
family
5 different water potentials with 10 replicates per family
3 means sharing a common letter are not significantly
different at an alpha level of 0.05
sources are significantly different at an alpha level of 0.06
It was most meaningful to determine the minimum tolerance levels
for in vitro growth by loblolly pine since this has not been reported
earlier. This was accomplished by pooling the family fresh weights for
each water potential and comparing the overall means (Table 3). Mean
fresh weight growth of the cot->cal system was 63% less than the control
when subjected to drought stress at -0.6 MPa and 96% less than the control
at -1.0 MPa. (Table 3). There was no growth at water potentials lower than
-1.0 MPa. Fresh weight of the cal->cal system was decreased by 64% at -0.6
MPa and 74% at -1.0 MPa compared to the control. However, final fresh
weight of the cal->cal tissue was smaller than the original inoculum fresh
weight at all stress treatments; only the callus at -0.4 MPa was larger than
the initial callus fresh weight (Table 3). At water potentials less than
-1.0 MPa the fresh weight of both systems remained relatively constant.
The minimum tolerance level for cot->cal growth was -1.0 MPa and was -0.6
MPa for cal->cal growth (Table 3).
68
TABLE 3- In vitro fresh weight growth of loblolly pine
as influenced by decreasing water potential.
Mean Fresh Weight (mg)
yw(MPa ) Cot->Call Cal->Call
-0.4 41.6 a2 182.9 a2
-0.6 Ihsyoaleenb 66.4 b
-1.0 1.6c¢ 47.9 be
silos ese 45.9 be
-1.8 SO ie syoll ©
— 255} LONE 3705 ©
hs 7) 0.8 c
1 yn = 60 with 10 replicates per family
2 means sharing a common letter are not significantly
different at an alpha level of 0.05.
Dry weight decreased in the cot->cal system when subjected to drought
stress, but cal->cal dry weight remained constant at all stress levels
(Table 4). Dry weight of the cot->cal system was decreased by nearly 90% at
-1.0 MPa compared to cultures at -0.4 MPa. Therefore, the decrease in fresh
weight of cot->cal (Table 1) was due to both water loss and dry weight
decrease (Table 3) whereas, the fresh weight decrease of cal-—>cal was
primarily water loss (Table 1) with very little change in dry weight (Table
3)
DISCUSSION
Two in vitro culture systems were used to compare growth between
loblolly pine families and sources. Both systems showed significant
differences between fast-growing and slow-growing sources during drought
stress. The family ranking of mean fresh weight growth under nonstressed
conditions (Table 1) was the same as their ranking under stressed conditions
(Table 2). Family A-1-14 from Louisiana consistently grew better under
nonstressed and stressed conditions compared to the other 5 families and
family Gl-8 from Texas was a consistent, poor performer. These
69
TABLE 4. In vitro dry weight growth of loblolly pine as
influenced by water potential.
Dry Weight (mg)
Ww(MPa) Cot->Cal! Cal->call
-0.4 7.2 a2 20.2 a2
-0.6 4.4 b 226 3ua
-1.0 0.8 c 20.5 a
-1.3 Dinsia
-1.8 2.3 a
-2.3 20.7 a
1 n = 60 with 10 replicates per family
2 means sharing a common letter are not significantly
different at an alpha level of 0.05
data indicated that fast growing families under nonstressed conditions also
perform well under stressed conditions and that Louisiana sources grew
significantly better than Texas sources. Although preliminary, these data
also show that source tolerance responses to drought may be different from
source avoidance responses.
Some interesting comparisons with field observations can be made here.
Louisiana sources are known for their rapid growth under field conditions
(Yeiser et al. 1981). Under drought stress, however they suffer excessive
mortality (Zobel and Goddard 1955), although surviving trees may grow quite
well for a period of time. Mortality of Louisiana source outplantings in
drought prone areas decreases per acre yields compared to seed sources
selected for drought resistance (van Buijtenen, unpublished). It will be
most interesting to pursue the possibility that in vitro growth is
indicative of growth rate observed under field conditions.
Callus proliferation from cotyledons appeared to be very limited
if the water potential was less than -1.0 MPa. This is similar to the
level of drought tolerance of other in vitro systems such as grain sorghum
(Newton et al., submitted). At water potentials less than -1.0 MPa, the
fresh weight of the cot->cal cultures (0.8 to 1.6 mg) were not significantly
different from the weight (1 mg) of the initial, inoculated cotyledons
70
(Table 3). Control medium in these experiments was at a water potential of
-0.4 MPa; higher water potentials were not tried. It would be most helpful
to determine if growth is increased at these higher water potentials.
Even though callus fresh weight produced from callus was severely
reduced by the slight stress of -0.6 MPa, the dry weight of the cal-—>cal
system was not affected (Table 4), indicating that the tissue was becoming
more dehydrated as it experienced more drought stress. For example, callus
tissue at a water potential of -0.4 MPa contained 160 mg of water with a dry
weight of 20.2 mg whereas tissue at -1.0 MPa had 27 mg of water with a dry
wt of 20-5 mg: (tabile? 35.4).
In conclusion, Louisiana sources sustained more growth under in vitro
drought stress than did Texas sources. The minimum tissue water potential
below which callus growth ceased was -1.0 MPa. In vitro drought stress
techniques may have applicability for predicting growth potential in the
field under stressed and nonstressed conditions. Furthermore, the data
indicate that this technique may be suitable for investigating drought
tolerance mechanisms which appear to be distinct and separate from drought
avoidance mechanisms. Additional research is in progress to confirm these
preliminary results.
LITERATURE CITED
Bilan, M. V., Hagan, C. T., and Carter, H. B. 1977. Stomatal opening,
transpiration, and needle moisture in loblolly pine seedlings from two
Texas seed sources. Forest Sci 23: 457-462.
Bilan, M. V., Leach, J. He, and Davies, G. 1978. Root development in
loblolly pine (Pinus taeda L.) from two Texas seed sources. [In] “Root
Form of Planted Trees” (E. van Eerden and J. M. Kinghorn, eds.), Joint
Rep. No. 8, ppe 17-22. British Columbia Ministry of Forests/Canadian
Forestry Service, Victoria, British Columbia.
Cannell, M. Ge. R., Bridgewater, F. E., and Greenwood, M. S.~ 1978. Seedling
growth rates, water stress responses and root-shoot relationships
related to eight-year volumes among families of Pinus taeda L. Silvae
Genet 27:237-248.
Ching, T. M. and Parker, M. Ce. 1958. Hydrogen peroxide for rapid viability
tests of some coniferous tree seeds. Forest Sci 4(2):128-134.
Gresshoff, P. Me and Doy, C. H- 1972. Development and differentiation
of haploid Lycopersicon esculentum (tomato). Planta (Berlin)
107:161-170.
71
Hennessey, T. C. and Dougherty, T. M. 1984. Characterization of the
internal water relation of loblolly pine seedlings in response to
nursery cultural treatment: Implication for reforestation success.
Seedling Physiology and Reforestation Success. (Mary L. Duryea and
George N. Brown, eds.), Martinus/Dr. W. Junkie Publishers.
Ppa 225—243;6
Hodges, J. D. and Lorio, P. Le 1969. Moisture stress and composition
of xylem oleoresin in loblolly pine. Forest Sci 21:283-290.
Kaloyereas, S. Ae 1958. A new method of detemining drought resistance.
Plant Physiol 33:232-233.
Knauf, T. A. and Bilan, M.- V. 1974. Needle variation in loblolly pine
from mesic and xeric seed sources. Forest Sci 20:89-90.
Knauf, T. A. and Bilan, M. Ve 1977. Cotyledon and primary needle variation
in loblolly pine from mesic and xeric seed sources. Forest Sci
ZBI 3—3Ol.
Mott, Re L. and Amerson, H. V. 1982. A tissue culture process for the
clonal production of loblolly pine plantlets. N.C. Agric
Res. Serv. Tech. Bul. No. 2/71.
Newton, R. J. and van Buijtenen, J. P. 1984. Evaluation of stress
resistance of loblolly pine with seedlings in controlled environment
chambers and tissue culture. TAPPI Research and Development Division
Conference. Appleton, WI, Sept. 30-Oct. 3.
Thames, Je. Le 1963. Needle variation in loblolly pine from four geographic
sources. Ecol 44:168-169.
van Buijtenen, J. P., Bilan, M. V. and Zimmerman, R. H. 1976.
Morpho-physiological characteristics related to drought resistance in
Pinus taeda L. [In] Tree Physiology and Yield Improvement
(M. G. R. Cannell and F. T. Last, eds.). Academic Press, New
York. p. 348-359.
Venator, C. R. 1976. Natural selection for drought resistance in Pinus
caribeae Morelet. Turrialba 26:381-387.
Wells, C. 0. and Wakeley, P. C. 1966. Geographic variation in survival,
growth, and fusiform rust infection of planted loblolly pine. Forest
Sci Monogr 11, 40 pp.
Woessner, R. A. 1972a. Crossing among loblolly pines indigenous to
different areas as a means of genetic improvement. Silvae Genet
2335-39).
72
Woessner, R. A. 1972b. Growth patterns of one-year-old loblolly pine seed
sources and inter-provenance crosses under contrasting edaphic
conditions. Forest Sci 18:205-210.
Wright, J. W. and Bull, W. T. 1963. Geographic variation in Scotch
pine: results of a 3-year Michigan Study. Silvae Genet 12: 1-25.
Yeiser, J. L., van Buijtenen, J. P., and Lowe, W. J. 1981. Genotype x
environment interactions and seed movements for loblolly pine in the
Western Gulf Region. Silvae Genet. 30:196-200.
Youngman, A. Le 1965. An ecotypic differentiation approach to the study of
isolated populations of Pinus taeda in south central Texas.
Ph.D. thesis, University of Texas, Austin: Diss. Abstr. Int. B. 27,
3006 (1967).
Zobel, Be J. and Goddard, G. E. 1955. Preliminary results on tests of
drought hardy strains of loblolly pine (Pinus taeda) L. Research Note
No. 14, Texas Forest Service. 22 pp.
73
CONIFER GENETICS |
LONGLEAF PINE
MODERATED BY MS. PAT LAYTON
University of Florida
74
TECHNIQUES FOR SUCCESSFUL ARTIFICIAL REGENERATION OF LONGLEAF
PINE
MARC G. ROUNSAVILLE./
Abstract.--Refinement of longleaf pine (Pinus palustris)
artificial regeneration techniques over the past ten years on
the Black Creek Ranger District has resulted in a five point
program for success.
Longleaf pine has been reduced to less than 10% of its original range.
This reduction is partly due to the problems and failures associated with
artificial regeneration. The U.S.Forest Service aggressively attacked these
problems in the early to mid-1970's. Five areas were identified as keys to
improved survival.
METHODS
The five keys to successful artificial regeneration of longleaf are:
1. Well prepared sites; 2. Large, healthy, fresh seedlings; 3. Proper care
and handling of planting stock; 4. Proper planting procedure; 5. Post
planting care and management.
The site preparation method used on the
Black Creek was usually shear, rake and disk. Other methods used are drum
chopping, burning and sometimes disking. The method of choice is the one
that will take control of the site and will allow access for the planting
machines at the least cost; soils, topography, and the amount and kind of
vegetation to be controlled also influence the choice of site prep method.
Herbicides will replace disking for control of root competition in FY 1986.
Adequate survival
cannot be expected with seedlings less than 0.4" root collar diameter. The
length of storage also impacts survival. The seedlings grown at Ashe
Nursery are graded to 0.4" RCD. White in his 1978 study on length of
storage, RCD and their affect on survival showed that acceptable survival
could not be expected with trees less than 0.4" RCD and those seedlings
with RCD's in the 0.4" class would not survive if stored for three weeks or
more. Brownspot needle blight control is enhanced by the use of Benlate at
the nursery.
Seedling storage on the Black Creek is limited to less than one week
by making seedling orders small and frequent. The nursery coordinates
lifting with the districts and destroys seedlings after they have been in
storage for ten days. Planting large seedlings will shorten the time
seedlings remain in the grass stage since height will not start until the
RED Sis Ol MacomOMs
Vsiiviculture Assistant, Black Creek Ranger District, National Forests
in Mississippi, USDA Forest Service, Wiggins, Mississippi.
75
Contracts on the
Black Creek require the contractor to use an insulated storage box to
transport the seedlings and store them in on the planting site. The box
not only reduces exposure but also protects the seedlings from
contamination by fuel, oil or other substance that may be in the bed of a
truck. Seedlings are picked up daily from cold storage and any trees left
at the end of the day are returned to cold storage.
Machine planting reduces the chances of exposure when compared to hand
planting. Close administration of the contract ensures that the contractor
meets the care and handling requirements of the contract.
4 Proper planting procedure. The planting contracts require that
the seedlings be planted within 0.5" of the depth that they were grown in
the nursery. Seedlings that are planted deeper than they were grown may
survive for one to three years but will be much slower in intiating height
growth than those planted at the correct depth. The seedlings planted too
shallow will expose roots making them more susceptible to drought and fire.
The planting machine generally pushes up a small berm around the
seedling as it is packed which will wash away from the seedling. The depth
of planting should be based on the location of the root collar after the
berm has settled away from the seedling.
Almost any planting machine in good working order Will dow
satisfactory job of planting longleaf. The machines used on the Black
Creek are Reynolds double coulter machines.
The planting rated should be based on past survival and the number of
trees per acre that achieve height growth in three to four years. One
thousand trees per acre were planted until the 1984-1985 planting season.
This has been reduced to 850 trees per acre for the 1985-1986 planting
season.
Annual checks are made using
0.01 acre plots to determine survival, brownspot infestation, number of
height growth seedlings and release needs. These plots are installed in
grid fashion to yield a 1% inventory. The prescription for the stand is
based on the information collected during this inventory. Brownspot
control burns are made as needed in winter under conditions that will yield
a "cool" burn. Normally this would mean burning with a head fire one to
two days following frontal passage.
Cattle should be excluded from plantations until 300 trees per acre
have achieved height growth. Fencing is the method normally used.
RESULTS
Prior to initiating these measures survival ranged from 57 to 67
percent on 3666 acres planted from 1973 to 1977 with the average for these
years being 62 percent. This was not considered acceptable since the
failure rate for these plantations ranged from 40 to 100 percent for this
same period. These five "keys" were gradually phased in starting in 1973
as they were developed and refined (which continues today). The
improvement in survival and plantation success rate is illustrated by the
76
results of the fiscal year 1984 planting season and the absence of
plantation failures since fiscal year 1980. The survival for 793 acres
planted in fiscal year 1984 was 97 percent.
Longleaf pine plantations can be established by paying close attention
to these five keys and using good basic tree planting techniques.
LITERATURE CITED
White, J. B. 1979. Longleaf Pine Survival Influenced by Seedling Size. In
Proceedings of the Longleaf Pine Workshop, p. 26-29. USDA Forest Service,
S&PF, SA, Technical Publication SA-TP-3.
77
LONGLEAF PINE TREE IMPROVEMENT IN THE WESTERN GULF REGION
T. D. Byram and W. J. Lowe WY
Abstract.--Longleaf pine collected from North Louisi-
ana, Southeast Texas, South Louisiana and South Mississippi
were outplanted in seven locations in the Western Gulf area.
Family heritabilities across locations were 0.56 for second
year survival, 0.54 for grass stage emergence and 0.72 for
brown-spot needle blight resistance. Coefficients of genet-—
ic prediction indicated a positive relationship between
survival and grass stage emergence but not between these
traits and brown-spot resistance. Tentative conclusions
based on one year's plantings indicate that North Louisi-
ana, Southeast Texas, South Louisiana and South Mississippi
can be considered one breeding zone for improving survival.
Southeast Texas, South Louisiana and South Mississippi can
be considered one breeding zone for improving emergence
from the grass stage and brown-spot resistance.
Additional keywords: Pinus palustris, Scirrhia acicola, genotype
by environment interaction, survival, grass stage emergence.
In recent years, increased interest has been shown in using long-
leaf pine (Pinus palustris Mill.) in artificial regeneration programs
in the Western Gulf Region. This stems from 1) the increasing value of
poles, pilings and other solid wood products, 2) the increasing losses
to fusiform rust (Cronartium quercuum [Berk.] Miybe ex Shirai f. sp.
fusiforme) on slash pine (Pinus elliottii Engelm. var elliottii), 3) an
increasing emphasis on planting the proper species on appropriate
sites, and 4) longleaf's suitability for planting in high fire hazard
areas.
Many of the problems traditionally associated with the establish-
ment and early growth of longleaf pine are manageable by improved nurs-
ery techniques, seedling care and silvicultural practices (Shipman
1960, Smith and Schmidtling 1970). Traits related to establishment are
also under genetic control. Family heritabilities for early survival
were estimated as 0.73 at one location (Rockwood and Kok 1977) and 0.35
across a wide range of environments (Goddard and Bryant 1981). Goddard
and Bryant calculated that selecting the top one-half of the families
in their study resulted in a 6.5 percent gain in survival. Family heri-
tabilities for height initiation at two years of age across several
environments ranged from 0.47 to 0.68 (Layton and Goddard 1982).
1/Assistant Geneticist, Western Gulf Forest Tree Improvement Program,
Texas Forest Service, College Station, Texas; and Associate Geneticist,
Texas Forest Service, and Assistant Professor, Texas Agriculture
Experiment Station, College Station, Texas. The authors wish to
acknowledge the members of the Western Gulf Forest Tree Improvement
Program working with longleaf pine for their efforts to establish,
maintain and measure the test plantings reported in this study.
78
Snyder and others (1977) reported heritabilities for height initiation
of 0.48 and 0.52 at three years of age. They also reported family
heritabilities for brown-spot needle blight (Scirrhia acicola [Dearn.]
Sigg.) resistance of 0.30 and 0.57.
Goddard and others (1973), and Snyder (1969) noted that selection
of phenotypically superior plus trees was not particularly effective in
improving juvenile traits and recommended a two step testing program.
The Western Gulf Forest Tree Improvement Longleaf Pine Program is
similar to the two step selection procedure developed by the Florida
Cooperative Forest Genetics Program (Goddard and others 1973, Goddard
and Rockwood 1978).
Western Gulf Longleaf Pine Program
Approximately 100 low intensity selections will be made from each
of four provenances - Southeast Texas, South Louisiana, North Louisi-
ana, and South Mississippi. The 400 selections will be included in
short duration tests (three years) to evaluate juvenile traits. Fami-
lies that exhibit acceptable performance will be established in long
term, good growth and form progeny tests. Both stages of the testing
program are designed to determine the effects of different provenances,
the relative amount of family variation, and the presence of genotype
by environment interactions.
Provisions have been made to meet seed needs with the establish-
ment of seedling seed orchards concurrently with the establishment of
the short term tests, in conjunction with the long term tests or by
establishing clonal orchards based on the long term test results.
This paper presents the second year data from the first plantings
of the short-term tests. The objectives are to estimate heritabilities
and examine genotype by environment interactions for survival, grass
stage emergence and brown-spot resistance.
Materials and Methods
In the spring of 1982, 100 families plus four bulk checklots were
sown in three nurseries for outplanting at seven locations (Figure 1).
Seedlings were sown at an initial density’ of nine per square foot and
grown according to standard nursery procedures for longleaf pine. The
two North Louisiana plantings were grown at one nursery, the two South
Mississippi plantings were grown at a second nursery and the two South-
east Texas plantings and the South Louisiana planting were grown at a
third. Sixty-four of the 100 families were common to all seven
locations and ninety-three families were in at least five of the seven
plantings.. Field design consisted of eight replications at each
location with four or five trees per row plot depending on the planting.
Spacing was two by 10 feet at four locations, three by nine feet at two
locations, and two by eight feet at a single location. Sites ranged
from dry sand ridges to poorly drained flatwoods.
79
At the end of the second growing season survival, percent of liv-
ing trees initiating height growth, and percent of foliage infected by
brown-spot needle blight were scored. Brown-spot infection was scored
on a 0 to five scale with 0 representing a brown-spot free individual.
Higher scores represented the amount of foliage infected in 20 percent
increments.
Bienville &
8 LaSali
TX
Tyler a a” Stone
Hardin ® Pearl River
Figure 1. County/parish locations of the short-term
longleaf pine plantings.
All percent data was transformed by the arc sine of the square
root and each location was analyzed separately. The 64 families common
to all seven locations were combined in one analysis across locations
to examine the relative amount of genotype by environment interaction.
Variance components were calculated for family within provenance and
provenance effects to determine the relative importance of geographic
variability in selection. Family heritabilities for survival, percent
height initiation, and brown-spot severity were calculated at each of
the locations with significant family effects as well as for the com-
bined analysis according to the following formulas:
One location Multiple locations
627 625
677 + 6B/y 67 + 67R(p)*L/1 + 6°E/Ir
Whe re
625 = variance among family means within provenances
80
627 (P)*L = variance among families within provenance by
planting location means
= error variance
number of locations
number of replications
62
1
ie
The coefficients of genetic prediction (CGP) were used to examine
the relationship between traits (Baradat 1976). They were also calcu-
lated for the same trait across locations to examine the amount and
direction of genotype by environment interaction.
RESULTS AND DISCUSSIONS
Single Locations
Second year planting survival ranged from 26 to 95 percent (Table
1). There were differences between families within provenances in six
of the seven locations. Family heritability for survival varied from
0.21 to 0.55. -Low heritability estimates at the Tyler and Hardin Coun-
ty, Texas, tests are primarily caused by lack of variation due to uni-
formly good survival. ;
Table 1. ULocations, averages, family heritability estimates, and
standard errors for survival, grass stage emergence and
brown-spot severity at seven plantings of two-year-old
longleaf pine in the Western Gulf region.
Trait
Location Grass Stage Brown-spot
County/ Parish Survival Emergence Severity Code
State
2) Gohet, SE i h2+ SE x h2 SE
Stone, MS 78 E16 58 sore le 134) 704 715
Pearl River, MS 32D tameel 45 ~ 23+ 18 On67 3495518
Tyler, TX Oreste LO 44 40+ .16 2288 69+ .15
Hardin, TX O55 29+ Fale 93 SSS 09) Wa23cb re 6
Vernon, LA 72 aie 72) «2 2b neh JOD!) iS 2 cel'6
Bienville, LA We Sie ols 64 SSS 250
LaSalle, LA 75 ~ 28+ .16 82 oS ot Oa
Combined location -56+ .18 54+ .18 72+ .18
81
The percent of living trees emerging from the grass stage ranged
from 27 to 93 percent. The Hardin County, Texas, test showed the bene-
fit of intensive competition control with both very high survival and a
high percentage of living trees initiating height growth. There were
differences between families within provenances for grass stage emer-
gence in four of the seven locations. Family heritabilities ranged
from 0.22 to 0.40.
Percent of trees infected with brown-spot needle blight varied
from only 5 percent at the Hardin County, Texas, test to 96 percent at
the Tyler County, Texas, test. In this planting, the average tree had
almost 50 percent of its foliage infected. Brown-spot severity score
was used for analysis because it was more heritable than the percent of
trees infected. The average planting severity code ranged from a low
of 0.09 to a high of 2.88. There were differences between families
within provenances in five of the seven tests. Family heritabilities
varied from 0.23 to 0.70. The low heritability estimate for the Hardin
County test was primarily caused by the lack of infection. Brown-spot
resistance was the most heritable of the three traits scored.
Combined Locations
When the 64 families common to all plantings were analyzed, there
were differences between families within provenances for all three
traits. There were also differences attributable to provenances for
survival and grass stage emergence. The North Louisiana and Southeast
Texas sources had slightly higher survival and grass stage emergence
(Table 2). There was no provenance by planting location interaction
for either trait. While there was no provenance effect for brown-spot
infection, there was evidence of a provenance by planting location
interaction (Figure 2). This interaction was statistically significant
but not operationally meaningful. Although there were some changes in
ranks at the Louisiana plantings, where the differences among
provenances were small, the interaction resulted primarily from changes
in magnitude. North Louisiana sources tended to be more susceptible
while South Mississippi sources were affected least.
Table 2. Provenance means for survival and grass stage emergence
for two-year-old longleaf pine planted in the Western
Gulf region.
Survival Growth Initiation
Provenance (Percent) (Percent)
North Louisiana 70.8 62.0
Southeast Texas 70.4 2 5 7
South Louisiana 68.1 59 .6
South Mississippi 65.0 56
82
MOCO <HA—TM<mMH AOCVMZEOIDHV
HARDIN BIENVILLE LASALLE ; VERNON PEARL RIVER STONE TYLER
PLANTING
LEGEND: ZONE eee N. LA ~--@--@ S. LA w-t--+ S. MS +—+-+ S.E. TX
Figure 2. Provenance (zone) by planting averages for brown-spot
severity in two-year-old longleaf pine.
There was a family within provenance by planting location interact-
ion for all three traits. This interaction accounted for approximately
10 percent of the total phenotypic variation and was considered unimpor-
tant because of the large amount of family variation. Family heritabil-
ity across all locations was 0.56 for survival, 0.54 for grass stage
emergence and 0.72 for brown-spot resistance (Table 3). These moderate
to strong heritabilities and the distribution of variation between the
provenance effect and family within provenance effect (Table 4) indi-
cate that emphasis should be placed on selecting the best individuals
regardless of seed collection zone.
Table 3. Heritabilities and coefficients of genetic predition from the
combined locations analysis for seven plantings of two-year-
old longleaf pine in the Western Gulf region.
Grass Stage Brown-Spot
Trait Survival Emergence Severity Score
Survival 0.56 0.18 0.07
Grass Stage
Emergence 0.54 0.05
Brown-Spot
Severity Score 0372
83
Table 4. Distribution of variation in percent between provenances and
families within provenances for two-year-old longleaf
pine planted in the Western Gulf region.
Grass Stage Brown-spot
Type of Variation Survival Emergence Severity Code
Provenance 28 35 10
Family (Provenance) 72 65 90
The coefficient of genetic prediction indicates that survival and
the percent of trees initiating height growth are positively related
(Table 3). Selection resulting in a one standard deviation increase in
the phenotypic value for percent survival would be accompanied by an
0.18 standard deviation gain in breeding value for grass stage emer-
gence. Brown-spot resistance at this age does not appear to be related
to either survival or percent of trees initiating growth. This patho-
gen causes fatality by repeated defoliation and at least one of the
mechanisms for escaping brown-spot needle blight is early height initia-
tion. It may be that the relationship between these traits has not had
sufficient time to develop in this study.
By considering the same variable across locations as different
traits the coefficients of genetic prediction can be calculated across
environments. This is a good device for examining genotype by environ-
ment interactions as suggested by Burdon (1977) for genetic correla-
tions and demonstrated by Yeiser and others (1981). It is also useful
in formulating seed movement recommendations and delineating breeding
zones. The danger, when examining longleaf pine in this manner, is
that most of the juvenile traits are strongly affected by nursery treat-
ment. In this study, plantings in each zone were grown at the same
nursery and can be expected to have similarities related to common nurs-
ery culture. Planting locations for which family variation was statis-
tically insignificant, indicating no detectable additive genetic vari-
ance, were dropped from the CGP matrix.
CGP's for survival across locations are shown in Table 5. Because
of the high overall survival at the Hardin and Tyler County, Texas,
plantings, CGP's with these tests are very low. If these tests are
ignored, it becomes apparent that survival at all of the other loca-
tions is positively correlated. For example, families selected for a
one phenotypic standard deviation increase in survival at Stone Coun-
ty would have an increased breeding value of 0.46 standard deviations
if planted at Pearl River County. In this example, it is impossible to
separate the planting location effects from those contributed by a com-
mon nursery. Comparisons to the heritabilities along the diagonal indi-
cate the relative efficiency of indirect selection.
84
Table 5. Coefficients of genetic prediction for survival in
two-year-old longleaf pine across different test locations
in the Western Gulf area.
Test Stone Pearl River Tyler Hardin Bienville LaSalle
MS MS TX TX LA LA
Stone, MS 0.47 0.46 0.01 0.22 0.24 0.26
Pearl River, MS 0.55 =0.12 0.16 0.21 0.34
Tyler, TX 0.21 0.00 0.00 -0.05
Hardin, TX 0.29 -0.03 0.11
Bienville, LA 0.34 0.25
LaSalle, LA 0.28
Feeincactinss SsnSnnnnennn
Table 6 contains the CGP's for grass stage emergence. There ap-
pears to be a positive relationship between all test locations. This
implies no special breeding zones are needed for east-west seed move-
ment in the Western Gulf region. Seed movement recommendations for
North Louisiana could not be made because the CGP's for these tests
could not be calculated.
Table 6. Coefficients of genetic prediction for grass stage emergence
in two-year-old longleaf pine across different test locations
in the Western Gulf region.
Test Stone Pearl River Tyler Vernon
MS MS TX LA
Stone, MS 0.37 0.51 0.48 0.34
Pearl River, MS 0.23 0.25 0.14
Tyler, TX 0.40 0.37
Vernon, LA 0.22
Nae ee ere
The coefficients for genetic prediction for brown-spot severity
are listed in Table 7. CGP's for brown-spot are very similar to those
for percent height initiation. There is a strong positive relationship
between all test locations with possible exceptions of the North Louisi-
ana tests for which CGP's could not be calculated. Again, positive
gains in brown-spot resistance at any of the five southern tests would
result in positive gains at any of the other southern tests.
Table 7. Coefficients of genetic prediction for brown-spot severity
in two-year-old longelaf pine across different test
locations in the Western Gulf region.
Test Stone Pearl River Tyler Hardin Vernon
MS MS TX TX LA
Stone, MS 0.70 0.41 0.68 0.35 0.45
Pearl River, MS 0.49 0.36 0.20 0.26
Tyler, TX 0.69 0.23 0.52
Hardin, TX 0.23 0.59
Vernon, LA 0.32
85
CONCLUSIONS
1. Good gains can be made in longleaf pine through selection for
survival, grass stage emergence, and brown-spot resistance.
2. Family selection will result in twice as much gain as prove-
nance selection.
3. Survival and grass stage emergence are positively related
while brown-spot resistance is independent of either trait at this
early age.
4. The same selection criteria will improve survival for all
areas within the Western Gulf region.
5. Southeast Texas, South Louisiana and South Mississippi can
also be considered one zone when selecting families for grass stage
emergence and brown-spot resistance.
LITERATURE CITED
Baradat, P. 1976. Use of juvenile-mature relationships and infor-
mation from relatives in combined multitrait selection. IUFRO,
Joint Meeting on Advanced Generation Breeding, Bordeaux. p. 121-138.
Burdon, R. D. 1977. Genetic correlation as a concept for studying
genotype-environment interaction in forest tree breeding. Silvae
Genetica 26(5-6):168-175.
Goddard, R. E. and Bryant, R. 1981. Genetic variation in survival of
of longleaf pine. Proc. 16th South. For. Tree Imp. Conf. p. 136-142.
Goddard, R. E. and Rockwood, D. L. 1978. Cooperative forest genetics
research program progress report 20. Univ. of Fla., School of For.
Res. and Conserv., Rept. No. 28. 20 p.
Goddard, R. E., Hollis, C., Kok, H. R., Rockwood, D. L., and
Strickland, R. K. 1973. Cooperative forest genetics research
program progress report 15. Univ. of Fla., School of For. Res. and
Conserv. Rept. No. 21. 19 p.
Layton, P. A. and Goddard, R. E. 1982. Environmental and genetic
effects on duration of the grass stage of longleaf pine. Proc.
Seventh North Am. For. Bio. Workshop. p. 131-136.
Rockwood, D. L. and Kok, H. R. 1977. Development and potential of a
longleaf pine seedling seed orchard. Proc. 14th South. For. Tree
Imp. Conf. p. 78-86.
Shipman, R. D. 1960. Survival and growth of graded longleaf pine
nursery stock. J. Forest 58:38-39. 42.
86
Smith, L. F. and Schmidtling, R. C. 1970. Cultivation and fer-
tilization speed early growth of planted southern pines. Tree
Planters’ Notes 21(1):1-3.
Snyder, E. B. 1969. Parental selection versus half-sib family selec-
tion of longleaf pine. Proc. 10th South. For. Tree Imp. Conf. p. 84-
88.
Snyder’, E. B., Dinus, RR.” J., and Derr, H. J. 1977. ° Genetics of long-
leaf pine. USDA For. Ser. Res. Pap. WO-33. 24 p.
Yeiser, J. L., van Buijtenen, J. P., and Lowe, W. J. 1981. Genotype x
environment interactions and seed movements for loblolly pine in the
Western Gulf Region. Silvae Genetica 30(6):196-200.
87
POLYMORPHIC ISOENZYMES FROM MEGAGAMETHOPHYTES AND POLLEN OF
LONGLEAF PINE: CHARACTERIZATION, INHERITANCE, AND USE IN
ANALYSES OF GENETIC VARIATION AND GENOTYPE VERIFICATION
Stuart E. Dubal/
Abstract.--Segregation of isoenzyme variants of 13 enzyme
systems, assayed in longleaf pine (Pinus palustris Mill.) indicated
control by 19 separate loci in megagametophytes and embryos of
control-crossed and open-pollinated seeds. Along with megaga-
metophyte evaluations, pollen contributions to embryos proved to be
suitable for genetic evaluation and determination of genotypes. A
unique multi-locus genotype was determined for each parent involved
in control-crosses. Exact parentage of hybrids was determined from
female and male contributions to hybrid multi-locus genotypes.
Unique genotypes were determined for 62 of 68 parents evaluated.
Considerable genetic variation in isoenzymal characteristics was
found among 24 natural populations from the Central Gulf Coast.
Allele frequencies per locus differed significantly among popu-
lations although in most cases one allele was more frequent in all
populations. Numbers of polymorphic loci per population ranged
from 31.6 to 57.9 percent and were correlated with latitudes at
which the populations occurred (r= 0.63, P<0.002).
Additional keywords: Pinus palustris Mill., electrophoresis,
number of alleles, polymorphic loci, genetic distance.
Isoenzyme analyses, by means of electrophoretic separation, are no longer
a novelty in forest genetics investigations, yet the number of species and
populations that have been studied is limited. Loblolly (Pinus taeda L.) is
the only southern pine that has been dealt with in much detail (Adams and
Jolly 1980, Conkle and Adams 1977, and Florence and Rink 1979). Conkle and
Adams (1977) included some longleaf pine (Pinus palustris Mill.) seeds in a
survey of southern pine banding patterns and concluded that similar genes were
probably present in all of them. However, inheritance of isoenzymes must be
established before they can be used in genetic analyses. A very limited
amount of isoenzymal inheritance data has been gathered for longleaf pine
(Snyder and Hamaker 1978). To be of much utility in arriving at unique
genotypes for a large number of parent trees, a large number of polymorphic
loci must be available and their inheritance must be understood. Part I of
this study was designed to elucidate the inheritance of 19 loci representing
13 enzyme systems in longleaf pine when evaluations were made of both the
female gametophyte and the contribution due to pollen. Although isoenzymes
may be expressed in both embryo and megagametophyte tissues, the pollen
1/Research Associate, School of Forestry, Alabama Agricultural Experiment
Station, Auburn University, AL 36849. Support of this research by the
Georgia Forestry Commission is sincerely appreciated.
88
contribution to the embryo genotype is not always apparent because of
confounding of the banding patterns. The reliability of determining genotypes
from pollen contributions of control-crosses was determined. Both embryo and
megagametophyte tissue were used to establish unique genotypes for parents and
verify the parentage of crosses.
Genetic variation has been demonstrated among seed sources of longleaf
pine for metric traits such as height growth (Wells and Wakeley 1970) and
disease resistance (Synder and Derr 1972). An appreciable amount of genetic
variation has been found for height growth which has allowed for the
differentiation of geographic zones based on growth potential. The question
arises whether similar variation is present for isoenzymal characteristics and
if so, what is its extent and distribution. Also, if it is present can it be
used to differentiate populations. Isoenzymal analyses are currently used to
estimate heterozygosities, genic diversity, and the extent of population
differentiation.
Isoenzymal variation has been demonstrated for single trees of loblolly
pine (Adams and Jolly 1980) and for longleaf pine (Duba, 1983). Variation has
also been demonstrated for particular isoenzyme loci of loblolly pine
(Florence and Rink 1979), pitch pine (Pinus rigida Mill.) (Guries and Ledig
1982), Norway spruce (Picea abies K.) (Lundkvist 1979), and Douglas-fir
(Pseudotsuga menziesii (Mirb.) Franco) (Yeh and O'Malley 1980) populations,
although when averaged over several loci the variation has not always shown
extensive population differentiation. Some clinical trends have been
indicated (Guries and Ledig 1982, Yeh and O'Malley 1980), but more isoenzymal
genetic variation has been found to exist within populations than between
them. Native conifers differ in the kind and amount of isoenzymal genetic
variation they contain (Conkle 1980) and, in general, possess high levels as
would be expected when considering their life history characteristics (Hamrick
et al. 1979).
Longleaf pine is a long-lived, wind pollinated species with a large
natural range that does not span dramatic climatic differences. Still,
patterns of variation in growth characteristics have been associated with
climatic factors, particularly temperature and rainfall. In connection with a
study of geographic variation in growth potential of longleaf pine, various
sources were utilized in Part II of this study to evaluate variation in
isoenzymal characteristics. The level of genetic diversity within the species
was determined and patterns of variation among populations were evaluated by
analyzing protein polymorphisms that were revealed by electrophoretic
separations.
MATERIALS AND METHODS
Part I
Control-cross seeds were utilized for the majority of these analyses, but
observations from open-pollinated seeds were also included. The control-cross
seeds represented 10 crossing combinations from 5 parents that were included
in the U.S. Forest Service's longleaf breeding program at Gulfport,
Mississippi. Open-pollinated seeds represented 63 longleaf parents from
sources located in Alabama and adjoining states.
89
For a detailed description of the electrophoretic run conditions, consult
Duba (1983). Mobility of electrophoretic bands was used to identify differing
zones of activity. Segregation of band patterns in each zone was evaluated in
conjunction with mobility of the zones of activity to identify separate loci
and alleles (variants) at each locus. Inheritance of the enzymes was
postulated based on segregations observed in band patterns of both megaga-—
metophyte and embryo tissues. Whenever a locus was determined to be
heterozygous, chi-square values were calculated to evaluate the goodness-
of-fit to the expected 1:1 ratio of segregation. The segregation in pollen
gametophytes was compared to that from megagametophytes to evaluate the
suitability of using pollen contributions to detemine genotypes.
From analyses of megagametophytes, the multi-locus genotype of the 5
parents involved in the 10 crosses and of the 63 source parents was
determined. Multi-locus genotypes and segregation ratios of progeny these
crosses were evaluated. Verifications of the parents responsible for
particular crosses were made.
Part II
The primary center of sampling was the central gulf coast region of the
natural range of longleaf pine. The entire sample consisted of 22 populations
distributed through Alabama, southeast Mississippi, southwest Georgia, and the
panhandle of Florida, plus 2 distant sources, 1 in central Florida and 1 in
North Carolina (Figure 1). Each source was evaluated as a separate population
to determine the extent and distribution of isoenzymal genetic variation.
Figure 1.-- Species range of longleaf pine and relative locations of sampling
points.
90
Direct count allele frequencies at each locus were obtained for each of
the 24 populations. Within each population, genetic variation was quantified
by determining the average number of alleles per locus, the proportion of
polymorphic loci, and the Hardy-Weinberg expected proportion of heterozygous
loci per individual. Heterogeneity chi-square values were calculated for
allele frequencies among all populations to determine if frequencies were
different from one population to another. Linear correlations were also
computed between certain isoenzymal characteristics and latitude and longitude
as well as growth potential.
RESULTS AND DISCUSSION
Part I
From analyses of electrophoretic banding patterns, 19 consistently
staining zones of activity were observed in 13 enzyme systems. Evidence
collected from segregation analyses of megagametophytes and pollen
contributions to embryos of control-cross seeds (Table 1) demonstrated
directly that 8 zones (ALAP, LAP-1, PGI-2, GOT-1, GOT-3, SKDH, MDH-2, AND
PGD-1) were each controlled by a single locus. Evaluations from megaga-
metophytes of open-pollinated seeds indicated control of 8 more zones (ADH,
FLEST, LAP-2, PGI-1, PGM-1, PGM-2, GPD-3, and PGD-2) by a single locus each.
Indirect evidence from embryos of open-pollinated seeds indicated control of
the final 3 zones (GDH, GPD-1, and IDH) also by a single locus each.
Analysis of embryo bands in control-cross and open-pollinated seeds gave
evidence that pollen contributions can be reliably ascertained (Table 1),
although caution was required in evaluating one allelic combination at the
MDH-2 locus. Pollen contributions can be utilized to verify male parents in
hybrids and evaluate allele frequencies in population studies where the
contributions to embryos can be consistently scored.
Table 1.-- zyp i i n n duced
Allelic Observed
Enzyme Estimated combination nupber Deviation
locus from® x Y x Total (1) P
ALAP G 1 2 18 39 0.23. >.50
P 1 2 12 24 0.00 >.90
LAP-1 G 1 2 17 40 0.90 >.25
P 1 2 1] 24 0.17. »>.50
PGI-2 G 1 2 20 44 0.36 >.50
P 1 2 15 24 1.50 >.10
G 1 3 56 15 0.08 »>.75
P (I 3 42 84 0.00 >.90
GOT-1 G 1 3 18 40 0.40 >.50
P 1 3 14 24 0.67 >.25
GOT-3 G 1 2 22 40 0.40 >.50
P i 2 15 24 1.50 >.10
SKDH G 1 Z3 51 89 1.90 >.10
? i 2 20 42 0.09 = >.75
MDH-2 fe} 1 2 17 41 1.39 8=6>.25
P i 2 7 18 0.89 »>.25
G 1 5 38 89 1.90 >.10
P 1 5 15 31 0.03 »>.75
G 2 5 14 40 3.60 >.05
P 2 5 43 $5 6.78 <.01
PGD-1 G 1 4 16 41 1.97 >.10
P 1 4 il 21 0.05 oes
G 1 5 80 120 13.33 <.01
P 1 5 40 62 Sia2e S002
|
|
|
® Estimated from female gametophyte (GC) or pollen (PF) contributions.
b Allele 1 is the most frequent allele s: each locus.
91
Segregation distortion was demonstarted in two combinations of PGD-l.
Both combinations showed a deficiency for the same allele. Normal segregation
was indicated for the other alleles at this locus. Therefore, caution must be
exercised when using this locus for genetic evaluations. Part of this problem
may have been due to band resolution between two alleles. The allele that was
deficient migrated to a location between the two alleles that had excesses
when in heterozygous combination. Although bands for this locus were very
clear, the closeness in position of these bands may have been partly
responsible for the excess numbers observed.
Every locus evaluated in longleaf pine had at least two electrophoretic
variants (Figure 2) although the second variant (allele) at several loci was
very rare. The reason for this observation may be that samples from a large
portion of the longleaf pine natural range in Alabama, Florida, Georgia, and
Mississippi were included in the open-pollinated seed collection, thus
constituting a very broad sample.
ati + Pave
60 E 50
° c}
© 50 S 40
x 40 ° k
= E 30 -
30 re >
eS. 20
20 5
o
10 2 10
Origin :
Peper esprprespeprsprepepesepepapese] “MSs
Allelic Designation Allelic Designation
Figure 2.-- Band patterns and their allelic designations for 19 loci in long-
leaf pine.
An important utility of isoenzymal analyses to applied tree breeding has
been the identification of parents (Adams and Jolly 1980). Offspring
genotypes and frequencies have been predicted from genotypes of the parents
involved in control-crosses. If each parent involved had a unique genotype,
then the parents of any particular cross should be verifiable. The number of
loci evaluated in this study was sufficient to allow for the determination of
unique genotypes for each parent involved in control-crosses (Table 2). These
multi-locus genotypes were suitable for evaluations of male and female
contributions to hybrid embryos and the exact establishment of parentage
(Table 3). The discovery that 62 of a total of 68 parents had unique
genotypes was reassuring. If a large enough number of loci were included in
evaluations, determinations of unique genotypes for applied breeding
utilization is feasible.
92
Table 2.-- Unique genotypes of the five parents used in control crosses
see coes SS SSeS
Genotype Mumber of alleles by which
(by locus) two parants oi‘ter
Parent ADH ALAP FLEST LAP] LAP2 PGI] PGI2 PGMl PGM2 GH GOT] GOT3 GPD] GPD3 SKDH IDH HOH2 POD) PCD?
1 D2) V5, 81s5. 1 4
}e
1
2 1 1 1 1 1 1 1,2 1 1 i 1 1 1 1 1 1 1,2 1,4 1 !
; is
6
Tel 2 owl e255), kl) ee) 7
7 3
et Tien 4
=
~
~
N
~
~
i)
~
~
-
~
-~ —
—
~
.
w
~
.
wn
~
-
~
~
er ut at sa. Bo ae te OS SF
Soe e sete fe Ocean Meets hte kt de kk
epee ona tO ee tO Oe
ce ee crea Pi ot Pee a
peel ra Ger Oa ierte ae ok che koh te eg
Spee ser tae te tt tt te eS
ices arg te ee a a Ek ee
Peony ot etiam th CE PRPS es a) ae ee a a
MMe etal vtewite te te atte Bt tt eS
LG Be SLRE PLT esr tas Win Cy Sis Wes a Pay cS Ys es Kana Was US Cee Pee
22 ee mim i Ds PRIS syne NTT op SNR Ae ye sm ae ie yeoman TErgl ny
The utility of isoenzymes in genetic studies has been described
previously (Adams 1979, Allard et al. 1975). Results on the inheritance of
isoenzymes in longleaf pine allow for its addition to the list of species for
which isoenzymal evaluations can enhance genetic analyses.
Part II
Allele frequencies were determined for each population as a first measure
of genetic variation (Table 4). At least two loci were observed in each
population sample, although several loci were essentially monomorphic. A
single allele was more frequent in all populations for 16 of the 19 loci.
Contingency chi-square analysis indicated significant differences for all loci
except GPD-3 and PGI-1 (Table 5). Thus, in the allele frequency data, an
appreciable amount of genetic variation was reflected among populations, but
the distribution was not readily apparent.
93
Table 4.-- Allele frequencies at 19 loci in natural Populations of longleaf pine
Locus Allele 3 5 7 9
ADH 1 1.00 1.00 1.00 0.97
2 0.03
ALAP 1 0.80 0.95 0.94 1.0
2 0.20 0.05 0.06
FLEST 1 0.97 0.93 0.97 0.87
2 0.03 0.07 0.03 0.13
LAP-1 1 0,91 0.71 0.92 0.86
2 0.09 0.29 0.08 0.14
LAP-2 1 1.00 1.00 1.00 1.00
2
3
PGI-1 1 1.00 1.00 1.00 1.00
2
PGI-2 1 0.50 0.34 0.42 0.33
2 0.42
3 0.03 0.53 0.48 0.63
4 0.05
S 0.12 0.10 0.04
Pa-1 1 1.00 1,00 1.00 0.95
2 0.05
PGM-2 1 1.00 1.00 1.00 0.88
2 0.12
3
GDH 1 1.00 1.00 1.00 1.00
2
cor) 1 0.97 0.94 0.91 0.85
3. 0.03 0.06 0.09 0.15
cor-3 1 0.91 0.97 0.95 0.91
Zu 0/.09 0.03 0.06
3 0.03 0.02 0.03
4
GPD-1 1 1.00 1.00 1.00 1.00
2
GPD-3 1 1.00 0.98 1.00 1.00
2 0.02
SKDH 1 0.69 0.46 0.71 0.58
2 0.31 0.54 0.26 0.42
3 0.03
4
IDH 1 1.00 0.94 1.00 1.00
2 0.03
3 0.03
MDH 1 0.48 0.35 0.30 0.52
2 0.32 0.08 0.34 0.33
5 0.20 0.58 0.36 0.15
PGD-1 1 0.89 0.82 0.77 0.89
2 0.04
3. 0.06 0.15 0.08
4 0.02 0.03
5 0.03 0.18 0.03
PGD-2 1 1.00 0.97 0.98 1.00
2
3 0.03 0.02
ll
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00 1.00 1.00 1.00 0.98 1.00 1.00 1.00 1.00 1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.30
0.65
0.05
1.00
1.00
1.00
1.00
1.00
0.02
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.75
0.08
0.17
1.00
1.06
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.42
0.32
0.03
0.23
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.46
0.18
0.36
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.60
0.09
0.09
0.22
1.00
1.00
1.00
1.00
94
1.00
1.00
1.00
1.00
0.98
0.02
1.00
1.00
1.00
0.79
0.08
0.13
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
0.86
0.05
0.09
1.00
1.00
0.02
1.00
1.00
Table 5.--Contingen hi-squ analysis 19 loci pooled over
populations
No. of
Locus alleles Chi-square D.F. P
ADH 2 69.95 23 0.000
ALAP 2 161.89 23 0.000
FLEST 2 58.71 23 0.000
LAP-1 2 175.73 23 0.000
LAP-2 3 195.06 46 0.000
PGI-1 2 23.22 23 0.448
PGI-2 5 725 .86 92 0.000
PGM-1 2 91.03 23 0.000
PGM-2 3 178.91 46 0.000
GDH 2 98.96 23 0.000
GOT-1 2 69.98 23 0.000
GOT-3 4 166.73 69 0.000
GPD-1 2 48 .60 23 0.001
GPD-3 2 27 .66 23 0.229
SKDH 4 320.74 69 0.000
IDH 3 136.56 46 0.000
MDH 3 272.28 46 0.000
PGD-1 5 517.87 92 0.000
PGD-2 3 79.20 46 0.002
As a second measure of isoenzymal genetic variation, the average number
of alleles per locus, the percentage of polymorphic loci, and the average
heterozygosity per individual were compared among populations (Table 6).
Although 2 or more alleles were found for every locus over all populations,
the range in number of alleles per locus for individual populations was only
1.5 to 1.9. There were no clear-cut differences between any two populations
in number of alleles, but it was evident from the data that populations
differed in the presence of specific alleles and in the frequencies of these
alleles. The percentage of polymorphic loci per population ranged from 31.6
to 57.9 percent and was correlated with latitude of the populations (r =
-0.63, P < 0.002). The more southern populations had a higher percentage of
polymorphic loci.
95
Table 6.--Variation of isoenzyme characteristics in 24 natura! populations of longleaf pine
Percent heterozygosiry
per individuai
Average number of
alleles per locus
Percentage of
Population loci polymorphic)
3 Washington, FL 1.7 + 0.2 36.8 13.8 + 4.7
5 Okaloosa, FL 1.8 + 0.2 47.4 15.8 + 4.3
7 Santa Rosa, FL 1.8 + 0.2 36.8 14.1 + 4.9
9 Escambia, FL 1.8 + 0.2 47.4 16.2 + 4.5
11 Baldwin, AL 1.9 + 0.2 57.9 16.2 + 4.5
13 Mobile, AL 1.8 + 0.2 47.4 17.9 + 4.8
16 Stone, MS 1.8 + 0.2 42.1 16.0 + 5.1
19 Worth, GA 1.9 + 0.2 42.1 17.6 + 5.1
23 Geneva, AL 1.9 + 0.2 47.4 18.6 + 5.6
25 Pike, AL 1.8 + 0.2 36.8 14.3 + 5.0
29 Conecuh, AL 1.9 + 0.2 42.1 14.8 + 4.7
30 Washington, AL 1.8 + 0.2 42.1 15.6 + 4.6
31 Wayne, MS 1.7 + 0.2 42.1 17.2 + 4.9
32 Perry, MS LO + O)o2: 31.6 13.5 + 4.6
36 Lawrence, MS 1.5 + 0.2 31.6 8.8 + 3.8
37 Taylor, GA 1.6 + 0.2 31.6 14.0 + 5.2
38 Meriwether, GA 1.6 + 0.2 31.6 12.6 + 4.8
41 Autauga, AL 1.8 + 0.3 42.1 14.0 + 4.9
42 Bibb, AL 1.7 + 0.2 31.6 15.2 + 5.3
43 Hale, AL 19+ O02 31.6 14.1 + 5.0
45 Scott, MS 1.9 + 0.2 42.1 18.0 + 5.2
46 Tallapoosa, AL Lo7 te Ol2 36.8 13.6 + 4.2
50 Marion, FL 1.8 + 0.2 36.8 13.9 + 4.4
51 Richmond, NC 1.8 + 0.3 42.1 15.0 + 4.5
a) Considered polymorphic if the frequency of the most common allele does not exceed 0.95.
As a third measure of genetic variation, the genetic distance between
pairs of populations was determined (Table 7). Genetic distance coefficients
combined over all loci ranged from 0.0 for 2 populations (populations 3/7 and
43) from the same latitude but separated in longitude, to a high of 0.048
between the North Carolina population (population 51) and a Mississippi
population (population 31). In general the genetic distance coefficients were
small and of the same order as those for pitch pine (Guries and Ledig 1982),
but were much larger than those for Douglas-fir (Yeh and O'Malley 1980).
Distribution of the largest coefficients was essentially random although for
some pairs, such as 31 and 51, they also were separated by a large geographic
distance.
Genetic distance coefficients also were calculated for each locus
separately to evaluate specific locus contributions to overall coefficients.
There were six loci (ALAP, LAP-1, MDH, PGD-1, PGI-2, and SKDH) that seemed to
contribute the most to the overall coefficients. Of these six, PDG-1 and
PGI-2 both had coefficients ranging from 0.0 to approximately 0.98, and were
the largest contributors to overall distances. The wide variation
attributable to separate loci indicated the necessity for evaluating large
numbers of loci in order to correctly evaluate population differentiation.
Although these analyses identify substantial variation, the distribution among
populations suggested a generally random distribution with only slight
population differentiation.
96
Table 7--- Genetic distance!) coefficients between longleaf pine populations
3 ences
5 0.027 *****
7 0.015 0.011 *****
9 0.022 0.011 0.007 *****
ll 0.011 0.015 0.004 0.003 *****
13 0.020 0.019 0.012 0.008 0.007 *****
16 0.014 0.023 0.015 0.025 0.012 0.016 *****
19 0.028 0.001 0.014 0.010 0.013 0.013 0.022 *##*¢
23 0.024 0.015 0.009 0.009 0.010 0.015 0.029 0.015 *****
25 0.012 0.020 0.003 0.005 0.002 0.011 0.018 0.021 0.006 *****
29 0.025 0.004 0.006 0.005 0.009 0.014 0.029 0.005 0.007 0.010 ***e#
30 0.013 0.024 0.009 0.009 0.004 0.012 0.009 0.022 0.018 0.004 0.019 ****"
31 0.047 0.043 0.020 0.029 0.030 0.019 0.047 0.038 0.024 0.023 0.031 0.033 **#e
32 0.009 0.020 0.007 0.012 0.004 0.011 0.013 0.024 0.013 0.003 0.018 0.007 0.031 **s*e
36 0.011 0.036 0.013 0.024 0.010 0.021 0.012 0.042 0.027 0.008 0.032 0.008 0.043 0.002 eee
377 0.012 0.010 0.002 0.009 0.004 0.011 0.009 0.011 0.012 0.004 0.009 0.006 0.030 0.004 0.009 **s**
38 0.011 0.025 0.010 0.014 0.004 0.013 0.012 0.023 0.013 0.004 0.019 0.006 0.037 0.003 0.006 0.005 esse
41 0.011 0.016 0.008 0.021 0.012 0.028 0.016 0.024 0.017 0.012 0.017 0.020 0.048 0.011 0.015 0.007 0.012 *****
42 0.021 0.025 0.005 0.016 0.009 0.016 0.018 0.024 0.011 0.004 0.019 0.010 0.013 0.009 0.015 0.007 0.010 0.018 se#+*
43 0.009 0.015 0.002 0.011 0.004 0.014 0.009 0.017 0.012 0.002 0.010 0.006 0.033 0.004 0.006 0.0 0.004 0.005 0.008 eeeee
45 0.015 0.009 0.004 0.011 0.006 0.015 0.008 0.011: 0.010 0.007 0.008 0.010 6.035 0.010 0.016 0.003 0.008 0.005 0.012 0.002 see
4 0.011 0.033 0.016 0.022 0.009 0.020 0.015 0.035 0.020 0.011 0.032 0.015 0.042 0.005 0.008 0.014 0.006 0.011 0.016 0.013 0.017 e#ee2
so 0.021 0.032 0.023 0.034 0.023 0.021 0.009 0.035 0.039 0.024 0.043 0.016 0.040 0.012 0.013 0.015 0.018 0.026 0.021 0.019 0.021 0.017 *##e
51 0.018 0.025 0.020 0.032 0.022 0.022 0.003 0.029 0.037 0.027 0.035 0.017 0.049 9.018 0.017 0.016 0.024 0.022 0.028 0.017 0.0186 0.022 0.008
1) unbiased genetic distance (D) according to Nei, 1978.
LITERATURE CITED
Adams, W.T. 1979. Applying isozyme analyses in tree breeding programs.
Pages 60-64 in Proc. Symp. Isozymes N. Amer. For. Trees For. Insects.
USDA For. Serv. Gen. Tech. Rep. PSW-48. 64pp.
Adams, W.T., and R. J. Jolly. 1980. Alloyme studies in loblolly pine
seed orchards: clonal variation and frequency of progeny due to
self-fertilization. Silvae Genet. 29:1-4.
Allard, R.W., A.L. Kahler, and M.T. Clegg. 1975. Isozymes in plant
population genetics. In Isozymes IV Genetics and Evoluation. C.
L. Markert, Ed. Academic Press, New York. p. 261-272.
Conkle, M.T. 1980. Amount and distribution of isozyme variation in
various conifer species. Proc. 17th Meet. Canadian Tree Improv.
97
Conkle, M.T., and W.T. Adams. 1979. Use of isoenzyme techniques in
forest genetics research. Proc. 14th Southern Forest Tree Improv.
Conf. Gainesville, Fla. p. 219-226.
Duba, S.E. 1983. Genetic variation in growth and isoenzymal charac-
teristics of longleaf pine from various sources. Ph.D. Thesis.
Auburn University. 123pp.
Florence, L.Z., and G. Rink. 1979. Geographic patterns of allozymic
variation in loblolly pine. Proc. 15th Southern Forest Tree Improv.
Conf. Starkville, Miss. p. 33-41.
Guries, R.P., and F.T. Ledig. 1982. Genetic diversity and population
structure in pitch pine (Pinus rigida Mill.). Evolution
36 :387-402.
Hamrick, J.L., J.B. Mitton, and Y.B. Linhart. 1979. Levels of genetic
variation in trees: Influence of life history characteristics.
Pages 35-41 in Proc. Symp. Isozymes North Amer. For. Trees and For.
Insects. USDA For. Serv. Gen. Tech. Rep. PSW-48. 64pp.
Lundkvist, K. 1979. Allozyme frequency distributions in four Swedish
populations of Norway spruce (Picea abies K.) I. Estimations of
genetic variation within and among populations, genetic linkage, and
a mating system parameter. Hereditas 90:127-143.
Snyder, E.B., and H.J. Derr. 1972. Breeding longleaf pines for resistance
to brown-spot needle blight. Phytopath. 62:325-329.
Snyder, E.B., and J.M. Hamaker. 1978. Inheritance of peroxidase
isozymes in needles of loblolly and longleaf pines. Silvae
Genet. 17:125-129.
Wells, 0.0., and P.C. Wakeley. 1970. Variation in longleaf pine from
several geographic sources. For. Sci. 16:28-42.
Yeh, F., and D. O'Malley. 1980. Enzyme variations in natural populations
of Douglas-fir, Pseudotsuga menziesii (Mirb.) Franco, from British
Columbia 1. Genetic variation patterns in coastal populations.
Silvae Genet. 29:83-92.
98
CONIFER GENETICS Il
MODERATED BY DR. TONY SQUILLACE
Universtiy of Florida
99
Genetic and Cultural Factors Affecting Growth Performance
of Slash Pine
G. L. Reighard, D. L. Rockwood, and C. W. Comer
Abstract.--Five- to seven-year growth performances of genetic-
ally select slash pine progenies planted at five northern
Florida sites were evaluated for differences due to family,
provenance, plantation site, competition (pure and maximum),
plot design (block and Nelder), spacing (472-43,100 trees/ha),
and age. Significant family differences were found for
growth. Family x site interactions were important on poor
sites. Intergenotypic competition and plot design did not
affect family performance. Spacing influenced diameter and
volume, but a family x spacing interaction was not apparent.
Growth trends detected at age five continued at age seven, but
variation among families decreased.
Additional Keywords: Pinus elliottii var. elliottii, genetic
tests, genotype x environment interaction, spacing.
INTRODUCTION
Numerous genetic tests of slash pine (Pinus elliottii var.
elliottii Engelm.) have been established in the southeastern United
States during the past 25 years. Data from these tests have been used
to select fast-growing, rust-resistant genotypes for clonal seed or-
chards. Recently, Franklin (1979, 1983) and Stonecypher and McCullough
(1981) have advocated shortening the evaluation period in progeny tests
by planting at high, non-conventional densities to create competition at
earlier ages. Furthermore, factors such as plot design, age of measure-
ment, and environment interactions require additional study to determine
if present genetic evaluations are appropriate for wide geographic
plantings of selected slash pine families. The objectives of this paper
were to evaluate the effects of family, provenance, site, intergenotypic
competition, plot design, spacing, and age on growth performance of
slash pine progenies at five northern Florida sites.
1/
— Assistant in Forest Biomass, Associate Professor, and Assistant
Research Scientist, respectively, Department of Forestry, University of
Florida, Gainesville, FL. Research reported here was supported by the
Cooperative Forest Genetics Research Program, Oak Ridge National Labora-
tory under subcontract No. 19X-09050C, and a cooperative program between
the Institute of Food and Agricultural Sciences of the University of
Florida and the Gas Research Institute entitled "Methane from Biomass
and Waste."
100
Bit hee ka eg ;
Dia Mga, AC CLE A
ia eh oe, ‘EEO oe Rae
% 3 te oy . 4 ~ ay) ?
: cure ae ’
i i we i ) a aS re bred
' ° rae h y : ef .
"ea ess k :
yey x ¥ A ies ATE
re) r
co
‘
tat in Re
2 i we i
3 i a
| 4 s We) ¢
M fia ; ipo
vas) oe te
oni ear aI
, he A
.
$ \
ip
5 ee © vag A
ao ht ivy
/ ae * Ee ol Y
< yin Paha
fr bi Be CE
& hey ee
;
. %
Oe * ' } ey
; of ys
i [3
Oral a oer ‘a
: nee i WM ae oy
Lj u : i"
i R eee
i +4 he
Tee t
vb 4 7
ih iT ‘ y
MATERIALS AND METHODS
Nine slash pine progeny tests located at five northern Florida
sites were measured in 1984 (Table 1). Height, diameter at breast
height (DBH), and survival were recorded. Individual tree volume was
calculated using an equation developed by Goddard and Strickland (1968)
for five-year-old slash pine. Six tests had block plot designs with
tree densities ranging from 1,121 to 10,000 trees/ha. The three remain-
ing tests were Nelder designs with eight spacings. One Nelder test had
densities ranging from 472 to 3,089 trees/ha, whereas the densities of
the other two ranged from 4,800 to 43,100 trees/ha. Detailed descrip-
tions of the tests and test sites are given by Rockwood (1983).
Analyses of variance were performed on data from either randomized
complete block or split-plot designs in each study. Coefficients of
variation were determined for height and DBH data from five tests.
Spearman's rank correlation coefficients were calculated for volume data
between common families of comparison tests. Rank correlations were
also calculated between the volume data from each of the nine tests and
the volume clonal evaluations assessed by the Cooperative Forest
Genetics Research Program at the University of Florida.
RESULTS AND DISCUSSION
Family and Site Factors
Growth varied by progeny test (Table 1) and was influenced by site
quality. Good growth occurred in tests 8-78-1, 8-78-2, and 0-58, which
were planted on similar lower coastal plain sites. Tests 0-59 and 0-60
were planted on less fertile, flatwood sites and had only average
growth. Growth differences due to site were also evident between the
tests at Gainesville, FL (10-N, 10-P) and Trenton, FL (11-N, 11-P).
Early survival, however, was not affected by site quality.
Growth differences due to family were statistically significant in
every test except 10-P and 11-P. In tests that contained a commercial
checklot (i.e., 0-58, 0-59, 0-60), selected slash pine families
generally grew and survived better than the check; although an average
of 19% of the selected families had less plot volume than the check for
a given site. Since some families occasionally performed poorly on
particular sites, family mixes would be preferred over family blocks for
certain sites if no families have demonstrated superiority.
Family ranks for individual tree volume in each of the tests were
not statistically correlated with the standardized clonal evaluations
(Table 2), which are based on row-plot progeny tests planted at opera-
tional spacings. The low correlations for sites other than Cantonment
suggested that genotype x environment interactions were present.
Earlier comparisons of other families across various test sites also
suggested such interactions (Goddard et al. 1976, Goddard et al. 1982)
whenever site differences were large. Most highly-rated families were
consistent, however, in these tests. Since fifth-year performance in
101
Table 1. Mean growth performance for all slash pine families in each
progeny test.
Progeny Florida
Test No. Location Age Ht. DBH Vol. Surv.
(eae) b pre bin Brat (yrs) (m) (cm) | (3) (%)
8-78-1 Cantonment 6 oy 10.4 -0299 86
8-78-2 Cantonment 6 oe: 10.7 0297 90
0-58 Cantonment d/ USS 10.6 0296 87
0-59 Perry i 6.0 8.8 0189 89
0-60 Yulee 7 Dio 8.6 -0182 89
10-N Gainesville 5 yen 358) 0064 96
10-P Gainesville 5) 3.6 358) 0074 95
11-N Trenton 5 gd) 555) 6 (QIU 96
11-P | Trenton 5 5.8 S10 0121 94
Table 2. Spearman's rank correlation coefficients between progeny test
performance and composite clonal evaluations.
Progeny Florida No. Common
Test No. Location Families Volume
—= rt--—
8-78-1 Cantonment 7 Se.
8-78-2 Cantonment 7 -6l
0-58 Cantonment 22 -40
0-59 Perry 22 04
0-60 Yulee 19 S25
10—N Gainesville 32 -08
10-P Gainesville 21 -.08
11-N Trenton 25 19
11-P Trenton 17 - 06
102
conventionally spaced tests is not as well correlated with subsequent
measurements as are 1l0th-year data, the higher correlations for the
Cantonment sites may be due to the excellent growth which was equivalent
to that of older trees.
Genotype x environment interactions were less apparent in the
comparisons between common families in the nine tests (Table 3).
Excluding comparisons involving test 10-P, rank correlations were
positive 6 (r=sl62)= 9. 73))- Therefore, family rankings for growth
performance within these tests did not change significantly. The
failure of the standardized clonal evaluations to demonstrate genotypic
stability for family performance across sites limits the usefulness of
the evaluations for selecting superior families for planting. Further
study of the methodology used in family evaluation is needed, since
comparisons between the progeny tests (Table 3) indicated genotypic
stability did occur at a modest level.
Test comparisons (Table 3) emphasizing site differences rather than
plot designs generally had positive rank correlations (r=.33 - .57).
Since plot designs and spacings were identical, the largely non-signifi-
cant correlations from these paired tests suggested family x environment
interactions were occurring. These interactions affected growth of some
families on the poorest site, test 10-P. Correlations from the three
paired tests involving 10-P were negative, and two were significant.
Possibly the slash pine families that grew well on moderate to good
sites may have been more sensitive to site quality and therefore, were
physiologically predisposed to site and spacing interactions on poor
sites.
Family performance was also linked to geographic origin in tests
having commercial spacings (i.e., 0-58, 0-59, 0-60). Families from
southeastern Georgia and northeastern Florida consistently grew the
best. Volume growth of these families averaged 3% more than southern
Alabama and Mississippi families and 9% more than north-central Florida
families. However, these comparisons may be influenced by the unequal
and limited sample sizes from the geographic regions. Slightly differ-
ent geographic patterns were observed in the fifth-year data (Goddard et
al. 1982).
In addition to family performance across sites, intergenotypic
competition was investigated in test 8-/78-1. Contrary to findings by
Williams et al. (1983) for loblolly pine, no differences in growth were
found between pure plots (one family) and mixed plots (nine families)
which agreed with results reported by Franklin (1983) for studies on
loblolly family competition. Survival, however, was significantly
higher (+9%) in the mixed plots.
Plot Design and Spacing Factors
Comparisons of tests from the same site but having different plot
designs, from different sites with the same plot design, and from
different sites with different designs showed that, with the exception
of the 10 tests, plot design affected family volume rankings less than
did site differences (Table 3). Families in Cantonment tests 8-78-1
103
Table 3. Spearman's rank correlation coefficients for volume data of
common families across different tests and plot designs.
Test Plot Design No. Common
Comparison Comparison Families Volume
(m3)
ape
8-78-1/8-78-2 Block/Nelder 9 73%
10-P/10-N Block/Nelder 20 -. 46%
11-P/11-N Block/Nelder 13 ~57%
0-58/0-59 Block/Block 14 Sais
0-58/0-60 Block/Block 14 47
0-59/0-60 Block/Block 14 . 36
10-P/11-P Block/Block 17 -.05
10-N/11-N Nelder/Nelder 25 233
10-P/11-N Block/Nelder 18 -. 48%
10-N/11-P Nelder/Block 1155 SG
*Significant at the 5% level.
and 8-78-2 were significantly correlated (r= .73) despite the differ-
ences in plot design and spacings between the two tests. The 11-P and
11-N test comparison gave similar results. The similarity of family
volume production across two plot designs with different spacings
suggested family x spacing interactions were not influencing family
rankings. This was in contrast to Stonecypher and McCullough's (1981)
observations from a eight-year-old Nelder test of Douglas-fir. They
found family x density interactions at spacings (735 to 26,300 trees/ha)
greater than those tested in the 10-N and 11-N tests. There were,
however, significant spacing x family interactions for height in tests
8-78-2 and 10-N. Therefore, some slash pine families were affected
differentially by spacing at ages five and six, although most family
ranks changed little.
Spacing significantly affected family height and/or diameter (DBH)
growth in all spacing tests except 8-78-1. DBH was influenced by
spacing in tests 8-78-2, 0-58, 0-59, 0-60, 10-N, and 11-N. Height was
affected in tests 10-N and 11-N.. In the Nelder tests, spacing influ-
enced height, DBH, and volume in a manner similar to that reported by
Stonecypher and McCullough (1981) for Douglas-fir. DBH and volume
increased with each subsequent decrease in density. Height also
increased with decreasing density up to the two lowest densities where
it decreased slightly. Therefore, growth differences between the
progeny tests reflected spacing as well as site differences.
Age Factors
Spearman correlations between family height, DBH, and volume data
for years three versus five, four versus six, and five versus seven were
significant (r=.75 - .93) in all tests. Family performance rankings
changed little over two years, irregardless of the different spacing
treatments.
104
The coefficient of variation (CV) for growth data recorded in 1982
and 1984 from four tests planted at two operational spacings decreased
over time (Table 4). The CV for height of trees planted at wide spac-
ings varied from 0.9% less than to 1.5% more than the CV of narrow
spacing trees. The CV for DBH was 1.5-1.7% larger at the narrow
spacings in three of four tests. Similarly, the CV for DBH at five
spacings (4800, 8400, 14,600, 25,100, and 43,300 trees/ha) in Nelder
plots from test 11-N decreased from the densest to the widest spacing at
both ages three (34% to 25%) and five (27% to 21%).
Table 4. Coefficients of variation (CV) for height and DBH of all
families tested at two spacings and measured in 1982 and 1984.
CV for Height Al} CV for DBH
Progeny Test Wide/Narrow Spacings— Wide/Narrow Spacings
(Planting Yr.) 1982 1984 1982 1984
a a a (4) ---------------------
8-78-1 (1978) 13)52/ 14.4 Tile oy 18.4/20.7 eye OVALE)
0-58 (1977) LSO/el3 22 11.9/11.9 NSU /LS.3 14.8/16.5
0-59 (1977) 2 LSe5 Gao /5 51 DEY [PAS GIA. PPC MY JANE 8}
0-60 (1977) DPA PP ES L6no/ 17014 2973/3 ede 2025/22 40
a/
Spacings were 1,223 and 2,446 trees/ha, respectively.
Even though high densities appeared to encourage expression of
intrafamily variation at three, five, and seven years of age, no family
xX spacing interactions for DBH were found in the Nelder tests (8-78-2,
10-N, 11-N) where high densities were tested. Furthermore, the Nelder
data and the decreasing CV with time suggested that significant family x
spacing interactions may not occur in these tests in the future. Thus
the merits of using narrow spacings and alternate plot designs to evalu-
ate the growth potential of slash pine families at young ages have yet
to be demonstrated.
LITERATURE CITED
Franklin, E. C. 1979. Model relating levels of genetic variance to
stand development of four North American conifers. Silvae
Genetica 28(5-6):207-212.
Franklin, E. C. 1983. Patterns of genetic and environmental variance
in a short-term progeny test of loblolly pine. Proc. 17th S.
For. Tree Imp. Conf. pp. 332-343.
105
Goddard, R. E. and R. K. Strickland. 1968. Volume and weight tables
for five-year-old plantation grown slash pine. Univ. Fla.
Sch. For. Res. Conserv. Res. Rpt. No. 14. 7 pp.
Coddard,. Rk. E., .D.2 Lb. wRockwood, Ho URan Koka Cer Ane HOW lic wanda mene
Hendrickson. 1976. Cooperative Forest Genetics Research
Program 18th Progress Report. Univ. Fla. Sch. For. Res.
Conserv. Res. Rpt. No. 25. 20 pp.
Goddard, R. E., D. L. Rockwood, and H. R. Kok. 1982. Cooperative
Forest Genetic Research Program 24th Progress Report. Univ.
Fla. Sch. For. Res. Conserv. Res. Rpt. No. 33. 25 pp.
Rockwood, D. L. 1983. Alternative designs for progeny testing slash
pine. Proc. 17th S. For. Tree Imp. Conf. pp. 179-185.
Stonecypher, R., and R. McCullough, 1981. Evaluation of full-sib
families of Douglas-fir in a Nelder design. Proc. 16th S.
For. Tree Imp. Conf. pp. 55-76.
Williams, C. G., F. E. Bridgwater, and C. C. Lambeth. 1983. Perfor-
mance of single family versus mixed family plantation blocks
of loblolly pine. Eroc., 7th) (Si. Por. free) impr Gonia
pp. 194-201.
106
TWO-STAGE EARLY SELECTION:
A METHOD FOR PRIORITIZATION AND WEIGHTING OF TRAITS
Cheryl B. Talbert*/
Abstract--A simulation approach was used to evaluate the impact of a
two-stage early-screening-plus-field-test selection program on predicted
total selection-progress in a ‘mature-stand’ selection-goal. Early family-
screening , when followed by field testing for a specified time period, can
lead to more, less or the same expected gain than a standard ‘field-test-
only’ program, depending upon the correlations among the early-screening,
field and mature-stand traits chosen. If the early-screening and field-
test criteria chosen are either strongly positively or strongly negatively
correlated, both total gain and gain per unit time may be less than if the
early screening had never been carried out. This effect can be reduced by
appropriate allocation of selection intensity between the early-screening
and field steps. Economic analysis will be necessary to evaluate whether
the benefits of early screening (in quality of the early-test environ-
ment(si, reduced field-test size, and/or larger family-size and greater
selection efficiency for a given field-test size) will outweigh its neg-
ative impacts under the conditions faced by a particular organization.
INTRODUCTION
Selection considerably prior to harvest-age is an operational reality
in loblolly pine tree improvement, as a result of persistently high alter-
native rates of return and the crushing resource costs of carrying large
field tests over long periods of time. The excellent research of the past
decade into methods of greenhouse, laboratory and nursery selection for
improvement of later field performance (reviewed by Lambeth, 1983 and by
Talbert and Lambeth, 1984) has not yet produced results conclusive enough
to cause operational programs to move away from conventional field tests of
4 to 8 years duration. However, a number of organizations are seriously
planning one- or two-year greenhouse, lab or nursery trials to ‘screen out’
their poorest families for growth, quality and/or adaptability, in order to
reduce the size of their field-tests.
A likely scenario for such a program is selection in two or more
stages, where some proportion of a population of half-sib and/or full-sib
families would be discarded at some early age, or at several early ages, on
the basis of seedling traits. After this initial truncation, field tests
would be olanted with reserve seed from the remaining families, and the
final set of orchard parents or selections tor the subsequent generation
would then be chosen from those tests based upon survival, pest-response
and height or volume. This discussion will use the term early screening to
define truncation-selection based upon seedling traits.
Considerable theoretical work in the agronomy literature has shown
that, when different selection criteria are correlated, culling of the
population for one criterion can drastically impact gain-potential from
other criteria, and the total progress that can be achieved is strongly
if Scientist, Forestry Research Division, Weyerhaeuser Company WIC-2H2,
Tacoma, Washington 98477.
107
affected by the reiative correlations of each criterion with the desired
mature-stand traits, and by the relative economic importance of those
mature-stand traits (Jain and Amble, 1962; WNamkoong, 1970; Cunningham,
TPS) Muilti-stage culling can actually reduce total ‘value gain’ if the
ordering of traits and the selection pressure applied to each trait is not
properly matched ta the genetic and economic characteristics of the popu-
lation. At the same time, early family-screening can provide a number of
advantages over canventional ‘field-only’ testing, which could oaffset
these disadvantages (Talbert and Lambeth, 1984) - advantages in quality of
the test enviranmentis) (for example, the ability to evaluate rust resis-
tance in @ high-inoculum environment), in reduced field-test size, and/or
in larger family-size (and greater selection efficiency) for a given field-
test 51276.
This report will explore the impact of early truncations aon total gain
from a fairly simpie two-stage ‘early-screening + field-test’ selection
program, and will explore alternatives for maximizing gain from such a
program in populations with a variety of characteristics.
The Quantitative Basis for Two-Stage Selection
Selection is usually justified by the expectation that some desired
change will occur as aresult, in one or more correlated traits ithe
simplest example would be genetic value for the original trait). By the
same token, any early truncation of a population will influence the mean
and variance of later-staqge, correlated traits, thereby affecting the
progress achievable from selection on those later-stage traits. The impact
of sequential truncations on predicted gain can be quantified based upon
the characteristics af the multivariate-normal distribution (Eisen, 1983).
The general theory 15 adapted to the current example below.
Because early selection i5 most commonly carried out to improve av-
erage genetic value for one or more harvest-age traits, gain in a harvest-
age trait M resulting from early selection for the same or different trait
J 15 appropriately described using an equation for indirect selection:
MGn,5= isCoviM,d) /Ops3 3 (1)
where lg the selection intensity practiced on Jd,
Cov(H,J) = the covariance between genetic value for the harvest-age
trait and phenotypic value for the early trait, and
On; = the phenotypic standard deviation for J.
For individual selection this formula reduces to the familiar form:
AGn,3= ighshwarGns Opn 3 (2)
where fi = square root of the heritability, and
rGHJ = genetic correlation between the mature trait M and the early
‘gieehe. dle
108
Now, instead of one early-selection trait J, consider the case of two
early-selection criteria J and F. In the current context, J would be the
greenhouse/lab/nursery trait and F would be a field-test trait; therefore,
J must be assessed prior to F. If a proportion of families p(J) is selec-
ted based upon J, retaining a reduced population having a standardized
selection differential ig, the variance of the field trait F is reduced
whenever J and F are correlated:
Vee = Ver COON ES 8 og ( lass) (3)
Nal
where # = designation of adjustment for ist-stage selection,
Ve = the variance of F prior to selection on J,
Cov(dJ,F) = the covariance between J and F prior to selection on J,
anid: (2 aC = the standardized truncation point for J,
In addition, the selection intensity which can be applied to F after selec-
GHOM OM ly Limit is reduced, regardless of the correlation between J
and F, due to the fact that the proportion left for field selection out of
a fixed total proportion P to be selected, is reduced.
The cavariance between F and M are also affected by prior selection
whenever either F or AW are correlated with J:
Cov(F,M)* = Cov(F,M) - ECCov(d,F) x Covid,M) x iglig-ts)]
Vy
(4)
The correlated gain in M resulting from the selection on J can be
predicted by equation (i). However, the gain which can be obtained from
subsequent selection on F will be altered by the first-stage truncation,
because of the impact of the truncation on if, on the variance of F,
and on the correlation of F with H. Therefore, gain in M resulting from
two-stage truncation selection on J and F will be:
AGmh,a+e = AGu,s + ieCov(F,M)* / On-* s
The Simulation Analysis
To illustrate the impacts of first-stage selection on total progress
from a two-stage selection program, several simulated populations were
carried through equations 1-5. A number of combinations of genetic cor-
relations between J, F and M were used, chosen to represent a range of
combinations rather than to model any specific biological scenario (Table
ye For purposes of the simulation, individual-tree narrow sense herita-
bilities are assumed to be 0.2 for J, F and M, and the phenotypic variances
for J, F and 4 are held at 1, 10 and 100, respectively, for all of the
Simulations. Finally, it is assumed that each population consisted of 100.
half-sib families, with a constant family-size of 200 for the early screen-
ing and 60 in the field. Half-sib family selection is used in the early-
screening stage, and a combined half-sib-family and within-family index is
used in the field stage. It i5 important to note that, since reserve seed
109
used to plant the field-test in our example, none of the same trees
would be measured for both ¢ and F, although half-sios would be.
Table i. Ranges of selection intensity and inter-trait genetic correlations
used in the simulatian analysis.
Farameter Definition Range
of selection intensity for J Wood = ty 7s
By proportion of families selected for J FOE Sei:
rGse genetic correlation between J and F -0.4, G@.1, 0.4, 0.8
rGJM genetic correlation between J and i SUG a4 Wate Oot}
rGem genetic correlation between F and A 7052, 5054, 0.4, 082
RESULTS
Figures ia and ib illustrate the impact of increasing intensity of
early half-sib family screening for a trait J on the phenotypic variance of
a field-trait index F, and on the correlation between genetic value for M
and the phenotypic index value for F, when the genetic correlation between
J and M and between F and M prior to the early screening are equal in
ahsciute value at rG& = +0.4 or -0.4.
Regardless of the genetic correlations between J and fe and between F
and M, when the genetic correlation between J and F is strongly positive,
the variance of the field-trait index F decreases to less than two-thirds
of its ariginal (rG@(J,Fi=6) value as the proportion of families selected
for J reaches 20 percent (or 20 of 100 families selected). If the genetic
correlations between J and WH, F and M and J and F prior to the early-
screening are of the same sign, selection for J resuits in a decrease in
the absolute value af the correlation between F and 47%. On the other
hand, aif the genetic correlations between J and M and between F and M are
of the same sign and the correlation between J and F is of the oppasite
sign, then early screening for J will decrease the absolute genetic correl-
ation petween F and M if rG(JM) is positive, and will increase that
absolute correlation if rG(Jf) is negative. Finally, if r&(JM) and rG(FR)
are of opposite sign, then early screening for J will increase the absolute
value of rG&(FR) if the genetic correlation between J and F is positive, and
will decrease the absalute value of rG(FA) if rGtJH.F) is negative.
z/ The importance of changes in rGiFM) result fram the direct
proportionality of erG(FA) with expected second-stage gain. If expected
gain from selection on a given variable turns out to be negative, then
selection on the opposite end of the scale for that variable will make the
gain positive, and of the same magnitude. Therefore, absolute correlations
and expected-gain values are the quantities of interest.
110
1 "Oz P—- zyx
——---
--_——_—---
--_—_——-
—.
-_————— - -$ —————-~
’
zn 2POO EMD
- mana
ee
oe
—7>
aoa ©
= ae
pone
vrreerere haahidaaad sahaliahad Laldaaadid bdadaddada baseddiadd baaadadadd sadahadied baaddaanad haadaannad
40 45 so ss 60 6s 70 73 40 rT} 20 Fe -] 30 F.) a “4 60 Ss 60 68
WT Fi
PERCENT SELECTED FOR J PERCENT FOR J
FIGURE 1. The impact of early screening for a trait J on (A) the variance
of a correlated field-trait F, and (B) the correlation of F with a selec-
tion goal M. (Assumptions and definitions in text and in Table 1)
How can selection for one variable actually increase the correlation
between two other variables? This can happen because, under certain con-
ditions, removal of entries on the low end of the scale for J means removal
of entries which are simultaneously low for M (due to a positive correl-
ation of J and A) and high for F (due to a negative correlation of J
With By An opposite but equivalent example exists if J and M are nega-
tively correlated and J and F are positively correlated. The outcome is an
upward pressure on the absolute value of the correlation between F and fA.
The combined impact of these effects is illustrated in Figures 2a-Zc,
in terms of the predicted progress in the mature trait M which would obe
expected fram the two-stage selection program relative to that which wouid
be expected from selection for the field-trait index alone, when the gen-
etic correlations of J with M and of F with M are again equal in absolute
value at +0.4 or -0.4,
Regardless of the proportion selected in the earily-screening step,
when the genetic correlation between J and F is less than 6.4 or negative
and the correlations of J and F with the selection goal are both either
+0,.4 or -0.4, selection for both J and F yields more expected gain (in M)
than does field selection alone. In fact, under these conditions, if the
genetic correlation between J and F is negative the expected gain in MN from
the second-stage, field-selection step alone is greater than the total
expected gain would have been from a single-stage field selection on F. On
the other hand, if J and F are strongly positively correlated (30.4), and
both r&(JM) and rG(FM) are either +0.4 or -0.4, two-stage selection yields
Jess expected gain in M than does one-stage selection for F alone.
The opposite behavior occurs when the genetic correlations between J
and M and F and WM are opposite in Sign and equal in magnitude at :0.4;. In
111
One-Stage: 10.03
11.23
=z =e B= FD
20
One-Stage: 10.03
15.69
12.34
2,2. @
Neeees:
OO
OOO
OO
42)
:
%
20x
Ke -0.4 —4 Fk 0.1 —4 f-— 0.4 —4 -— 08——{ RIF
FIGURE 2. Predicted gain in a mature-stand trait M resulting from two-stage
‘early-screening-pius-field’ selection for a seedling trait J and a field
Penis HR compared to field-selection only, for two early-screening
intensities: (a) when rG(JH) and rG(FRA) are both 0.4 or both -0.4; and (b)
when rG{JH) and rGt(FH) are apposite in sign but equal in magnitude at
(G.4:. —_ (Assumptions and definitions are contained in Table i)
such a scenario, unless J and F are strongly negatively correlated, two-
stage selection yields greater expected gain than one-stage selection.
As expected, gain from two-stage seiection varies in magnitude with
degree of selection-pressure placed on the early-screening trait d,
ted. For every population scenario tested, rigorous early screening yields
greater gain overall than does low-intensity early screening. This occurs
despite the fact that greater early-screening gain sometimes results in
reduced again from field-test selection; the increase in gain resulting from
the increased early-selection intensity even in these scenarios is greater
than the decrease in gain from the field-selection step. Therefore, it
appears that it is possible to allocate selection intensity between early-
screening and field-test steps in a two-stage selection program so as to
make two-stage selection at least as effective, and often much more effec-
tive. in overall expected gain than a one-stage field-test-only program.
Needless to say, it i5 not realistic to presume that the genetic
112
correlation between a chosen early-Screening trait J and the ‘mature-stand’
selection goal M will always be the same as the genetic correlation between
the chosen field-test trait (F) and M. Figures 3a-3d illustrate the effects
of differences between J and F in the strength of their genetic relation-
ships with M, ranging from a case where the early-screening trait J is
less-strongly correlated with M than is the field-trait F (at r& = 0.2 and
0.4 respectively), to a case where J is twice as strongly correlated with M
than is F (rG = 0.4 and 9.2 respectively). (Again, the attempt here is to
utilize a range of theoretically-possible scenarios, rather than to pick a
single empirical example, soa that readers will have maximum latitude in
finding their own ‘most useful’ case.) Of course, when rG(FM) is 0.21,
the absolute gain fram field selection in both the one-stage and the two-
stage examples is less than it is when rG(FM) is {0.41. Relative to the
‘conventional’ one-stage approach, however, two-stage gain is affected in a
predictable way. In situations where the early screening has a detrimental
impact on gain from field-selection, a lower genetic correlation between J
and ff reduces expected gain from early-selection, and therefore decreases
the detrimental impact; the two-stage approach ‘looks better’ relative to
one-stage selection. In situations where early screening actually in-
creases expected gain from field-selection, a stronger genetic correlation
between J and M increases the magnitude of the positive effect, and two-
stage selection becomes proportionally better relative to one-stage selec-
tion. Even under these altered scenarios the greatest two-stage gain is
produced when rigorous selection is applied in the early-screening step.
It 15 important to note that the relationships and results discussed
here are dependent only upon underlying biological relationships, and are
independent of scale. If the scale of J, F or # is reversed, ail of the
relationships of that variable with other variables in equations 3-5 will
be affected, and the combined effects will cancel one another. re as
assumed in this analysis, however, that selection will be carried out
appropriately relative to scale: in other, words, that when rG(JH) or
rG(FM) is negative, the ‘lowest value’ entries for Jj or F will be selected,
5G that expected progress in M will always be positive.
DISCUSSION, RECOMMENDATIONS
Buantitative methods are most appropriately used not in making de-
cisions, but rather as tools to augment the knowledge and experience of the
decisionmaker . In the case of two-stage screening, the quantitative tools
usec in this analysis provide four guidelines:
Early family screening for a seedling trait J, when followed by
selection for a trait F in a conventional field-test, may actually lead to
reduced total expected gain in a ‘mature-stand’ selection goal (M) from the
two stages combined, compared to what would be expected from field selec-
tion alone. This will occur if the genetic correlations between J and M&M
and F and M are of the same sign, and the early-sScreening trait is strongly
positively correlated with the field-test trait (rG > 0.4), or alternative-
iy, if rGiJHh) and rG{FM) are opposite in sign and rG(JF) is strongly
negative (rG ¢ -0.4)., In all other scenarios tested, two-stage selection
provided at least as much and generally more total expected gain than one-
stage ‘field-only’ selection.
113
(“I aLqel ult pue 3x94 UL UPALH Bue SUOLZLULJapP pue SUOLydwNssy) ‘paLidde st uolq9.e
-|as A[ues yeam pue artzebau A, Huouzs St (4f)94 uayM (p) pue Spattdde st uolqoealas A Aesd SnouoblLu pue sALy
-Lsod A, buouzs St (4f)94 uayM (9) Sparidde st uolzoaLas A|Uea yeaM pue BALZebaU SL (4f)HA UaYM (q) ‘tpat{d
-de SL uOLzOa[as Aj uea snouocHbLu pue aAtzebau St (4f)94 UBYM (ke) 24 JLe42 7S97-plals e Aq paMo| [Oy CP 4Leug
BuLuaeuds-Aj ued ue AOJ SUOLZIALAS JHezS-OM} wWOuJ Hulyjnseu pw [eob uolqoejas e uL uLeb pazoLpadd “€ FYNdI4
sr BRD sC ee Gt tv
: 4 : “os
270-<M451 p-or=Hgu 204-454 p-oe=tgu p'0-e4o1 z-Orabigu poreM4qu 2-oreHgu z0-M4gs poral gu 2'0entgs y-pratt yu vo-aMgs zoel vrordtgs 2-o+- gu
aovis-2 39V1S-1 39viS-2 39Vv1S-1 B9viS-2 39ViS-1 39viS-2 39V1S-I Q5v19-2 39Vv1S-t 39viS-2 39V1S-1 35v18-t BoviS-2 39¥AS-1
SSO
es 55052
(HAKAS
BS
5e)
- BO
zr
wa-z
w<-—-
: ar
€0°01 sé 60°01
£1'st
er
16°21 @‘O-sr9y = 10Z-f NOX GBLITIS NOI lvoddud
a2
vt
@'G-sF98 = =6408"f BOS 03139735 NOL WOddUd
——.— - - ———_ -. — Dae o- -- - — See ee —. o - — ———— —- a
| poesey
66
tl z
2
. 5
i 20°S 20°S
EE
oS
4 = j
& I
W e
N
I
207
g 8 Vv 96°6 €0°O1 too | 7
Se
ra
; at 6t-2t
16'6 £0°O1 £0°O1 ‘
v ‘O-°3F9N xO2ef YO O31ITIS NOTLYOdded vt
v ‘O--4F9N xO8er YO G31939IaS NOL wOdOud Ze
er
Se
7
7
oat-=z
oct~-=z
114
. If a two-stage selection program is assumed to require i-2 more
years to complete than a one-stage program, most two-stage scenarios tested
yield less expected gain per year than one-stage selection, for the same
rG(FA).
The stronger the genetic correlation between J and A, the greater
the gain from first-stage truncation, and the greater the resulting impact
on second-stage gain. Under conditions where early screening is
detrimental to field-test-selection progress, the detrimental impact is
more severe when rG(JN) is increased, and less severe when rG(JM) is
decreased in absolute value. On the other hand, when early screening
increases expected field-test-selection progress, an increase in rG(JH)
increases gain from both the first and the second-stage.
. Regardless of the scenario, if a two-stage selection approach
is chosen, the greatest gain will result when strong selection pressure is
placed on the early-screening traits - the resulting increase in gain from
the early screening offsets the corresponding reduction in predicted gain
from the later field-selection step.
Host tree-improvement decisionmakers operate with some ‘feeling’ for
the general magnitude of interrelationships among selection-criteria, and
between selection-criteria and selection-goals; in many cases, there are
actual published relationships. Although experience-based parameter-
estimates will not always be unbiased, they are likely to be sufficiently
accurate to allow for decisionmaking based upon the above results. To
ignore these guidelines is to implicitly assume that the correlations among
early-screening and field traits are zero.
Because of today’s grueling economic conditions, eariy-screening may
appear very attractive as a means of reducing long-term field-testing
costs. These results show, however, that two-stage selection may not
always yield greater gain for the additional effort. The analytical methods
discussed here allow the tree improver to choose the most appropriate
relative magnitude of selection-emphasis to apply to different selection
criteria throughout a given testing-cycle, to determine the maxiaum pas-
sible two-stage expected gain under the specific conditions suspected to be
operating in the population of interest. It is critical that a careful
economic analysis then be carried out to determine whether two-stage selec-
tion is a desirable course of actian.
LITERATURE CITED
CUNNINGHAM, E.P. 1975. Multi-stage index selection. TAG 46: 55-641.
EISEN, E.J. 1983. Population genetics in animal improvement. Unpub. lecture
notes, N.C. State University.
JAIN, J.P. and YV.N. AMBLE. 1962. Improvement through selection at succes-
Sive stages. J. Indian Soc. Agric. Stat. 14: 88-109.
LAMBETH, C.C. 1983. Early testing - an overview with emphasis on loblolly
DUG On noe. CoO ree No. CON. — AtLhENS. GAs 297-51 hb.
115
Y i978. Index selection for genetic improvement of quantitative
characters. TAG SZ: 49-34.
WANEGONG, G. 197G. Optimum allocation of selection intensity in two stages
of truncation selection. Biometrics 26: 465-474.
TALBERT, C.B. and C.C. LAMBETH. 1984. Early testing and multiple-stage
selection. In: Prac. S23 Meeting on Progeny Testing, Baton Rouge, LA
{In Press). ,
116
DESIGN EFFICIENCIES WITH PLANNED AND UNPLANNED UNBALANCE
FOR ESTIMATING HERITABILITY IN FORESTRY
Barbara G. McCutchanz/
Abstract.--Both balanced and unbalanced data can be analyzed
for variance component estimation with Modified Maximum Likelihood
estimates in a unified approach. Design efficiencies are evaluated
for the estimation of heritability using this methodology, assuming
knowledge of the variance components. Rules for obtaining efficient
randomized block designs are established, The effect of number
of blocks, plot size, number of families, variance on family size
and total number of observations on design efficiency is examined
across the range of heritability and under 100%, 90%, 80% and 60%
survival.
Additional keywords: Modified Maximum Likelihood, Design allocation
rules.
INTRODUCTION
One of the problems that the experimenter faces in forestry designs for
the estimation of means and also variance components is the use of large
blocks in balanced experimental designs. Large blocks necessitate employment
of either a restricted set of environments where small plot variances can
be found or the inclusion of block=type variation among plots within blocks,
In this latter situation, the error. in estimating family means is increased,
effectively decreasing heritability.
Anderson (1975, 1981) and others suggest and evaluate intentionally un-
balanced designs for the estimation of variance components. These planned
unbalanced designs allow for the redistribution of the degrees of freedom
to the variance components of interest. Various unbalanced two-way designs,
useful for forest genetics trials of half-sib families with small randomized
blocks, are also possible to design (McCutchan 1985).
A complicating factor in most forestry experiments, and one which makes
design evaluation difficult, is that some level. of unplanned loss occurs in
a genetic trial subsequent to its establishment. Roughly 10% loss occurs
in loblolly pine (Pinus taeda L.) genetic trials after one year of field
growth, with up to 30% loss occurring by age 10, depending upon the incidence
of fusiform rust (Cronartium quercuum f. sp. fusiforme) (R. J. Weir, pers.
comm.). The efficiency of a cae for the estimation of particular variance
components or functions thereof, under states of loss, is of consequential
interest.
Ue a aa
— Quantitative Geneticist, Westvaco Forest Research, Summerville, SC. This
research was conducted while the author was a Graduate Research Assistant
at North Carolina State University, Raleigh, NC. The author graciously acknowl-
edges Drs. Gene Namkoong and Francis Giesbrecht for their guidance with this
research, the Statistics Department at North Carolina State University, Raleigh,
for providing the necessary computing funds, Ronnie Hise for his drawing of
the graphics, and the North Carolina State Hardwood Research Cooperative,
Raleigh, NC, for releasing the 1981 sycamore data.
117
The analysis of unbalanced data--either planned or unplanned unbalance--
is therefore important for two reasons: (1) after an experiment is completed,
an efficient estimator is needed and (2) before an experiment is conducted,
designs need to be analyzed for the possible efficiency with which parameter
estimates will ultimately be made (Namkoong 1981). Unfortunately, the most
commonly used analytical procedure, namely Henderson's Method 3 (1953), is
ambiguous as to which sum of squares is most appropriate; such estimators
retain only their unbiased property with unbalanced data, The maximum likeli-
hood type estimators have been known theoretically, but have not been avail-
able practically. Giesbrecht (1983) has written an efficient algorithm by
which Modified Maximum Likelihood (MML), Maximum Likelihood (ML) and MInimum
Norm Quadratic Unbiased Estimates (MINQUE) can be computed. The MML approach
is used for the remainder of this paper. The MML estimates, for which nor-
mality is assumed, are chosen because of their desirable properties regardless _
of the state of balance in the data. The estimates maximize the likelihood,
use the same information as the full ML estimates do and account, in some
sense, for the estimation of fixed effects; with balanced data, MML estimates
are also those obtained through the analysis of variance (AOV), which is not
true for ML estimates. The MML estimates are obtained by iterating the
MINQUE. The MML method is a unified approach to the estimation of variance
components and/or for comparing design efficiency.
The objective of this paper is to compare design efficiencies ofplanned
balanced and unbalanced designs for the estimation of heritability (h“). The
unbalanced designs allow for the inclusion of a large number of families in
relatively small blocks. The variance of the estimate of h2 (var(h2)) from
each design is compared to other designs across the range of h@ and with 10%,
20% and 40% random loss of individuals. Design efficiency is examined over
the range of h2 to indicate the quality of the design at any level of realized
h2 or for multiple traits which may have different h2 in the same experiment.
The design structure studied is a randomized block design on one location;
the treatments are unrelated half-sib families using either single-tree or
two-tree contiguous plots. The variance components are assumed known which
enables calculation of the variances of the variance components... An overview
of the results from McCutchan et al. (submitted) and McCutchan (1985) is pre-
sented,
METHODOLOGY
The notation and the computational methodology for Modified Maximum Like-
lihood follow Giesbrecht's (1983). His procedure for variance component esti-
mation is written as a temporary Statistical Analysis System (SAS) program
entitled Procedure MIXMOD.
The statistical model for each design considered is;
ns U Ae + Up ep t+ Ur en + Up ep t+ ey
nxb? bxie an nx fxd Cainxsesx le wiainxd
where Y is the column vector of n observations; p is the overall mean; Ups Ur
and Up are design matrices pertaining to block, family and plot effects, re-
spectively, with all elements equal to zero or one (where there are b blocks,
f families and s filled combinations of the families and blocks); for
118
single-tree plots, Up is the identity matrix of size n; Cg» &rs ep and OF
are independent column vectors of independent random variables, each with
: ; ; 2 2 2 2
zero mean and variance-covariance matrix I ops [.or, I 95 and I ows respec-
tively.
The variance-covariance matrix for the vector of observations (Y) is:
Pie We Les Z
V(Y) = UpUog + UpUror + Upon + Ilo ,
where . a of and on are the variance components due to the block, family,
plot and within-plot effects, respectively. Letting V. = UU: and for conve-
nience V= 1, V(Y), based on the parameters, can be rewritten as:
- ; 2 2 2 2
It is assumed that the f unrelated families chosen are a random sample
of the reference population, that those trees planted are all to be assessed
save for those lost and that blocks of different sizes are placed on different
parcels of land such that for a given set of variance components (04 of of da)
different sizes of blocks and plots can be compared,
If the variance components (05) were known, then the variance-covariance
matrix for the resulting MML estimates of the components (04 of of obs which
would then be MInimum Variance Quadratic Unbiased Estimates, would be:
=I ed 3
2ter(Q V0 oVs)3 i,j = B,F,P,W (2)
where Ci ee - ie i} aves) = los » and V . is defined in Eq. 1.
| oO oO Oo Oo o oO
The dispersion matrix (Eq. 2) of the variance components is a function of
the variance components themselves plus the design matrices (Uj). It is there-
fore possible to calculate this dispersion matrix for a given set of true
variance components and a design. The observational values (Y) are not needed
to calculate the dispersion matrix for the variance components. Heritability
is calculated based on the experimental structure as:
2 = doe/(oe + 0% + 04) = X/Y. (3)
h
Since variance component estimation is based on the experimental design in
only one environment, any environment by additive genetic interaction variance
is confounded with the estimates of the additive variance of Such an esti-
mate is appropriate for inferences only on this site type. The variance of
119
the estimate of he can be approximated by the variance of a ratio using a
Taylor's series expansion:
var(f?) = (1/¥)?var(X) = 2¢1/Y)(X/¥*)cov(X,¥) + (X/Y*)2var(¥). (4)
The computation of var(h°) is based on the calculation of var(X), cov(X,Y)
and var(Y) from the dispersion matrix (Eq. 2). The values of X and Y are
calculated from the set of variance component parameters. The approximation
of var(h*) relies on large sample theory,
The var (h?) is calculated for each of the designs in Figure 1 for two
types of variance component sets. Only the type of = of is reported here,
where of Se. oy = 1 and of = of take on values of .5, .1, .05 and .005 for ne
of 1.0, .33, .18 and .02, respectively. The actual assignment of the families
to the blocks is not given, as it_was found to be inconsequential in terms
of design efficiency (McCutchan et al. submitted). In each design the effect
of the random loss of 10%, 20% and 40% of the individuals on design efficiency
is examined. The var() reported for cases of loss is actually an average
of two independent samplings of individual loss,
ee bee PEE OLR EDT yt te ON Rie tay re
var(fs) = 0 var(fs) # 0
No. of Blocks No. of Blocks
g\t® 2°
ero 20(50) 40(25) pict 2 aoe) 40(25)
5 ; a
° ; = 60 page
fo) = = (20)
- Ee E
@ @
il =
e 7) ° ribet
_ 100 Latins
g & (10)
ico planned unbalanced design
Zz planned balanced design
(apes design not possible to evaluate
“4 No. of Blocks e No. of Blocks
sit git
ero* 10(50) 20(25) prot 10(50) 20(25)
2 TO
(oe)
° e 2 60
wo z E (10)
uN £ K
(= ff =
° °
a S 100
z gS (5)
Figure 1, Schematic diagram of designs studied for 1000 and 500 observations (n), and for equal (var(fs)=0)
and variable (var(fs)#0) family sizes. The var(fs) is proportional to the mean family size and
to n, The number in parentheses indicates the size of the effect; in the case of variable
family size, the number is the mean family size.
120
The effects of block size, plot size, family size, variance of family
size and total experimental size on design efficiency are each assessed at
four levels of survival across the range of h¢. To examine any one effect,
the var(h2) from each of two designs, which differ only in their levels of
this effect, are compared in a ratio. If the ratio is one, then one level
of the effect is just as efficient as the Bee level. If the ratio is
greater than one, then the design whose var(h“) is in the numerator of the
efficiency ratio is less efficient. In assessing the efficiency ratios for
each effect, the buffering to loss is discussed as is the comparison of
var(h2) to a particular criterion (Namkoong and Roberds 1974). The criterion
is established o = 50% = (Vvar(h2)’ /h2) x 100% for h2 > .2 and on
std(h2) = 10 Svar Roy tor h2 < .2, Ideally, a design is sought whose
var(h2) profile falls beneath that of the standard across the range of h2,
RESULTS AND DISCUSSION
All design Sols ratios are computed based on a given set of variance
components (of OF 95 OF) » regardless of block, plot or family size.
The effect of block size on design efficiency is illustrated in Table
1 for the 100-family Sales plot (STP) design with 1000 observations.
The ratio of the var(h¢) from the 20-block block design is less than that
from the 40-block design across the range of h2, The larger block design
is uniformly more efficient than a design with more smaller blocks, given
the same set of variance components.
The effect of random loss on such a comparison is shown in Table 2 for
h2 = .33. The ratio decreases with loss indicating that the larger block
design is better buffered to loss. This is true also across the range of h’,
ae specifics for other comparisons of the block effect are given by McCutchan
1985).
Table 1.--Design efficiency ratios Table 2.--Design efficiency ratios
as affected by number of Black: as affected by number of place for
for 100% survival, 100 tamilies wae ~30. LOO families, oIP and
STP and n=1000
n=
Number of Blocks
h Survival (%)
var (fc) EN =20 1/ var(fi*) ED =20
1.00 .0219 982 100 .0096 977
aS .0096 977 90 .0109 .974
18 .0068 .975 80 .0128 970
.02 0041 973 60 0195 958
I/ &) 09 is the ratio of the var(h*) from the 20-block design divided by that
from the 40-block design.
The initial motivation for examining the usefulness of smaller blocks
was the observation that. smaller homogeneous sites are more frequent than
larger homogeneous sites. In these comparisons designs have been examined
for the same set of variance components over the range of h@ and a range of
random loss, regardless of block and plot size. The designs with larger blocks
are 2% to 3% more efficient than those with 25-tree plots, and are better
buffered to loss, given the same set of variance components. The practical
application of these results includes consideration of the frequency at which
these larger sites can be found. For a fixed n, fewer larger sites would
be required than small sites; whether b large blocks could be found for a
given site type, of course, depends upon the site. Considering the use of 20
blocks, the results show that by using designs with blocks half the size,
a 2% to 3% loss in efficiency is incurred. (The loss in efficiency indicates
the increase in var(h¢) in having used 40 blocks versus 20 blocks). This
cost in efficiency has to be balanced against the cost of obtaining and main-
taining half as many blocks, each of twice the size. This latter cost may
include not only difficulties in locating such blocks, but also bias in
representing planting sites.
The effect of plot size on design efficiency is illustrated in Table 3
with the comparison of a single-tree plot (STP) design to a two-tree plot
(TTP) design given 100 families, 20 blocks and n=1000. othe STP design is
more efficient than that with TTP across the range of h” (Jable 3), with the
advantage in efficiency at 100% survival decreasing with h¢, The STP design
as is efficient with the imposition of random 10%, 20% and 40% loss
Table 4).
Table 3.--Design efficiency ratios Table 4,.--Design efficiency ratios
as affected b lot size for 100% as affected b fot size for he =
Survival, LOO families, 20 Dlocks 33, 100 families, 20 blocks and
and n= 1000 T= 1000
Plot Size Plot Size
2 STP rip 3 STP ITP
h ~ Survival (%) —-
var (he) ae var( fi) eee
1-00 20215 1.2351 100 -0094 te 1S15
533 0094 1.1815 90 .0107 1.1768
ss} 0066 1.1676 80 .0124 1.1685
02 .0039 1.1570 60 .0187 1.1675
pp is the ratio of the var(he) from the TTP design divided by that from
af Ue design.
The premise of using a TTP design is to protect the data set against
loss of plots. An AOV can be used for balanced data on a plot mean basis.
Loss of plots is not a computational or interpretative obstacle with the MML
methodology. There is a statistical cost to using TTP, as observed here,
which even at that fails to insure plot survival.
The large number of small family design is more efficient for 100% sur-
vival and high heritabilities than the small number of large family design
(Table 5, Ef-59). This result is reversed for low heritabilities, where the
larger family design is more efficient. The effect of random loss on the
design efficiency is given in Table 6 by Efe59 based on 10%, 20% and 40% ran-
dom loss. At each h@ given, the var( (iz) # Fae the 50-family design increases
122
less than that from the 100-family design. The buffering capacity of the
design to loss is greatly influenced with these size differences in families,
there being roughly twice as much buffering capacity at h@ = 1.0 for the
50-family design compared to the 100-family design, with this difference de-
creasing with h@, This greater buffering capacity with 50-family designs
is reflected in a decreasing E¢-59 with loss. The 50-family design, with
this lawgs difference in buffering capacity, becomes more efficient with 10%
loss at h© = .33 in contrast to the block or plot effects. In neither of
these designs are families lost through random loss of individuals.
Table 5,--Design efficiency ratios Table 6.--Design efficiency ratios
as affected by number of families as affected by number of families
for 100% survival, 40 Dlocks, SIP f Yd 33, 40 blocks. SIP and
OnMmno=, OCKS an
Number of Families Number of Families
2 100 50 . 100 50
h no Survival (%) ee
WA 1/ 2
var(h*) Esc var(h*) Ee_cg
1200 .0219 1.476 100 .0096 1.026
235 .0096 1,026 90 .0109 ~982
18 .0068 024 80 ,0128 2935
S020 =: .0041 917 60 01:95 .830
EN PR IDES SI SI ES SE
l/ £, ey is the ratio of the var(h*) from the 50-family design divided by
that from the 100-family design.
The ye ica ions for design recommendations are that STP and large blocks
provide low var(f2) across the range of h2, but that the family size that
should be employed depends on the heritabilities of interest. The 100-family
design is more efficient across a large portion of the range of h2, If design
allocations included only balanced designs, this efficient 100-family, 20-block
design would not be a viable alternative. If all the traits of interest have
low heritabilities, for example, less than .2, then a 100-family design would
not be the most efficient design to use. The 50-family design would be more
efficient in this range and have greater buffering to loss.
The effect of variable family size in contrast to equal family size on
design efficiency is illustrated in Table 7 for 100 families of average size
10. The equal family size design is more efficient at h2 > .33 than the vari-
able family size design at 100% survival. The variable family size design
is more efficient below this level of h@, increasingly so as h@ decreases.
These results confirm the suggestion (McCutchan et al. submitted) that in-
creased variance on family size might result in increased efficiencies for
low h2. They found that for 100 families of average size 10, the design with
var(fs) = 7 based on a binomial distribution with mean 10 was 2% less efficient
at h2 = 1.0 than the equal eau size design. The variable family size design
became more efficient at .25 > h© > .21 than the equal family size SES EL
having a var(h2) 6% less than that of the equal family size design at h¢ = .02
Variance of family size equal to 60 is examined here. The variable family
size design is less efficient at h2 = 1,0, 21% higher var(f2), and more effi-
cient at h¢ = .02, 35% lower var(fi2), than the equal family size design.
123
In addition to the 100% survival case studied by McCutchan et al. (submitted),
the effects of 10%, 20% and 40% random loss on this comparison are given
(Table 8). The variable family size design is better buffered to loss at
hgritabilities other than 1.0. The buffering is such that with 40% loss at
h* = ,33, the variable family size design becomes more efficient.
Table 7.--Design efficiency ratios Table 8.--Design efficiency ratios
as affected by variance on the as affected by variance on the
Family size tor 100% survival, 40 family size for h2 = .33, 40 Dlocks
blocks, SIP, 100 families and SUPT 100 families and n= 1000
Variance of Un Size Variance of Ean Size
ne Survival
var(h?) Eee ae (%) var(h?) EV 60
1.00 .0219 e206 Aa 100 0096 11052
433 -0096 1.1052 90 0109 1.0749
.18 .0068 9596 80 .0128 1.0432
402 0041 6534 60 0195 9633
1/ Ey=60 is the ratio of the var(h*) from the variable family size design
(with variance equal to 60) divided by that from the equal family size design.
The variable family size effect on design efficiency is an extended ver-
sion of the family size effect. Use of variation on the family size effec-
tively increases the average family size through an asymmetric effect of the
larger families. The deliberate use of variable family size can be viewed,
consequently, in a similar Hight as family size, in that its use depends upon
the portion of the range of h* in which interest in estimation lies, If family
sizes are unequal due to differential fecundity or survival, then for an aver-
age family size n, a variable family size design will actually be more effi-
cient at the lower range of h2 than an equal family size design,
A general comment can be made concerning n=1000 and n=500 designs in
UCU a to the criterion. Only at the 60% survival level for either
h¢ = .18 and/or h@ = .02 are var(f2) values from the 1000-observation designs
greater than the criterion. However, values from the 500-observation leaigiils
ane generally greater than the criterion for all levels of survival at h¢ =
OSC RANG OZ.
As an example of evaluating a given design for the estimation of he,
h2 and var(h ) are estimated from a North Carolina Forest Service installed
American sycamore (Platanus occidentalis L.) mother tree trial. The experi-
ment, located in McDowell County, N.C., has 7 blocks, 30 half-sib families
in 10-tree row plots and a total of 1866 surviving trees (89% survival).
Variance components were estimated on eight-year-old height data (ft.), con-
ver oyng in one iteration: h2 = .25 and var(h2) = .01. The estimated variance
on h“ is less than that suggested by the standard, namely .016. The results
of the Recall show that for a design epyecluihigiice primarily for the estima-
tion of h¢, an equivalent level of var(h*) can be obtained with half as many
total observations as in this trial.
124
CONCLUSTONS
Utility of efficient Modified Maximum Likelihood estimators is afforded
by recent computational methodology. Both balanced and unbalanced data can
be analyzed for variance component estimation in a unified approach. Design
efficiencies are evaluated for the estimation of heritability using this
methodology and assuming knowledge of the variance components. Rules for
randomized block design allocation are established based on using the same
set of variance components regardless of block, plot or family size. Single-
tree plots in large blocks are recommended if the plots within blocks have
small homogeneous variances=-smaller blocks if the above is not possible,
Recommendation of a particular family size depends on the portion of h2 range
in which estimation interest lies. Five hundred observations are insufficient
to achieve the set standard on estimating h2, One thousand observations will
achieve this standard if survival is at least 80%, The rules indicate that
there is not one design allocation which will uniformly provide a low var(h2)
across the range of h¢,
The research has based design efficiency on the estimation of he, Herit-
ability is but one function of the variance components; the methodology is
laid out for the examination of other functions of variance components, This
sort of a priori examination of design efficiency offers the experimenter
a strong tool in achieving experimental design objectives.
LITERATURE CITED
Anderson, R. L. 1975. Designs and estimators for variance components. In:
A Survey of Statistical ves ig and Linear Models. J. N. Srivastava (Ed.).
North-Holland Pub. Co., pp. 1-29.
Anderson, R. L. 1981. Recent developments in designs and estimators for
variance components. _In: Statistics and Related Topics. M. Csorgo,
D. A. Dawson, J. N. K. Rao, A. K. Md. E. Saleh (ras, North-Holland
PubiesCOl pps 3=22.
Gnesbnechta FG. 1983. An efficient procedure for computing MINQUE of vari-
ance components and generalized least squares estimates of fixed effects.
Commun. Statist.-Theor. Meth, 12(18):2169-2177.
Henderson, C. R. 1953. Estimation of variance and covariance components.
- Biometrics 9:226-252,
McCutchan, B. G. 1985. Design efficiencies with planned and unplanned un-
balance for the estimation of heritability in. forestry. Ph.D. disserta-
tion, North Carolina State Univ., Raleigh, NC. 177 pp.
McCutchan, B. G., J. X. Ou and G, Namkoong. A comparison of planned unbalanced
designs for estimating heritability in perennial crops. Submitted to
Theoretical and Applied Genetics.
Namkoong, G. 1981. Variance component. estimation. XVII IUFRO World Con-
gress, Div. 6, Japan. pp. 149-159,
125
Namkoong, G. and J. H. Roberds. 1974, Choosing mating designs to efficiently
estimate genetic variance components for trees, Silvae Genet. 23(1-3):
43-53,
Weir, R. J., Director, North Carolina State University-Industry Cooperative
Tree Improvement Program, NCSU, Raleigh, NC, 1985, Personal communica-
tion.
126
WITHIN-TREE VARIATION IN
CORTICAL MONOTERPENES OF SLASH PINE
Susan V. Kossuth and H. David Muse
Abstract.--Cortical monoterpene composition, bud diameter,
and length of the current flush from all buds (545) on a 10- to
12-year-old grafted, high gum-yielding Pinus elliottii Engelm.
(slash pine), were analysed to determine within-tree variability.
Approximately equal numbers of high-, medium- and low-vigor buds
were sampled from the upper, middle, and lower crown position
from the north and south aspects in the spring, summer, fall, and
winter. No north or south differences were found. Beta-pinene. —
content was significantly greater in the spring than in the other
seasons and the inverse was true for a-pinene and 8-phellandrene.
Alpha-pinene content was significantly greater in buds from the
lower part of the crown. Beta-pinene and f§$-phellandrene did
not vary with crown position. Alpha-pinene content decreased
from low- to high-vigor buds and the opposite was true for
8-phellandrene. Beta-pinene content was highest in low-vigor buds.
Bud diameters and lengths of the flushes decreased progressively in
size from the upper to lower crown, and from high- to low-vigor
buds. Bud diameters were similar among seasons, and lengths of
the current flushes were slightly longer in the spring than
summer.
Additional keywords: terpene, oleoresin, gum, clone, ramet,
seed orchard, a-pinene, B-pinene, 8-phellandrene, limonene,
myrcene, Q-phellandrene, Pinus elliottii.
1/ Supervisory Research Geneticist and Station Statistician, respectively,
USDA Forest Service, Southeastern Forest Experiment Station, 1143 Fifield
Hall, University of Florida, Gainesville, FL 32611, and Forestry Sciences
Laboratory, Carlton Street, Athens, GA 30602.
Appreciation is extended to John Munson, Annette Holliday, Junior Broomfield,
and Tillman Richards for technical assistance.
©
127
INTRODUCTION
Analyses of the composition of monoterpenes in conifers has been a
useful tool for identifying the geographic source of seed in plantations,
identifying hybrids and inbreeding, and for identifying ramets in seed
orchards (Squillace 1976; Kossuth and McCall 1984). Damage to trees from
grazing animals and resistance to insects and disease have also been linked
to the monoterpene composition of trees (Squillace 1976). In an early
study it was suggested that before large-scale studies of monoterpene
composition are undertaken, the best place to sample on a tree should be
determined so that repeated sampling would give consistent results (Hanover
1966) for individual tree phenotypes. The objectives of this study were to
determine the effects of season, aspect, position in crown, and bud vigor
on monoterpene composition of one slash pine (Pinus elliottii Engelm. )
ramet.
METHODS
One 10- to 12-year-old grafted ramet of high gum yielding slash pine
clone number 335 in a seed orchard at the USDA Forest Service, Southeastern
Forest Experiment Station, Olustee, Florida, was used for the study. It was
12.9 m tall, and had a live crown height of 9.8 m.
The ramet was divided into north and south aspects, and these were
subdivided into three equal lengths of crown for upper, middle, and lower
position. In the spring, all the buds within each of the six sections on
the ramet were classified as high, medium, or low vigor. A count was made
of the number of buds in each vigor class by section and ramet. One-fourth
of the buds from each section on the tree were sampled for cortical
monoterpenes in the spring (5/5/82), summer (7/12/82), fall (10/6/82) and
winter (1/12/83). All 545 buds were sampled.
Samples were taken by removing approximately the terminal centimeter
of the bud, collecting the cortical oleoresin that flowed out, and
immediately placing it in vials containing pentane. If the quantity of
oleoresin was low, the bud--with bud scales removed--was extracted in the
pentane. Gas-liquid chromatographic analysis was conducted according to
the method of Kossuth and Munson (1981) by using a 5840-A Hewlett-Packard
gas chromatograph with an automatic sample injector and programmable
integrator. The amount of each monoterpene is presented as a percentage of
the total monoterpenes since this method has been shown to have the least
variation (Powell and Adams 1973). At the time of sampling, each bud
diameter was measured.. The length of the current flush for the bud sampled
was measured in the spring and summer only.
The percentage data was subjected to an arcsin square root
transformation and analysed by analysis of variance (ANOVA) and Duncan's
multiple range test. If significant interactions were present, then
additional ANOVA was performed on each level of each factor involved to
further investigate main effects. Higher order interaction mean squares
were used to approximate associated error terms for each analysis.
&
128
RESULTS
The average monoterpene content for the tree was 30.8, 43.0, and 21.6
percent Q-pinene, B-pinene, and B-phellandrene, respectively. Of the other
constituents in the monoterpene fraction, camphene, myrcene, a-phellandrene
and limonene each contributed less than 4 percent and were not analysed
statistically. These low levels are probably of little physiological
significance. Alpha-pinene, 8-pinene and 8-phellandrene contents ranged
from 24.5-47.7, 18.2-54.5, 11.1-31.9 percent, respectively.
Overall, the content of the three major monoterpenes showed no
significant differences for the north and south aspects (Table 1).
Table 1. Significance levels for monoterpene composition, bud diameter, and
length of flush determined by analysis of variance, by source of variance.
Bud Flush
Source of variance Q-pinene f-pinene §-phellandrene diam. length
Aspect
Position %* KK sex
Season bated x¥ wx
Aspect x season
Position x season
Vigor week wevek KKK kK le
Vigor x aspect
Vigor x position wie seeker
Vigor x season
Pid ale
x, AK, ARK, RKKX indicate significance differences at the 0.05, 0.01, 0.001,
Bd 0.0001 levels, respectively.
Alpha-pinene content increased progressively with decreasing bud vigor and
the opposite was true for B-phellandrene (Table 2). Beta-pinene content
was significantly lower in the high- and medium-vigor buds than in the low
vigor buds. Alpha-pinene and 8-phellandrene were significantly lower in
the spring than in the other seasons, and the inverse was true for 8-pinene
for high- medium- and low-vigor buds (Table 3).
Alpha-pinene content was higher in the lower part of the crown than
in the middle and upper crown. Beta-pinene and 8-phellandrene content did
not vary with crown position (Table 2).
Bud diameter and current flush length of the buds sampled decreased
from high to medium to low-vigor buds, and from the upper to the lower
crown (Table 2, 4,). Bud diameter and flush length varied significantly
with position in the crown and bud vigor (Table 1). There was a significant
interaction of bud vigor and crown position for both bud diameter and flush
129
TABLE 2. Cortical monoterpene composition, bud diameter, and current length of flush for buds collected from
rafted high gum-yielding slash pine ramet number 335, by aspect, bud vigor, season, and position in crown.
Source No. o-pinene 8-pinene 8-phellandrene Bud Dia. Flush lgth. No.
ee as ercent Ae ck Wee cm
Overall 545 30.8+ 2.4 43.0+ 3.8 DP OTE). 7 NS} Goal slats) 259
Range “lis SONI 6 U 18. 2-54.5 dala 33 1259 OS,
Aspect
North 280 Wo Whe eck 43.4+ 3.5a 21.2+3.7a 13.0+ 9.1la 144
Range 195) 6 JAAN V7) 34.2-54.0 is2 3129 Ne SO36 7
South 265 30/564) 2..58 42.6+ 4.0a 22.1+ 3.8a 14.3+ 8.5a tS
Range 24.5-47.7 18.2-54.5 eel 30 5 ql 7252
Vigor:
High 62 ZOIDS 2 Ze UVa doe Sie yAo) 2372+ Shoa 27 -Dt14. 1a 24
Range 25.2-34.9 34.2-54.5 14.6-31.9 OS 7
Medium B55 30.64 2.2b 42.6+ 3.6b 227342 3.048 14.5+ 5.6b 94
Range DAS 370 B25 405 1258-3055 Ve = SrD
Low 148 31.7+ 2.8a 44.3+ 4.0a 19.5+ 3.8b (Goh eo Ne 75
Range 25.1-47.7 18.2-54.0 Pal 2955 Distal OS
Season
Spring 120 2859+. 254d 47.1+ 2.7a USiy2t Ze Lb 14.5+°7.4a 120
Range 24.5-36.3 40.2-54.5 P11 =24.58 1 3-47-56
Summer 138 31 0m 27a 41.1+ 3.1b 23.1+ 4.4a 12.94 9.8a 138
Range 26.1-40.2 34y. 2-51] <2 PLe9-3 159 1 Ike / S67
Fall 137. 31.0+ 1.5a 42.3+ 3.2b 22.6+ 3.2a ==
Range 27 38-36..9 Spal asoe) 12-9 -—297 5 22
Winter 150 Bl ..9t Zola 42.0+ 3.3b 22st Sy Oa oa
Range 29.1-47.7 SR 250). 2 eS =28i-4 i =
Position
Top 171 30.5+ 3.0b 42.7+ 4.la LO Dt S08 20 oe 0a 74
Range 24.5-47.7 18% 2-54..'5 IZ 9=30RS e037!
Middle 227 30.7+ 2.0b 43.5+ 3.6a ZAC lsht Se Gal 103 9s5).1 1a 3
Range 25:19 37 38 34.2-54.0 S23 19 SO 2 513
Lower 147 Si S22 3a 42.6+ 3.7a ZA ste Or Val 10.7+4.7c V2
Range 2594 0).2 34.8-50.4 11 1=29%.5 6 PADS)
130
—_—_—_——_.... nn —nmnm—n—n— nL
Mean + standard deviation. Column values within groups followed by different letters are significantly
different at the 0.05 level.
TABLE 3. Cortical monoterpene composition, bud diameter, and length of flush for buds
collected from one slash pine ramet, by bud vigor and season.
Bud vigor
and season No. d-pinene B-pinene f-phellandrene Bud Dia. Flush lgth.
é- Votes Pxcpenen to MMctetolte sPeCT CCM tCer. cane. « Memes mm cm
High os bot
Spring 12 Taf SN GE) 47.3+ 2.8 Teo 74 19 9.1+ 0.9 GP) 5(6)
Range Za 2-293 43.1-54.5 1436-2155 8.0-10.0 ES 7 MATL GS)
Summer 15 ZO sat Zk BOR 245203 26.4+ 2.4 Jems bal) IT she Tfe 72
Range 2651-33). 9 342-420 PSS MoS ikge) 7.0-13.0 1803 i7,
Fall 16 30.0+ 0.8 Bist O38 Mes Del) Over 2.1 Ls
Range 2856-3230 37.8-49.2 18.4-27.5 SG Weeahs ya (0) 30
Winter 19 319-5 4le72A4.8 22 otek SrOte ie ac
Range 29.8-34.9 394-4557 19'33:>26...4 7<0-12.0 =i
Medium
Spring 73 28.8+ 2.2 46.8+ 2.8 18.8+ 1.7 7.14 0.9 15.2 Ba
Range TA I NSS Hf 40.2-54.5 [ono = 24.6 5.0-10.0 Do SasiA6 ik
Summer 84 Noone! Ags 40.8+ 3. 24.0+ 3.9 Tel ctl 0 13n St Or
Range 26.4-37.6 36.2-48.7 WAS ISO BE) >0=10 IG Ps heh55)
Fall 79 30.84 1.3 42.0+ 3.1 231218 Tis Orta ere 2S
Range 27.8-30.0 815)9 (ODES 14.9-28.5 50-1050 AS
Winter 99 SGyae bod QD Ziad IRS SPS PAGS) 7.0+ 0.8 ae
Range 29.3-37.4 32.1-46.6 592 Si 4.0-10.0 2
Low
Spring 35 29.7+ 2.6 47.6+ 2.4 oO 26S) 5.44 1.2 8.3+ 4.1
Range 25).0-36:.3 43.3-54.0 JIS Wowk 63} A OE aSr0 GNA 3}
Summer 39 Sas ise Ao 42.6+ 3.1 ZO0LE 4.4 4.74 0.8 Se leas
Range 27.9-40.2 SIGS ekg 7 SO 27.7, SE0S* 7/50) of ASES)
Fall 42 Bibs Hoe hots 43; 74-3). Aeeae SoS) Soke Woe =
Range 28.0-36.9 3578-51210 13.0-30.0 JoW> 7a) a
Winter 32 32044 Sel 44.04 5.4 Oe ei 5.0+ 0.6 --
Range PAN NON Gif USRZS DOL I 8a26e4 A O0=10.10 as
Mean + standard deviation.
131
Table 4. Diameter and flush length of buds by bud-vigor class and position
in crown.
(Oi ee
eR AS a SP SS SE an VSD
Source of variance _ Bud diam. Flush length
VIGOR
High x
Upper 10.3a 39.1a
Middle 8.8b 20.7b
Lower 7.6b 16.1b
Medium x
Upper 7.9a 19.8a
Middle Zolb 1352_
Lower 6.5b UNV Sho)
Low x
Upper : 5.8a 9.4la
Middle 4.9b 5.7a
Lower 41D Sala
POSITION
Upper x
High 10.3a 39.1la
Medium 7.9b 19.8b
Low Saqtere 9.4c
Middle x
High 8.9a 20.7a
Medium Yo ile) 13.2ab
Low 4.9c Sanh
Lower xX
High 7.6a 16.la
Medium 6.5b 11.4b
Low 4.5¢ Dae
ee SE SSE SIS Ss cess sewn
Column values in each group followed by different letters are significantly
different at the 0.05 level.
132
length (Table 1). Further analysis by bud vigor to determine position
effects showed that any conclusions about these effects depended on bud
vigor (Table 5). However, analyses by position to determine bud vigor
Tabilieu Si. Significance levels of bud diameter and flush length determined by
analysis of variance of effects of bud-vigor class on position in crown, and
of position effects on vigor.
Variable Bud diam. Flush length
Vigor effects on position
Upper WHKK
Middle xX
Lower eke xx
Position effects on vigor
High week week
Medium *
Low x <5
pe a indicates Sipniticantly different, at) the, 0-05, 0-01. 0.001,
aia 0. 0001. levels.
effects showed that conclusions about position effects can be made without
regard to bud-vigor considerations. High-, medium- and low-vigor buds
tended to be larger in the upper crown and decrease toward the lower crown
(Table 4). Flush lengths followed the same general pattern.
DISCUSSION
The seasonal differences in monoterpene composition found in this
study are consistent with those found in P. taeda L. (Rockwood 1973) and in
Picea glauca (Moench) Voss, P. pungens Engelm. » P, mariana (MiB) eBeSe ee
and Abies balsamea (Ga) Mill., » when several samples from each tree were
combined for each determination (vonRudloff 1972, 1975a, 1975b; vonRudloff
and Granat 1982). Sampling in the spring seems to result in the most
variation and should be avoided. Based on the standard deviations and
ranges for the monoterpenes, 8-pinene and 8-phellandrene vary the most, and
a-pinene is the most stable.
Because of the significant differences in monoterpene composition
associated with bud vigor and the lack of any interaction with season, trees
133
should be sampled by using buds of the same vigor. Because low- vigor buds
are small and difficult to sample there is a greater chance of getting some
xylem oleoresin, which tends to be lower in §-phellandrene (Squillace and
Fisher 1966) in the sample. Sampling is easiest from the lower crown where
more low-vigor buds are found. Although d-pinene was somewhat higher in
content here, the differences were not large. Kossuth and Muse (1985)2/
determined that combining five buds is adequate to ensure an error of, at
most, 5 percent in phenotypic determinations of the three monoterpene
concentrations with a 95 percent confidence limit. Based on the data in
this study, it is recommended that sampling for determining individual tree
phenotypes be done in the summer, fall, or winter from the lower crown from
buds of similar vigor or size, preferably large buds.
_2/Kossuth, S. V., and H. D. Muse. Variation in monoterpenes among slash
pine ramets by season, aspect, position in the crown, and bud vigor.
(In process).
LITERATURE CITED
Hanover, J. W. 1966. Environmental variation in the monoterpenes of P._
monticola. Phytochemistry 5: 713-717.
Kossuth, S. V., and J. W. Munson. 1981. Automated terpene analysis with
an internal standard. Tappi 64: 174-175.
Kossuth, S. V., and E. McCall. 1984. Identification of seed orchard ramets
using monoterpenes. Proc. North Am. For. Biol. Workshop 8: 154.
Powell, R. A., and R. P. Adams. 1973. Seasonal variation in the volatile
terpenoids of Juniperus scopulorum (cupressaceae). Am. J. Bot. 60:
1041-1050.
Rockwood, D. L. 1973. Variation in the monoterpene composition of two
oleoresin systems of loblolly pine. For. Sci. 19: 147-153.
Squillace, A. E. 1976. Analyses of monoterpenes of conifers by gas-liquid
chromatography. Chap. 6, pp 120-157. In Modern Methods in Forest
Genetics. J. P. Miksche (ed.): Springer-Verlag, West Berlin.
Squillace, A. E., and G. S. Fisher. 1966. Evidences of the inheritance of
turpentine composition in slash pine. In Jt. Proc., 2nd Genetics Workshop
and 7th Lake States For. Tree Improv. Conf., USDA For. Serv. Res. Pap.
NC-6: 53-60.
vonRudloff, E. 1972. Seasonal variation in the composition of the
volatile oil of the leaves, buds, and twigs of white spruce (Picea
glauca). Gan. J! Bot. 50: 715905-1603.
134
vonRudloff, E. 1975a. Seasonal variation in the terpenes of the foliage
black spruce. Phytochemistry 14: 1695-1699.
vonRudloff, E. 1975b. Seasonal variation of the terpenes of the leaves,
buds and twigs of blue spruce (Picea pungens). Can. J. Bot. 53:
2978-2982.
vonRudloff, E., and M. Granat. 1982. Seasonal variation of the terpenes
the leaves, buds, and twigs of balsam fir (Abies balsamea). Can. J. Bot.
60: 2682-2685.
135
of
of
FIELD PERFORMANCE OF LOBLOLLY PINE
TISSUE CULTURE PLANTLETS
L. John Frampton, Jr., Ralph L. Mott and
Henry V. Amersonl.
Abstract.--Loblolly pine tissue culture plantlets of
cotyledon origin were compared to seedlings from the
same half-sib families after three growing seasons in the
field. Early growth in the field was slower for the
plantlets than the seedlings although the plantlets
appeared more resistant to fusiform rust.
Morphologically, the plantlets appear more mature than the
seedlings. Further studies to understand and manipulate
these differences are underway.
Additional Keywords: Pinus taeda, vegetative propagation,
tree improvement.
INTRODUCTION
A major potential benefit of tissue culture to forestry operations is its
use as a method of vegetative propagation of elite genotypes from tree
improvement programs. Seed for planting stock produced from currently applied
seed orchard technology captures only additive genetic effects; however,
commercial propagation of planting stock via tissue culture techniques could
utilize all (additive and nonadditive) genetic effects. In loblolly pine
(Pinus taeda L.), such technology could conservatively increase genetic gains
by one-third to one-half (McKeand 1981). Based on published estimates of
genetic variance components (McKeand et al. 1985), genetic gains for some
traits such as volume growth and disease resistance may double by employing
tissue culture technology.
In addition to greater genetic gains, a large decrease is expected in the
length of time between selection of improved individuals and production of
planting stock from tissue culture as opposed to seed orchard propagation.
Loblolly pine seed orchards require a minimum of eight to ten years from
grafting until large-scale seed production begins. Hopefully, tissue cultured
propagules from select trees could be mass-produced one or two years following
selection. Therefore, tissue culture technology offers not only the
opportunity to capture greater genetic gains, but also to utilize this genetic
gain earlier than with conventional seed orchard technology (McKeand and
Frampton 1984, Amerson et al. 1985).
These and other (Durzan and Campbell 1974, Mott 1981, Sommer and Brown
1979) benefits of tissue culture have great potential, but the technology
1/assistant Professor, Dept. of Forestry, Professor, Dept. of Botany and
~ Assistant Professor, Dept. of Forestry and Botany, respectively, North
Carolina State University, Raleigh, N. C. The laboratory production of
plant material by the technical support staff of Dr. Henry V. Amerson
is gratefully appreciated.
136
necessary for operational propagation of loblolly pine via tissue culture
is not yet available. Ultimately, embryogenesis or organogenesis from callus
or cell suspensions are desired to facilitiate mechanization of propagation
for mass-production as well as integration with molecular genetic
biotechnology. Currently, organogenic propagation of loblolly pine from
needle fascicles (Mehra-Palta et al. 1977), cytokinin-treated winter dormant
buds (Abo E1-Nil 1982) and cotyledon explants (Mott and Amerson 1981) is
possible. While the technology to propagate from other sources is less
reliable, clonal propagation of loblolly pine from cotyledon explants on a
research scale is routine. The use of cotyledons as starting material offers
less genetic advantage than propagation from older trees of proven genetic
value. However, until more reliable methods of propagation from older trees
are available, studies of the field performance of tissue culture plantlets
from cotyledon origin will be useful in identifying and understanding general
problems associated with tissue culture propagation and facilitate
amelioration of these problems when other propagation systems are employed.
For this reason, the Project on Tissue Culture of the Southern Forest
Research Center at North Carolina State University has established a series of
field plantings containing tissue culture propagules of cotyledon origin.
This paper compares the growth and development of these plantlets with
seedlings after three growing seasons in the field.
METHODS
Laboratory
The propagation system used to establish these studies was reported in
1981 (Mott and Amerson) and involves the timely application and removal (i.e.,
pulsing) of growth regulators to progress from shoot initiation through
rooting. Although this sequence has been and continues to be improved, a
summary of the process follows. Basal media used in this study was BLG (Brown
and Lawrence 1968) with glutamine (10 mM) substituted for NH, and NO3 and 10
mM Kel added, or GD (Gresshoff and Doy) based media diluted by one-half (Mott
and Amerson 1981).
Seeds scarified at the micropylar end were partially germinated in
hydrogen peroxide (typically three days in 1% aqueous H705 followed by one to
two days in 0.03% H»90 at 28-30°C). Subsequent to seed coat removal and
surface sterilization, the embryos were aseptically excised from the female
gametophyte. Next the cotyledons were surgically removed and planted
horizontally on a shoot initiation medium (BLG) which was cytokinin-rich
[typically 444M benzylaminopurine (BAP)]. Cotyledons were maintained on this
medium for 14 to 28 days. On this medium, cell divisions occurred in the
peripheral areas of the cotyledons producing a warty, meristematic surface.
Cotyledons were removed from this medium prior to the actual observance of
shoots and placed on a hormone-free (GD)°1/2) medium containing charcoal to
further aid cytokinin removal. On this medium, shoot apices became
recognizable on the cotyledons and the shoots began to elongate. Shoot growth
continued during further monthly subcultures on hormone-free (BLG) medium.
The multiple shoots crowded on the cotyledons were individually excised and
separated from the cotyledon for further growth.
Following growth, shoots about 1-2 cm in length were transferred to
auxin-rich (GD)°1/2) medium [typically ,-Naphthaleneacetic acid (NAA) at
137
2.5MM]. Shoots were freshly cut at the base, implanted upright and pulsed
for six to nine days on the auxin medium. Pre-root cell divisions formed near
the cambial region at the stem base, resulting in a swollen, callusy region.
To facilitate organization and rapid root growth, the shoots were transferred
to hormone-free (GD ,°1/2) medium. Plantlets typically were ready for
transfer to greenhouse soil three to five weeks after the root initiation
treatment.
Greenhouse
Plantlets were transferred from the agar medium to the greenhouse when
their total shoot lengths (including needles) exceeded 1-2 cm and their
individual root lengths exceeded 3-4 mm. Plantlets meeting these requirements
were carefully removed from the agar medium and planted in 164 cc RL Super
Cells containing a fine textured mix of peat, vermiculite, and perlite -
(2:2:1). The plantlets were grown in a mist bench the first three to six
weeks in the greenhouse. After the first week, they were fertilized three to
five times weekly with Peters 15-30-15 mixed at 30 ppm N. Plantlets in the
mist bench were sprayed weekly with a fungicide, Captan, to reduce damping off
and other disease problems. When necessary, the photoperiod of the plantlets
was extended to 16 hours using incandescent lights (approximately 4 Wm7 2) to
prevent dormancy.
Although the initial growth of the plantlets in the greenhouse was very
slow, after about six weeks, new vigorous growth appeared. At this time,
plantlets were removed from the mist bench and fertilization was changed to
Peters 20-19-18 mixed at 40 ppm N applied three to five times weekly. After
removal from the mist bench, plantlets were watered as needed with pH 5.5
water. Generally, plantlets reached a suitable size for field planting (about
20-30 cm in height and 3-5 mm in caliper) after six months in the greenhouse.
Using similar procedures, seedlings were also grown in the greenhouse to
use in field tests for comparison purposes. Seedlings generally required only
four months to attain plantable size so that it was often necessary to
manipulate the watering and fertilization regime of the plantlets and
seedlings in order to coordinate their growth.
Before field planting, both the plantlets and seedlings were gradually
adapted to conditions outside the greenhouse. The succulent growth was
hardened-off by first stopping fertilization and reducing watering. Next, the
trees were transported outside for two to four weeks in order to adapt to
direct sunlight, natural photoperiod and outdoor temperatures.
Field
The field tests were carefully site-prepared, hand-planted and
intensively managed. At the time of establishment, a soil analysis was
conducted and any nutrient deficiencies were corrected. Additionally,
approximately 50 g N usually in the form of ammonium nitrate was applied to
every tree during the spring to enhance growth. Weeds in the plantings were
controlled either by periodic mowing or with herbicides. Nantucket pine
tipmoth (Rhyacionia frustrana Comst.) which often kills young loblolly pine
shoot tips was controlled with Furadan applications. Spacing was typically
Sig (5) 5% S605) iii
138
The North Carolina State University Project on Tissue Culture has
established 16 field plantings across the Southeast (Figure 1). Over 3000
trees each of plantlets and seedlings have been planted representing over 25
half-sib families of loblolly pine. Results discussed in this paper will be
limited to the first series of eight plantings that were established in 1981
with brief mention of results from two of the plantings established in 1982.
All of the 1981 field plantings contain paired row-plots of plantlets and
seedlings from several half-sib families. The plantlets in a plot represent
one clone produced from the cotyledons of a single embryo. The trees of the
seedling plots were grown from seed of the same half-sib family from which the
plantlets were derived. The 1981 field plantings contained from 16 to 49
plots. Plot size varied from two to 46 trees depending on the number of
plantlets produced in a clone. Planting size varied from 158 to 324 trees.
Several of the field plantings established after 1981 in addition to row-plots
contain clonal block plots of 16 or 25 plantlets compared to block plots of
seedlings from the same half-sib family.
Total height and the incidence of fusiform rust (caused by Cronartium
quercuum (Berk.) Miyabe ex Shirai f. sp. fusiforme (Cumm.) Burds. et Snow)
were recorded annually in all field tests. Additional measurements of
morphological characteristics were also made in some of the plantings.
Figure 1. Location of the North Carolina State University Tissue Culture
Project's loblolly pine field plantings.
139
Data Analysis
Analyses of variance were utilized to determine differences between
plantlets and seedlings at each location. Although the field design
necessitated a different model for some plantings, height measurements at most
locations were analyzed using the following sources of variation: 'plant
type' (plantlets verus seedlings), 'family', ' plant type x family’,
'plot(plant type x family)' and 'tree(plot(plant type x family))'. Incidence
of fusiform rust was analyzed on plot means employing a similar model. Plant
type differences in the analysis of variance were tested using the 'plant
type x family' interaction as the error term. Plant type differences across
all locations were tested by a paired t-test where the difference between
plant type means at each location was weighted by its number of observations.
RESULTS AND DISCUSSION
Height Growth
Average second year survival exceeded 94 percent for both the plantlets
and seedlings in the eight 1981 field studies. Third year height averaged
2.72 and 3.38 m, respectively, for the plantlets and seedlings (Table 1). The
plantlet height averaged 74 to 84 percent of the seedling height and except
for one location was statistically (RES 0.05) shorter than that of the
seedlings.
Table 1.--Mean third year heights of the eight field plantings established
in 1981 by the North Carolina State University Tissue Culture
Project.
Total Height(m)
Year 3
Organization Location EB Ss P/S*100
Federal, PapersBoardlConmmincalimLunberton uN Cini 2= 9 n> InES 5 o> lniannnn7 > amma
Brunswick Pulp LandWCos Siem esup i i GAUn sanmrnnnnnn n=) aca ont? CE a
Westvaco Corp. |... ) mtlns De mnRESummeryidlleNS Cnn 45) nino 05 ame
Scott PaperiCo.) ll) |) LeeMonroe vile ty Alumnen Oo H5/n* 59 OOO
NC.) State Universi type aRalle leh N Cinna Om inn 0 enna 7 an
ITT Rayonter; Incs) #2 ylee lL) pvulleesur ly nz Oon SO nee
Champion uinternationalucor p aimilNewbe rr yas CH NnnnnLE GO mien nnn 7
Crown) Zellerbach iCorpai i lin abo alls ale aU AM mmm SnlnS ot On emn2 720 senna a
Weighted :
Mean Zaha x S35 80
I/p = plantlet, S = seedling, *
Ps -05 level.
significantly different at
140
Figure 2 compares the height growth curves of the plantlets and seedlings
averaged over the eight 1981 field plantings. Within each year, the seedling
height significantly exceeded the plantlet height (McKeand and Frampton 1984,
Amerson et al. 1985). Although every effort was made to establish the
plantlets and seedlings at similar heights and stages of development, the
plantlets averaged only 84 percent of the seedling height at the time of field
planting. The plantlet height fell further behind the seedling height during
the first and second growing seasons. However, after the third growing
season, plantlets had gained to only a net loss of four percent from the time
of establishment.
Figure 2.--Mean height growth curves of tissue culture plantlets and
seedlings in the eight field plantings established in 1981
by the North Carolina State University Tissue Culture Project.
4.0
5
3.0
= P
Fe
ee Ke 20
1.0
YEAR: ‘1981 1982 1983 1984
P/S-100: 84% 18% 16% 80%
\/p = Planclet, S = Seedling
Thus, the difference between the third year height of the plantlets and
seedlings is largely related to differences in initial planting size and a
first year lag in plantlet height growth in the field. In these studies, the
seedlings will most likely remain taller than the plantlets in absolute height
through rotation age. Since plantlet growth rates are similar to that of
seedlings during the third year in the field, cultural treatments which
overcome the plantlets' initial slow growth should yield plantlets of
comparable size as seedlings in future studies. Measurements will continue in
these tests to monitor long-term growth trends.
Fusiform Rust Resistance
After three growing seasons in the field, the plantlets had less fusiform
rust incidence than the seedlings in all eight of the 1981 field plantings
(Table 2). Overall, the plantlets averaged 27.7 percent infection while the
seedlings averaged 47.6 percent, a statistically significant difference
(P S0t 05). In some high hazard regions of the Southeast, such reductions in
fusiform rust incidence would be of great economic benefit and could offset
initial slower growth. The nature of the plantlets’ relative resistance is
not yet understood.
141
Table 2.--Mean third year incidence of fusiform rust in the eight field
plantings established in 1981 by the North Carolina State
University Tissue Culture Project.
Fusiform Rust
Infection (%)
Year 3 uy
Organization Location
12 S
Federal Paper Board Co., Inc. Lumberton, NC 1.8 8.4
Brunswick) Pullpland §Coslln i eSaphiCANN n= TIES O-0NNINs Oman
WestvacolCorpa D2 10 Ln NES Umnier¢ulMlien tS Ch ulnennS Gna nC mo man
Scottmpaper Conn ltt llmmnnEn Monroeuilllonm Al; mnnnNn7 ONOEoan/OM> ina
NC Gare Unholy Ne by oS
ITT Rayonierssines).0n le inneevullee )ohk) Galt heennEs4 Slecn>/aCmmn
Championiinternationallcorp memmNewberry sno Cll inna nl osOmnn
‘Crown iZel lerbachncorpel NallnlNBopalusas LAN NN—NO ON. mmnnGl yaaa
Weighted ;
Mean Mey CSE
1/P = plantlet, S = seedling, * = significantly different at
p < .05 level.
Morphological Characteristics
Many differences between plantlet and seedling shoot morphology have been
observed in the field plantings. Although the plantlets originated from
embryonic material, their morphology appears more mature-like than seedlings.
Older loblolly pine generally has larger needles, fewer branches per unit of
height, fewer growth cycles per season and slower growth rates than younger
material (Greenwood 1984). A detailed measurement of these and other
characteristics (after two growing seasons) at the 1981 field planting in
Jesup, Georgia, has verified quantitatively that the plantlets are expressing
more mature-like morphology than the seedlings (McKeand 1985). Examples from
that study are presented in Table 3.
Additionally, early production of female stobili has been observed on
many plantlet clones. One clone which is represented by 75 plantlets at each
of two of the 1982 field plantings, located at Rincon, Georgia, and
Cantonment, Florida, represents the extreme of this phenomenon. After one
full growing season in the field, 85 percent of the plantlets of this clone
had produced female strobili. Only five percent of the seedling plots from
the same half-sib family had produced female strobili at the Cantonment,
Florida, study while no trees in the seedling plots had produced strobili at
the Rincon, Georgia, study (McKeand 1985).
142
Table 3.--Second year measurements of the loblolly pine tissue
culture planting established in 1981 at Jesup,
Georgia (McKeand 1985).
begubie Plantlets Seedlings
Terminal Bud Length (cm) Be4 Be ili
Terminal Bud Diameter (mm) 6.3 xe Gada
Needle Length (cm) 1S Si NS 18.6
‘Needle Dry Weight (g) 0.16 * O.12
Percent Fusiform Rust Infection 47.4 NS 68.8
Number of Cycles 4.6 * ae
Vv * = significantly different at P 20.105 level.
The cause of this apparent early maturation of loblolly pine tissue
culture plantlets relative to seedlings is unknown.
CONCLUSION
Early results from field trials of loblolly pine tissue culture plantlets
have identified differences in growth, fusiform rust resistance and morphology
between plantlets and seedlings. In an attempt to better understand the
nature of these differences, new studies have been initiated. These include:
(1) exploring alternative treatments for producing shoots and roots in vitro,
(2) investigating the effect of cultural practices such as root pruning on
subsequent plantlet development, (3) establishment of field plantings to
compare seedlings having tissue culture-produced root systems, plantlet shoots
grafted onto seedling roots and seedling shoots grafted onto plantlet roots
and (4) excavation of both plantlet and seedling root systems after several
growing seasons in the field. Forthcoming results from these and other
studies will not only provide knowledge necessary to improve tissue culture
propagation but also provide some insight into the control processes of
development in loblolly pine. i
The use of trade names throughout this paper does not imply endorsement
of these products, nor criticism of products not named.
143
LITERATURE CITED
Abo E1-Nil, M. M. 1982. Method for asexual reproduction of coniferous
trees. U. S. Patent #4,353,184.
Amerson, H. V., L. J. Frampton, Jr., S. E. McKeand, R. L. Mott, and R. J.
Weir. 1985. Loblolly pine tissue culture: Laboratory greenhouse and
field studies. In: Tissue Culture in Forestry and Agriculture. Eds.
R. R. Henke, K. W. Hughes, M. J. Constantine, and A. Hollaender. Plenum
Press. Univ. Tenn. Symp. Poc. Sept. 9-13, 1984. pp. 271-287.
Brown, C. L., and R. H. Lawrence. 1968. Culture of pine callus on a defined
medium. For. Sci. 14:62-64.
Durzan, D. J. and R. A. Campbell. 1974. Prospects for the mass production
of improved stock of forest trees by cell and tissue culture. Can. J.
For. Res. 4:151-174.
Greenwood, M. S. 1984. Phase change in loblolly pine: shoot development as
a function of age. Physiol. Plant. 61:518-522.
McKeand, S. E. 1981. Loblolly pine tissue culture: Present and future
uses in southern forestry. School of For. Res. Tech. Rep. No. 64,
| NopGo State nUniv... 950 %p).
McKeand, S. E. 1985. Expression of mature characteristics by tissue culture
plantlets derived from embryos of loblolly pine. J. Am. Soc. Hort.
Sci. (in press).
McKeand, S..E. and L. J. Frampton, Jr. 1984. Performance of tissue culture
plantlets of loblolly pine in vivo. p. 82-91. In: Proc. Intl. Symp.
of Recent Adv. in For. Biotechnology. Traverse City, MI. 194 p.
McKeand, S. E., G. S. Foster and F. E. Bridgwater. 1985. Breeding systems
for pedigree-controlled production populations of loblolly pine. Proc.
S-23 Workshop on Adv. Generation Breeding & Current Status and Research
Needs. 1984. Baton Rouge, LA. (in press).
Mehra-Palta,-A., R. H. Smeltzer and R. L. Mott. 1977. Hormonal control of
induced organogenesis: experiments with excised plant parts of loblolly
pine. Tappi 61(1)37-40.
Mote “iRewbeed 198 loa Crees. In: Cloning Agricultural Plants Via In Vitro
Techniques. B. V. Conger ed. CRC Press. Boca Raton, Fla. pp. 217-254.
Mott, R. L. and H. V. Amerson. (1981. A tissue culture process for the clonal
production of loblolly pine plantlets. North Carolina Ag. Res. Ser.
Tech. Bul. #271. 14 p.
Sommer, H. E. and C. L. Brown. 1979. Applications of tissue culture to
forest tree improvement. p. 461-491. In: Plant cell and tissue culture
principles and applications (Sharp, W. R., P. O. Larsen and V. Raghaven,
Eds.) Ohio St. Univ. Press, Columbus. 892 p.
144
PARENT VS. OFFSPRING SELECTION: A CASE STUDY
Gary R. Hodge! /
Abstract.--A comparison was made between genetic gains and
program benefits to be expected from parent selection
after progeny testing, and offspring selection of the best
trees in the best families, given the constraints of the
N. C. State Industry Tree Improvement Cooperative
disconnected half diallel mating scheme. At moderate
heritability levels (h2 =-0.15), offspring selection
yielded higher expected genetic gains. However orchards
established with juvenile scion material from genetic
tests may not reach commercial production levels as
rapidly as orchards established with scion material from
older parent trees. Economic analysis shows that a delay
of-only one year in reaching full production levels can
make offspring selection less profitable than parent
selection, i.e., the cost of delay exceeds the increased
return of higher genetic gains.
INTRODUCTION
Parent selection is a selection strategy where individuals are selected
on the basis of the performance of their progenies. The best individuals (the
parents of the best progeny families) form the commercial production
population. Offspring selection is a strategy where the selected individuals
are the best members of the progeny families. The main advantage of offspring
selection is that it offers breeders the opportunity to do within family
selection at high selection intensities. This generally allows the
achievement of higher expected genetic gains, and thus is quite attractive to
breeders. The primary objective of breeding programs, however, is not to
achieve maximum genetic gains, rather it is to generate maximum dollar value
or economic return. In making a decision about which selection strategy to
use, it is important to consider factors other than those which maximize
genetic gain. A case study of the situation involving the N. C. State
Industry Tree Improvement Cooperative illustrates this point.
At this time, the N. C. State Cooperative plans to use offspring
selection to select its third generation population of loblolly pine (Pinus
taeda L.) (Anon., 1983). There has been some concern, however, that seed
orchards established after offspring selection may not develop strobili as
quickly as orchards established after parent selection. This possibility
arises because the two selection strategies will yield scion material of quite
different ages and in different states of sexual maturity. Loblolly pine
grown in the field does not begin to flower consistently until it is 10 to 15
years-old (Dorman and Zobel, 1973). Since the N. C. State Cooperative plans
on making selections in its genetic tests at age eight (McKeand and Weir,
1983), scion material from offspring selection will be sexually immature.
Scion material from parent selection would be sexually mature, and the two
T7Research Assistant, N. C. State University, Raleigh, North Carolina.
145
types of orchards could conceivably show different patterns of strobili
production over time. The objective of this case study is to compare genetic
gains expected from parent and offspring selection, and to examine the effect
of production delays on expected program benefits.
ESTIMATION OF GENETIC GAINS
The following assumptions were made in calculating genetic gains:
a. The seed orchard will be composed of 24 unrelated individuals.
b. The foundation population consists of 480 unrelated individuals.
c. The foundation population will be mated using six parent disconnected
half-diallels formed at random with respect to general combining
ability.
d. Non additive genetic variance equals additive genetic variance.
e. Field trials of the diallels will be conducted according to the
N. C. State Cooperative Genetic Testing Manual (Talbert et al.,
1981).
The specific calculations for genetic gains for both parent and offspring
selection are presented in the Appendix. Calculations for genetic gains from
offspring selection were made using separate formulae for family and within
family selection, following a technique outlined by Squillace (1973).
Estimates of genetic gains obtained from this technique are conservative, as
combined selection would give greater genetic gains (Falconer, 1981).
However, in calculating genetic gains from combined selection a priori, one
cannot account for the requirement to maintain unrelatedness. |
Expected genetic gains (Table 1) are presented in terms of phenotypic
standard deviations of individuals (9,), and eusts was assumed to be equal for
both parent and offspring populations.
Results of Genetic Gain Calculations
As expected, at most heritabilities offspring selection yielded higher
expected genetic gains than parent selection. At very low heritabilities
(individual, narrow sense), parent selection was more efficient, although a
change occurred between h* = 0.10 and h2 = 0.15. One should note that this
was primarily due to the effect of within family selection. As heritability
increased from 0.05 to 0.50, gains from half-sib, full-sib, and parent
selection increased approximately three-fold. Over the same range, gains from
within family selection increased ten-fold.
If the decision between the two selection strategies were simply a
question of maximizing genetic gains, one need only determine the heritability
of the trait of interest to make the correct decision. As the program
objective is to maximize economic return, however, other factors must be
considered.
146
Table 1.--Expected genetic gains®
Selection
Half Full Within
h2 sib sib Family Of fspring? Parent
3(0)5) APA) -0935 -0707 - 3839 ©4379
10 SSO Sia ~1442 5986 -6354
lis) 23903 elivalol 2209 7823 e758
o20 DIL 1991 - 3009 «9D22 -9107
5S) 005 2230 - 3847 1.1148 1.0209
30 -5556 - 2456 BOTA) 1.2739 1.1204
-40 -6426 2847 6632 ~ 1.5905 1.2967
~50 SUN Lol 8773 1.9156 1.4517
4values expressed in units of phenotypic standard deviation of individuals.
bgain from offspring selection equals the sum of gains from half-sib, full-
sib, and within family selection.
SEED ORCHARD DEVELOPMENT
Grafted seed orchards (using sexually mature scion material) have
generally reached commercial production levels 10 to 15 years after orchard
establishment (Anon., 1979), and a rule of thumb developed in the N. C. State
Cooperative is that 8 to 12 years usually elapse before meaningful production
occurs (Talbert et al., 1983). Good choice of seed orchard sites, along with
intensive irrigation and fertilization can promote the development of young
seed orchards (Jett, 1983). The possibility that seed orchards established
with sexually mature material may come into production sooner than orchards
established with juvenile material forces one to consider the economic costs
of a delay in production. To do this, one must know when the returns from the
increased genetic gains will be available, i.e. when the improved trees will
be harvested.
TIME LINES
The following assumptions were used in developing time lines scheduling
the harvest of improved trees:
a. Two years to establish orchard after selection.
b. Breeding and testing for the subsequent generation begins immediately
and will be completed in 14 years. A new orchard will then be
established.
c. Two years after first commercial cone harvest until actual planting of
the improved seedlings.
d. 25 years rotation.
e. Seed orchards go from zero to full strobili production in a single year.
Assumption e. is unrealistic and was made only to simplify analysis.
Using these assumptions, one can generate a timeline for an orchard
established after parent selection, assuming eight years to reach full
production (Pg):
147
Event Year
Year to finish orchard establishment 22
Year to first commercial production 10
Year to first planting 12.
Earliest that next generation orchard could be established 16
Time when next generation orchard would produce 24
Last year planting #45)
First year next generation material planted 26
Years of harvest Sf 20) 50
A similar time line can be developed for offspring orchards. Comparisons
were made between the Pg situation above and offspring orchards taking 9,
10, and 11 years to reach full production (09, Ojo, and 07], respectively).
One should note that in the Pg situation, harvests are made from year 37
to year 50. For all offspring orchard situations, the next generation orchard
is assumed to take eight years to develop. This would be the case if
offspring selection was utilized, and then followed by parent selection for
the next cycle of improvement. Thus for the 0g situation, the years of
harvest are years 38 to 50, and it is the harvests over these years that are
compared to the Pg Situation.
DISCOUNTING PROCEDURES
In order to compare the economic values of the two selection schemes, one
needs to convert the genetic gains calculated earlier into dollar values. The
information needed is the mean of trait p, the standard deviation of trait p,
and the relationship of trait p to dollar value. At the time that the
selections would be made, this information would be in hand. For the sake of
comparison in this general analysis, however, one can make the assumption that
trait p is linearly related to dollar value. This is a reasonable assumption
for volume growth with a product objective of pulpwood. It then becomes
possible to make relative comparisons between the two selection schemes,
simply treating genetic gain as if it were dollar value. Another assumption
is that the amount of land planted and harvested is equal each year and from
year to year. For example, the organization may be planting and harvesting
10,000 acres every year. The organization would then receive an annual
annuity over the years of harvest outlined above.
One can calculate the present value of a terminating annual annuity with
the formula:
(14+1)"8 - 1
VOC) ain aa (Lundgren, 1971)
where: a = value of annuity
i = interest rate
n = year the annuity terminates
The present value of an annual annuity to be received between the year n and
' year x is:
Vo(n to x) = Yo(x) ~ Yo(n-1)
148
including heritability, interest rate, economic value function, and generation
interval. The effects of heritability and interest rate have already been
discussed. Economic value function can also be important. In this study, an
assumption was made that the trait of interest is linearly related to dollar
value. In fact, the economic value function could be some type of stepwise
function where an increase of the trait beyond a certain point yields a very
large increase in dollar value. If the trait has this sort of relationship to
dollar value, the discounted genetic gain estimates for offspring selection
may be low relative to parent selection. The more the population mean is
increased, the greater the probability the population will reach the next step
in dollar value.
Generation interval is also important. In this study, it was assumed
that six years would be necessary to complete the matings, and eight years to
complete the field testing, for a generation interval of 14 years. If
selections were made at six years instead of eight, the generation interval
would be 12 years. Under this circumstance, the relative cost of missing the
first year of production is greater than with a 14 year generation interval.
Therefore, if a breeder expected orchards established with offspring scion
material to be slower in reaching full production than orchards of parent
scion material, a 12 year generation would tend to push him even more in the
direction of parent selection.
For this specific case study, the primary question becomes “Will there be
a difference in the development of seed orchards established with sexually
mature and immature scion material?” Although there is very little in the
‘literature on this subject, I suspect that there would be little differnce.
Consider that if selection occurs at eight years, it takes two years to
establish the orchard, and a minimum of six to eight years are necessary for
the trees to have enough vegetative growth to support full production levels,
offspring scion material would be 16 to 18 years-old, and would probably be
sexually mature. More concrete evidence is presented by Talbert et al.
(1982). In a study of four seed orchards, although sexually immature grafts
tended to produce more pollen catkins and less female strobili than mature
grafts, the differences were not statistically significant.
Other Time Factors
A difference in the rate of seed orchard development is not the only way
that a time difference could have an impact on the decision between parent and
offspring selection. Another source of difference might be the time required
for orchard establishment. It may take more time to establish offspring seed
orchards from single eight year-old trees than parent orchards from numerous
ramets kept in a greenhouse or clone bank.
Breeders should also consider that it is generally possible to identify
the best families in field tests earlier than it is possible to identify the
best individuals in those families. In the case of the N. C. State
Cooperative, it is likely that one could be nearly as effective in parent
selection at age four or six as at age eight. To identify the best
individuals in offspring selection, however, is very difficult at younger
ages. One could then argue that a breeder is imposing a two year delay on his
program in order to gain the additional benefit of within family selection.
-The results of this study would suggest that this is not worthwhile.
149
One can then discount the genetic gains presented in Table 1 by
multiplying by the appropriate value for Vo(n to x)° This allows a comparison
of the two selection schemes taking the delay in reaching full production into
account.
Discounted Genetic Gains
Genetic gains were discounted at interest rates of 6% and 9% (Tables 2,
3). Discounted genetic gains were calculated at h2 = 0.15. Solely on the
basis of genetic gains, offspring selection was more efficient at all
heritabilities in Tables 2 and 3. But comparing the Pg and Og situations at
6% interest, the cost of a one year delay in reaching full production levels
was enough to make parent selection more valuable than offspring selection at
heritabilities up to 0.25. Not unless h2 was as high as 0.30, did the
increased genetic gains from offspring selection offset the cost of missing
the first year of production.
Longer delays had higher costs. Comparing the Pg and the 0);9 Situations
in Table 2, a two year delay at 6% interest, only at a heritability of 0.50
was offspring selection more valuable than parent selection. For a three year
delay, Pg and 0;], parent selection was always more valuable.
The effect of higher interest rates was to place a higher premium on
reaching production earlier. At 9% interest with a one year delay, offspring
selection became more valuable than parent selection at h2 = 0.40, as opposed
to a h2 = 0.30 with 6% interest.
h? i pez 09 010 O11
15 8649 8019 7165 6349
20 1.0390 -9761 8721 7739
25 1.1647 1.1428 1.0210 9061
30 1.2783 1.3059 1.1668 1.0354
40 1.4794 1.6304 1.4567 1.2928
50 1.6562 1.9637 1.7545 1.5570
Table 3.--Discounted genetic gains at an interest rate of 0.09.
——— eee
h? Pg Og 10 011
SNS .2340 .2119 -1848 .1598
.20 .2811 . 2580 2249 1945
25 Piss. . 3020 2633 .2278
130 : ~3459 3451 . 3009 .2603
40 .4003 -4309 SUS 3249
50 8 44g 5189 GEES ney 3914
DISCUSSION
The most striking result of this study was that only a one year delay had
.a significant impact on the choice between parent and offspring selection. In
making this decision, tree breeders must consider a number of factors
150
CONCLUSIONS
In making a decision between parent and offspring selection, tree
breeders should consider when genetic gains will be available, in addition to
the size of those genetic gains. Even a delay of as little as one year can
change which selection scheme yields the highest overall benefit to the
program.
LITERATURE CITED
Anonymous. 1979. Twenty-third Annual Report of the North Carolina State
Industry Cooperative Tree Improvement Program. 55 p.
Anonymous. 1983. Twenty-Seventh Annual Report of the North Carolina State
Industry Cooperative Tree Improvement Program. 66 p.
Dorman, K. and B. Zobel. 1973. Genetics of Loblolly Pine. USDA Forest
Service Research Paper WO-19, 21 p.
Falconer, D. 1981. Introduction to Quantitative Genetics 2nd edition.
_ Longman Inc., New York. 340 p.
Jett, J. 1983. The impact of irrigation and supplemental nitrogen
fertilization on the development of a young loblolly pine seed orchard.
Ph.D. dissertation, Dept. of For., N. C. State Univ. 38 p.
Lundgren, A. 1971. Tables of Compound-Discount Interest Rate Multipliers
for Evaluating Forestry Investments. USDA Forest Service Research
Lush, J. 1943. Animal Breeding Plans. JIowa State College Press. Ames,
Iowa. 437 p.
McKeand, S. and R. Weir. 1983. Economic Benefits of an Aggressive Breeding
Program. Proceedings of the 17th Southern Forest Tree Improvement
Conference. Athens, Georgia. June 6-9, 1983.
Squillace, A. 1973. Comparison of some alternative second generation
breeding plans for slash pine. Proceedings of 12 Southern Forest Tree
Improvement Conference. Baton Rouge, Louisiana.
Talbert, J., F. Bridgwater and C. Lambeth. 1981. Genetic Testing Manual.
N. C. State University-Industry Cooperative Tree Improvement Progran,
Sch. of For. Res., N. C. State Univ. 37 p.
Talbert, J., R. Weir and R. Arnold. 1983. Loblolly pine tree improvement:
an attractive forestry investment. Proceedings of the Southern Forest
Economics Workshop. Mobile, Alabama. April 7-8, 1983.
Talbert, J., R. Wilson and R. Weir. 1982. Utility of First Generation
Pollen Parents in Young Second Generation Loblolly Pine Seed Orchards.
Proceedings of the 7th North American Forest Biology Workshop.
Lexington, Kentucky. July 26-28, 1982.
151
Appendix.--Genetic Gain Calculations.
A. Conditions
480 P) Selections, divided into 80 groups of 6 each.
Selections mated to prduce 5 full-sib families within each half-sib
family.
Field test design involves (4 locations) (6 reps/loc.)
(6 trees/family/rep) yielding 144 trees/cross.
n¢ = 144 = number of full-sib family members
ny, = 720 = number of half-sib family members
B. Offspring Selection Scheme
1. All half-sib and full-sib families are ranked.
2. The highest ranking half-sib family is identified (Family A).
3. The best of the five full-sib families involving Family A is
identified (Family AxB).
4. The best individual tree in the full-sib family AxB is identified
to be grafted into the seed orchard.
5. Half-sib families and full-sib families involving A or B are
eliminated from the list of candidate families.
6. Return to Step 2.
C. Genetic Gain From Parent Selection
Rp = ioph? ----------- = 2.063 oyh?
4 + (720-1)h2
Rp = expected response from progeny testing (parent selection)
i = selection intensity expressed in standard measure
Op = phenotypic standard deviation of individuals
h2 = individual tree heritability
n = number of progeny = 720
Selection of parents of the top 24 half-sib families = 24/480 yielding
i = 2.063.
D. Genetic gain from offspring selection - adapted from Squillace (1973).
Ro = Rus + Res + Ryr
where:
Ro = expected response from offspring selection
Rus = expected response from half-sib family selection
Rrg = expected response from full-sib family selection
Ryr = expected response from within family selection
152
1. Half-sib selection
7 1 + (np-l) ren an 1 + (720-1) (.2987)
“Ring © iG 0) | Sees 2.013 o,h? --------~------------------
(Rai GES pen) = J 720[1 + (720-1) (.3495h2) ]
where:
variables as before —
ny, = number in half-sib family
Yon = effective coefficient of relationship between members
. of half-sib families (Lush, 1943)
Eon bh Z999N—. OO17
Coli smissare oer =f ieaianes crc. = .2987
1 - Tph_ T0017
‘Twh = average coefficient of relationship within half-sib families
jaya Gays cae Ie 720) Copck el) in=2
Ce See Se ee) ee oe 2999
Oe Nay 3b) ee G20): 65) —
fy = ‘number of full-sib families per half-sib family
ph = average coefficient of relationship between half-sib families
3p - 2 C)K6) PS 2
SS 6 SSS SS = .0017
4(N-1) (p-1) 4(480-1) (6-1)
_where:
N = total number of parents = 480
p = number of parents per diallel group = 6
cher = phenotypic correlation between trees in half-sib families = .3495 h2
If all genetic variance was additive, t could be calculated for a half-sib
family by multiplying the n2 by the average coefficient of relationship:
t =h*r. In this case, however, non-additive variance equals additive
_ variance, and may contribute to phenotypic correlation as the half-sib
families are not truly half-sibs, but have full-sibs in them. Full-sib
families have a covariance of 1/2V, + 1/4Vq- If all non-additive variance is
dominance, one can calculate t for full- ~sibs as tr = (1/2V, N/A yh2,
Thus t¢ = 0.75 h2, One can then determine th for half-sib Families by
weighting ty, and t¢ by the probability of selecting two half-sibs and two
full-sibs, respectively, when randomly choosing two members of a half-sib
family.
t, = Pr(FS) + tg + PR(HS)ty
th = 143/719 (. 2) + [(719-143)/719](.25h2) = .3495h2
153
Selection Intensity: We are selecting 24 half-sib families, but our
selection scheme eliminates two half-sib families each time through the cycle.
We may not be able to actually select the top 24 half-sib families. It is
possible to determine, however, the probability that a diallel contains zero
of the top 27 half-sib families:
(480-27) 480
Pr(diallel has zero of top 27) =C 6 AG 6
(453 x 452 x 451 x 450 x 449 x 448)/(480 x 479 x 478 x 477 x 476 x 475)
0.7052 :
0.7052 x 80 diallels = 56 diallels that contain none of the top 27 families;
therefore 24 different diallels contain the top 27. We can then be confident
of selecting the top 27 half-sib families from 480 yielding i = 2.013.
2. Full-sib selection
; 1 + (ng-l)te¢ , 1 + (144-1) .333
Rpg = i¢ph@ | =es-e === = 1.1630,h2 _ -------------------=
(all Ga Der T 144{1 + (l44=1)(.75n2)]
where:
variables as before
lef = effective coefficient of relationship between members of full-sib
families
Gere e Oar — 0225
= Se = 3s — = 0.333 |
l-rp¢ LR = O25 |
Yy¢ = coefficient of relationship within full-sib families = 0.5
= coefficient of relationship between full-sib families = 0.25 |
phenotypic correlation between members of full-sib families = 0.75h2
rR
ho
rh
"oo
Selection intensity is 1 of 5 yielding i = 1.163.
3. Within family selection
2.784 oph2(0.5)
= j 2 =
Ryr = igph2 (1-0.5)
144(1 - .75h2)
where: i
_ variables as before :
2.784.
Selection intensity is 1 of 144 yielding i
154
GENETIC VARIATION IN LOBLOLLY PINE
ROOT GROWTH POTENTIAL
L.E. DeWald, P.P. Feret, and R.E. Kreh?
Abstract. --Half-sib families of 1-0 loblolly pine seedlings were
evaluated for genetic variation in their ability to regenerate new
roots (root growth potential (RGP)) when outplanted. Half-sib
family variation was significant with heritabilities ranging
0.34-0.37 for two independent samples of seedlots lifted in March
1983 and 1984. Variation patterns of root growth potential were
different for half-sib families lifted at different times during
-the nursery growing season, and RGP of March lifted half-sib
families was related to both first- and second-year field
performance.
Additional k ywords: RGP, field performance, height growth,
nursery management, lifting season, Pinus taeda.
| - INTRODUCTION
Loblolly pine (Pinus taeda L.) is typically regenerated in the southern
UNS. by planting bare-root 1-0 seedlings. Although nearly one billion
loblolly pine seedlings are planted annually, (Johnson et al. 1983), highly
variable transplanting success has resulted in decreased survival rates since
1960 (Weaver et al. 1981).
Analysis of the ability of bare-rovt seedlings to regenerate new roots
(root. growth potential) has improved the understanding of plantation
establishment failures of western North American conifers (Ritchie and Dunlap
1980, Jenkinson and Nelson 1978). A direct relationship between root growth
potential (RGP), and field survival and height growth has been established for
many western conifer species. Among the factors found to affect western
conifer RGP are genetic origin, nursery practices, and handling procedures
(Ritchie and Dunlap. 1980). Jenkinson (1980) determined that one of the keys
to successful plantation establishment of western yellow pines is knowing when
to lift seedlings . from the nursery. Similar results have been obtained for
Douglas-fir (Pseudotsuga menziesii (Mirb. ) Franco) (Jenkinson 1984, Jenkinson
and Nelson 1983) and for other western species (Stone and Norberg 1979).
_Subsequently, optimum lifting periods or "windows" have been established for
different species which are based on genetic background and climatic data.
Stimulated by the success of RGP research in the west, and the studies in
the southeast which indicate a relation between loblolly pine RGP and field
performance (Feret and Kreh 1985, Feret et. al 1985) a project was undertaken
to examine the role of genetic variation in RGP among half-sib families of
loblolly pine seedlings. The specific objectives were to determine if root
growth potential is affected by genetic orgin; to -describe the relationship
between root growth potential, genetic origin, and field performance; and to
1 Graduate Research Assistant, Associate Professor, and Research Associate,
respectively, Dept. of Forestry, VPI & SU, Blacksburg, VA 24061.
155
determine if there is a lift date by genetic origin interaction.
MATERIALS AND METHODS
Two separate studies were established to examine genetic variation in
loblolly pine RGP. The first study, conducted in 1982-1983, measured the
genetic variation in RGP among 15 half-sib families and determined the
relationship between RGP, genetic origin, and field performance. The second
study, which measured genetic variation in RGP over the nursery lifting
season, was conducted in 1983-1984 using 14 half-sib families completely
independent of those used in the 1982-1983 study.
The 1982-1983 study used seedlings grown from seed donated by the N.C.
State Cooperative Tree Improvement Program, and a Virginia Division of
Forestry (VDF) nursery mix. The seedlings were grown in four replicate
nursery beds at the VDF nursery in Providence Forge, Va. using standard
nursery practices, except they were not undercut or top pruned.
On March 15, 1983, 90 seedlings per half-sib family (seedlot) were hand-
lifted; 30 seedlings per seedlot were evaluated for RGP and 60 outplanted on
the Virginia Piedmont Plateau, in Patrick County, Va. The outplanted
seedlings were randomly assigned to 4 five-tree plots per seedlot within each
of three blocks using a 0.5 by 1.0 m_ spacing. The outplanting site was a
southwest facing slope originally containing a mixed pine-hardwood stand that
was clearcut, followed by a chop and burn site preparation. The soil was a
Hayesville fine sandy loam (SCS 1973). Seedling height and survival were
‘measured at the start of the growing season and again in December of 1983 and
1984 when growth had ceased.
Root growth potential was measured by root pruning the seedlings to 12 cm
below their root collars and then planting 15 seedlings per seedlot into two
acrylic trays (46 x 10x 40 cm) containing Promix Bx®, a commercial growth
media containing peatmoss, pearlite and nutrient additives. The trays were
watered to field capacity, sealed and placed in a waterbath at 20 C for 24
days under a 16 hour photoperiod in a greenhouse maintained above 15 C at
night and below 24 C during the day.
After 24 days, seedlings were excavated from the trays and the number of
new reots greater 0.5 cm (easily distinguished from old roots by their white
color) were removed and counted. Root growth potential was expressed as the
number of new roots produced by each seedling. Each .seedling was also
characterized at harvest by measuring root collar diameter, and shoot and root
dry weights (dried at 70 C to constant weight).
The 1983-1984 study used 14 half-sib families donated by the VDF plus a
VDF nursery mix. The seedlings were grown operationally from seed at the same
VDF nursery as the previous study in 8 replicate nursery blocks. - They were
top pruned but not undercut. Severe 1983 spring storms caused poor survival
in the nursery and, consequently, only four lift dates were possible. On
October 25, November 22, 1983 and on February 2 and March 13, 1984 two
randomly selected nursery field replicates were handlifted, common seedlots
from both replicates were combined, and the RGP of 28 seedlings per seedlot
measured.
156
A simpler less expensive RGP testing system, which gives similar relative
results to the soil system used in the 1982-1983 study (DeWald et al. 1985),
was used in the 1983-1984 study. In this system, 4-seedling plots per seedlot
were grown hydroponically in a greenhouse under a 16-hour photoperiod in 7
replicate By Sul stericish aquariums for 15 days. The seedlings were suspended
in aerated tap-water by inserting them at their root collars in floating
sytrofoam blocks. The water temperature was maintained at ambient air
temperature (minimum of 16 C nights and up to 27 C days) and 0.5 g of 20-20-20
(nitrogen-phosphorus-potassium) was added to the water (approximately 13 ppm
final concentration). After the 15 days of hydroponic growth the RGP of the
seedlings was quantified in the same manner as the 1982-1983 study.
Root growth potential and field performance (expressed as total height,
height increment and survival) were analyzed using analyses of variance.
Seedlot means were separated with Duncan's multiple range tests. Heritability
of root growth potential was calculated (Falconer 1983) using data from the
half-sib families (excluding the VDF nursery mix). Regression analysis was
-used to elucidate the relationship between RGP and field performance of the
different seedlots. Since the 1982-1983 seedlings were not top pruned
regression analyses were conducted using annual height increments (obtained by
subtraction).
ree anile RESULTS
1982-1983 Study
The number of new roots averaged over all seedlots ranged from 14.6 to
23.5 with a mean of 17.9 (standard deviation = 3.4), and seedlot variation was
significant (a=0.01). The heritability estimate for RGP was. 0.34 (+0.12).
First-year height increment averaged over seedlots ranged from 14.9 to 22.2 cm
with an overall mean of 17.1 cm (standard deviation = 3.38). Second-year
height increment averaged over seedlots ranged from 43.4 to 60.7 cm with a
‘mean of 50.4 (standard deviation = 5.1). Survival after the first growing
season was 95.9 percent and remained unchanged after two years. The
regression of field performance on number of new roots was significant for
both first- and second-year height increment (%=0.01), with R? values of 0.59
(standard error = 1.86 cm) and. 0.28 (standard error = 4.46 cm), respectively.
The relationships of RGP and height increment are illustrated in Figures 1 and
z. ; :
Regression analyses of mean seedlot shoot dry weight, root dry weight,
root-shoot ratio (based on dry weights), and root collar diamter versus mean
number of ._new roots revealed that only root-shoot ratio was significantly
(%=0.05), although weakly, related (r=0. 25).
1983-1984 Study
Root growth potential varied significantly (2=0.01) among seedlots and
among lift dates. In general, RGP and heritability were lowest for October
and highest for the February lift. There was little RGP variation among
seedlots for these two lifts, with only one seedlot differing significantly
(a=0.05) from the rest. RGP variation among seedlots for the November and
March lift dates was similar and greater than in October and February. The
heritability estimates for the March and November lift dates were also
157
24
© a
il
=
5 :
i
et
Onn
aw 15
a
Sh
>
ane s) Y = 6.87 + 0.58 X
= R2 = 0.59 se = 1.68
ll 14 VW; 20 23 26
NumBeR oF New Roots
Figure 1. Relationship of first-year field performance and root
growth potential of loblolly pine half-sib families
lifted and outplanted on the Virginia Piedmont Plateau,
Patrick County, Va.5 in) March 1983.
SEconND-YEAR HEIGHT INCREMENT (cM)
12 14 16 18 20 De. 24
NumBer oF New Roots
Figure 2. Relationship of second-year field performance and root
growth potential of loblolly pine half-sib families
lifted and outplanted on the Virginia Piedmont Plateau,
Wyck Gower, Weg shay iene cla Abs
158
This
This
well as a complete reversal type
SE
Heritability
h2
0.15 +40.03
0.30 +0.04
0.50 +0.05
0.37" +0)..05
The specific direction
seedlot.
for each lift are summarized in
Range
Zoll=9.29
as
(Seedlot Means)
My
Number of New Roots
0.44 0.14-1.14
8.97 6.54-18.36
8.32 4.39-13.42
6.
In general, the October RGP was always the lowest and either
Mean
the February or March lift dates had the highest RGP.
Significance
a = 0.05
a = 0.01
a = 0.01
a = 0.01
Summary of half-sib loblolly pine seedling root growth potential
Seedlot Variation
in RGP from date to date varied depending on the
performance over the 1983-1984 nursery lifting season.
The RGP results and heritabilities
The seedlot by lift date interaction was also significant (a@=0.01).
relationship is illustrated in Figure 3.
interaction was both a relational (rate)
Time
Oaelinsfate
November
February
October
March
of interaction.
of change
Tabilkey 1c;
samilar:
Table 1.
+ ZENG
|
OPAL ACLEL LL Lh REL
1 TE 40. A
GAA ALARA A
oO
a LA (RL ALLL RL CLL EL EA
o DAA DRA 9 Zz Sc
NSE SSNS SSIS SS S
CELLALLAL ELL LE Loh Leah nA hhh hk
a AOA SAAS RAS c
OLLLZLPLLL LL LLL LEE TATA EIA
EEE SESE
OBA AAAS AAAS LoSsm
=)
ULZZZZLLEL LL ALL LAA
| TAT ARTE Gs Se TAS fa)
a CER RRAERERERS SI vSH
QPPIATPZZL LLL
SSS SSeS 9
SURE EEE EES oso
°
OQOZZZZZZZZLL LLL
( DBE EEEEDOBARUEAEY
2 AAAAAABLALA LLS4
[s)
SU72ZZ7Z7i LaLa Aaa eel eadeaaaglariged
( | RBBB BBPeBsBBwBBWseBwoiBBaBeBaewewVwa 9
Fe) La OR RS A v G nN
oO
OQOQTZZZTILZALLL LL LRA
Cj PRADA EDA DWESLeABEDBASBABWSVBWeBeaecag 6
5) POLLEN ESE SNR BENS LSS STEIN css
Oo
“PZT ZZZZ LL LLL aaa aaa dhaaag
OESSSEASEASS SEES SERENE
a BEE IRIN ¢ c S uy
Oo
UW OPTZ77 ZL LALA AAA
( SREB 7
2 ABAARAADN LLG
O°
OPT ILI Z ZL LLL ZL LALLA LL
a SEEN ENN SLSANEN SS L
Te)
WAS iLeL LLL LA AMAL AA
i / RBDWABDABBBBVweVBi
2 SLR ESAS ENS 8 OS H
Oo
OVZZIZZLL LLL LLL LLL
(:] De OAS AAS
ol BALLS AA AAS S LY La
5
UP ZIZI LL ALAA LN
\] ASG EDS ASAT Gv L
Te) ae Vv
Oo
oO wo oO
—
S]00y MeN FO TequNN
20
15
0.05).
Means followed by the same letter
159
Seedlot
Loblolly pine seedlot root growth potential performance over the
within each seedlot do not differ significantly (a
1983-1984 lifting season.
Figure 3.
DISCUSSION
Results from both years indicate strong genetic control for loblolly pine
RGP and selection would probably be successful for improving this trait. The
heritabilities for March lifted seedlings are similar for both years even
though two completely independent groups of half-sib families were used. The
different heritabilities for the four lift dates indicate that different
degrees of environmental control exist over the growing season, but this
difference is in part due to the lift date by seedlot interaction.
The interaction between time of lift and genetic origin suggests that
different optimum lifting and planting dates exist for different loblolly pine
seedlots. It appears that lifting "windows", such as those described for
western species (Jenkinson (1980, 1984), also exist for loblolly pine.
Field performance, genetic origin, and RGP appear to be closely linked
and the results indicate that seedlots with superior RGP will have greater
height increment the first growing season, with this effect lasting at least
two years. . The relationship between field performance and RGP is consistent
with other loblolly pine studies where treatments with low RGP had poor field
performance (Feret and Kreh 1985, Feret et al. 1985). When interpreting the
field performance data of individual seedlots, it must be kept in mind that
some of the RGP and subsequent height differences may have been due to March
not being an optimum time for lifting some seedlots.
The differences in RGP responses over the lifting season suggest that
family-block plantings in the nursery have an advantage over composite seedlot
plantings since time of lift could be scheduled to maximize RGP for groups of
families that respond similarly. In the western U.S. nurseries that utilize
family-block plantings can use lifting windows ensuring appropriate lifting
schedules for each seedlot, thereby increasing the probability of plantation
success (Jenkinson 1984).
In addition to lifting schedules, seedlot blocks in the nursery also
allow cultural practices to be tailored to maximize RGP. For example, certain
cultural treatments such as’ fertilization might result in increased RGP of
some seedlots at a time when their RGP would otherwise be low. These seedlots
could then be lifted at what would have been a non-optimal time.
The genetic control of RGP permits screening of seedlots for an
indication of potential relative field performance. This can be particularly
useful for the selection of seedlots for difficult sites or difficult years.
In addition, the lift date by seedlot RGP interaction may be a partial cause
of obscure genotype by environment interaction in progeny tests.
The differential RGP response among seedlots over the lifting season
should be considered in the establishment of progeny tests. Since all
seedlots are not at their optimal RGP for lifting simultaneously, progeny
tests should perhaps be established over a period of weeks to minimize seedlot
differences caused by non-optimal lifting dates. This procedure would allow
more precise estimates of the actual field performance rankings of improved
families by minimizing error variance due to non-optimal lifting from the
nursery.
160
CONCLUSIONS
Results of this research show RGP is a genetically controlled trait, and
suggests that RGP should be considered in improvement programs as well as in
nursery management of loblolly pine seedlings. Genetic selection for improved
RGP should be successful and yield seedlings better able to withstand
transplanting. If the relationship between RGP and height growth continues
beyond the first two years, RGP may be a screening tool for the early
selection of superior genetic stock.
LITERATURE CITED
DeWald, L.E., P.P. Feret, and R.E. Kreh. 1985. A 15-Day hydroponic system
for measuring root growth potential. in proc. Southern Silv. Res.
Conf. , Atlanta GA. (in press).
Falconer, D.S. 1983. Introduction to Quantitative Genetics. Second Ed.
Longman Group Limited Publ. 340 p.
Feret), P.P. ‘and R.E. Kreh. 1985, Seedling root growth potential as an
indicator of loblolly pine field performance. For Sci (in press).
Feret, P.P., R.E. Kreh, and C. Mulligan. 1985. Effects of air-drying on
survival, height and root growth potential of loblolly pine seedlings.
S. J. Appl. Forestry. (in press).
Jenkinson, J.L. 1984. Seed source lifting windows improve plantation
establishment of Pacific slope Douglas-fir. Pages 115-141 In M.L.
Duryea and G.N. Brown (Eds.) Seedling physiology and reforestation
success. Martinus Nijhoff/Dr. W. Junk Publishers. The Netherlands.
Jenkinson, J.L. 1980. Improving plantation establishment by optimizing
growth capacity and planting time of western yellow pines. U.S.D.A.
For. Serv. Res. Pap. PSW-154.
Jenkinson, J.L. and J.A. Nelson. 1983. 1-0 Douglas-fir: A bare-root
planting option. Pages 63-76 In Proc. W. For. Nur. Council Conf.
Southern Oregon Regional Services Institute, S. OR State College.
Jenkinson, J.L. and J.A. Nelson. 1978. Seed source lifting windows for
Douglas-fir in the Humboldt Nursery. Pages B-77-95 In W. For. Nursery
Council and Intermtn. Nurseryman's Assoc. Combined Nurseryman's Conf.
and Seed Processing Workshop. U.S.D.A. For. Serv.
Johnson, J.D., D.L. Bromlett, R.M. Burns, T.A. Dierauf, S.E. McDonald, and
JoMumestone, . 1982. Pine seedling production in the South: A problem
analysis. Publ. No. FWS-82, Sch. For. and Wild. Sci. V.P.1I. and S.U.
Blacksburg, VA. 33 p.
Ritchie, G.A. and J.R. Dunlap. 1980. Root growth potential: Its
development and expression in forest tree seedlings. New Zea. J. For.
Sede) 10:28 =248.
Soil Conservation Service. 1973. Hayesville series. National Cooperative
Soil Survey Established Series Description Sheet. U.S.D.A. Soil
Conservation Service, Fort Worth, Texas. 4 p.
Stone, E.C. and E.A. Norberg. 1979. Use of root growth capacity in
developing propagation regimes, storage criteria, and nursery stock
certification. In Proc: A sympo;ium on regeneration and management of
of young true fir stands. The True Fir Management Cooperative, Redding,
CAvet Soaps
Weaver, C.H., B. Izlar, K. Xydias, and F.S. Broerman. 1981. Trends in
southern forestry industry pine plantation survival 1960-1978. In Proc.
161
of A.S.A.E. Symp. on Engineering Systems for Forest Regeneration.
A. S- AUB. Bulb: 10-81h 53-55.
162
RESPONSE OF FOUR SOURCES OF LOBLOLLY PINE
TO SOIL ACIDITY EXTREMES
tighael Ge email
Abstract--Four sources of loblolly pine from Florida and ad-
jacent regions were grown on two soil regimes, one very acidic and
the other nearly neutral, in a greenhouse. All sources had greater
height, root collar caliper, and dry weight on the acidic soil.
There was no significant interaction of sources or families within
sources with acidity levels. The Gulf Coast source was significant-
ly shorter in height than the other three sources. While no
important differences were found between the two Florida sources,
some Gulf Hammock families performed well on the soil with acidity
level of pH 6 to 7.
Additional keywords: Marion County, edapic ecotypes, G X E inter-
action, Pinus taeda.
INTRODUCTION
Much interest has been expressed in Florida loblolly pine (Pinus taeda
L.) because of its status as the southern-most source and its reputation for
fast growth (Ladrach, 1980 and Draper, 1975). It is a logical choice for
planting as an exotic in more tropical regions of the world.
Over the years, foresters have come to refer to Florida loblolly pine as
either the Marion County or Gulf Hammock source. The Marion County material
is typically found on deep sandy sites with soil acidity levels of pH 4 to 5.
In contrast, Gulf Hammock refers to Levy and Dixie Counties on the Gulf Coast
and is typified by wet marl soils with soil acidity of pH 6 to 7. Since soil
acidity is such a dominant factor in soil chemistry and the Florida sources
are under intense selection pressure by virtue of the fact they are on the
edge of species' natural range, an interesting question is whether the two
sources have evolved differently into edaphic races.
Another concern involving Florida Loblolly pine is the restricted
breeding population and dim prospects for locating more select phenotypes.
Presently, about 140 selections have been located by the organizations working
with Florida loblolly pine. This is far below the 200 to 400 genotypes
generally suggested as the minimum for a base breeding population (Krug, 1979;
Burdon, et al., 1977; and McKeand and Beineke, 1980). The dearth of natural
stands precludes more mass selection to bolster the breeding population.
This paper reports on a greenhouse study to determine if the two Florida
sources are distinct genetically and whether they could be combined with
adjacent sources of loblolly pine in an expanded breeding population.
1/ ; ARS
— Tree Improvement Project Leader, Chesapeake Corporation, West Point, Virginia.
163
MATERIALS AND METHODS
Fifteen half-sib families were chosen from seed orchard seed lots to
represent ortets originally selected in each of the following geographical
seed sources: Gulf Hammock, Florida (GH); Marion County, Florida (MC); the
Guif Coastal Plain of Mississippi - Alabama (GCP); and the Atlantic Coastal
Plain of Georgia - South Carolina (ACP). The GH seed was collected from
Georgia - Pacific's Gulf Hammock orchard and the MC seed came from clonal
collections at Brunswick Pulp Land Company's Florida Loblolly Seed Orchard.
The other seed lots were selected at random from seed in storage at North
Carolina State University. Relative quality was ignored in choosing seed
li@eSe
A split-plot design was used with three replications. Two soils with
acidities of pH 4.5 and pH 6.5 were the main plots, seed sources were sub-
plots, and the half-sib families were nested within sources. Families in the
main plots were represented by 10 seedlings. A base soil media of pH 4.5 was
developed using equal parts of river sand, peat moss, and coarse vermiculite.
Calcium carbonate was used to adjust the soil acidity to pH 6.5. The seedlings
grew for 16 weeks in cylindrical containers of 164 cc volume. The temperature
settings in the greenhouse were 26 C days and 22°C nights.
Poor germination caused the deletion of 3 families of the Atlantic
Coastal Plain and 1 family of the Gulf Coast source. Seedlings were given
optimum conditions for growth, other than soil acidity, which proved to be
fairly easy to regulate. Approximately two whole pH units were maintained
between the main plot treatments over the course of the study, although they
varied from the target pH levels of 4.5 and 6.5.
At the end of the growth phase, the growth variables looked at were
seedling height, root collar caliper, and stem dry weight. An analysis of
variance was done on plot means for each trait.
RESULTS
Seedling Height
Seed weight accounted for only a small part of the variability associated
with seedling height (r= -0.19) and was not significant. The analysis showed
a significant acidity effect. Seedlings on the low pH soil averaged 27.77 cm,
13% taller than seedlings on the high pH soil (Table 1). There were marked
differences in appearance as well. Seedlings looked healthy and vigorous at
the. lower pH level while those growing under the acidity level of pH 6.5
looked distressed and were less bushy with needles that were a paler shade of
green.
Source rankings were exactly the same under both acidity regimes, with no
Significant interaction of acidity and sources (Table 1). This suggests that
none of the sources tested are especially adapted genetically to either high
or low soil acidity. The fact that Marion County was tallest and Gulf Hammock
was second is in agreement with previous seed source studies (Draper, 1975 and
Labrach, 1980) and is encouraging evidence that the Florida sources respond
Similarly to soil acidity.
164
Table 1.--Seedling height for all four seed sources, by pH level and combined
over pH levels.
a
Seedling Height
Seed Source pH 4.5 pH 6.5 overall
= Se Centimeters, - <= (= -)-.—
Marion County 28.76 25.44 27.10 b 2/
Gulf Hammock 28.26 24.93 26.60 b
Atlantic Coastal Plain 27.97 24.87 26.42 b
Gulf Coastal Plain 26.08 22.95 (aluesoyl | {c
Means 27.77 24.55 26.16
a/
—" Means within a soil pH level no sharing the same superscript are signi-
ficantly different at the 1% level.
The source effect was highly significant in the analysis. A multiple
comparison procedure using Waller-Duncan's Bayesian K-ratio t-test revealed
that the Gulf Coast source was significantly different at the K=100 level
(approximately equal to the 0.05 level). This result was surprising since the
GCP source had the heaviest seed and was ranked first in height at age ten
weeks. This apparent genetic difference in height growth would be a consider-
able handicap if it persisted well into the rotation.
There was a large and highly significant family within source effect.
The large amount of variation within sources indicates that mass selection
would be effective in a breeding program.
While there was no significant family X acidity interaction, a closer
look at the two Florida sources revealed interesting family rank changes.
Using Spearman's Coefficient of Rank Correlation (Steel and Torrie, 1980),
families of the MC source exhibited a high degree of stability over acidity
regimes (rg = 0.85). The rank of GH families over acidity regimes was very
poorly correlated (rs = 0.08), indicating a lack of stability to soil acidity
extremes. These results are supported by linear correlation r values of 0.90
for MC and 0.36 for GH.
Some GH clones apparently are genetically adapted to either low or high
Soil pH values. For instance, clone 23-341/ ranked second under pH 4.5 but
Groppedeaco, last. in, pil 6.5. while. ;clones22-33,,, 22-32, 22-23, wand) 22-4
performed well on the high pH soil.
Root Collar Caliper
Root collar caliper is more sensitive than seedling height to stocking
density (McGilvary and Barnett, 1981). It has limited value as a measure of
growth because of the high seedling sensity in the study (527 seedlings per
Square meter) and the small amount of variation detectable with the measuring
1/ . ; :
— Clone number assigned by the North Carolina State University - Industry Co-
operative Tree Improvement Program.
165
instruments on small stems. Also, stem dry weight accounted for 76% of the
variability associated with root collar caliper.
Diameter growth was significantly better on the low pH value soil where
the mean caliper was 8.5% greater. The analysis found no interaction between
soil acidity and sources or families within sources, although there was some
rank change amoung the sources between pH levels (Table 2). The Gulf Coastal
Plain source had the largest caliper on both pH levels, with no clear second
place. The fact that seed weight accounted for 39% of the variation in
caliper (r = 0.63) and that the Gulf Coastal Plain had the heaviest seed may
explain the good showing of this source material.
Table 2.--Root collar caliper for all four sources by pH levels and combined
over pH Tevels. ©
Root Collar Caliper
_Seed Source pH 4.5 a Gas overal i
S.5.9.5 MUG Ss = Sess =
Gulf Coastal Plain 2585 2.65 21/49) af
Marion County 2.82 2u58 2.7/0 b
Atlantic Coastal Plain 2.84 2.56 Zev Oeb
Gulf Hammock Cad 2.60 2.69 b
Means 2.82 2.60 (oo It |
al”
Means within a soil pH level sharing the same superscript are not
Significantly different at the 0.05 level.
For this trait, sources were basically equal, with no significant
differences. However, once again there were important family within source
differences, indicating enough potential variation to make mass selection
worthwhile. Variability among families ranged from a high of 3.00 mm for
family 24-1 (GCP source) to 2.45 mm for family 22-22 (MC source), a difference
of 22% in root collar caliper and 150% in cross sectional area.
As the data in Table 2 shows, the GH source had slightly larger caliper
(2.60 mm vs. 2.58 mm) than MC on the pH 6.5 soil. When the GCP and ACP
sources were dropped from the analysis, the Florida sources exhibit a
Significant interaction with acidity levels at the 0.05 level. The Guif
Hammock source may be better adapted to high pH levels; some families did show
the ability to grow well on the adverse soil of pH 6.5. Families 22-32 and
22-33 had larger root collar caliper on the high pH value soil than on the low
pH value soil, and family 22-23 was nearly the same on both soil acidities.
These three Gulf Hammock clones were also top performers for seedling height.
This suggests that these clones are tolerant of soil acidity in the range of
pH 6.5. At the same time, they are adaptable enough to show good growth on
Soil with acidity more typical of loblolly pine sites.
Each Florida source had good correlation of family ranks between acidity
extremes when using Spearman's Coefficient of Rank Correlation (r, values were
166
0.53 and 0.86 for GH and MC). A linear correlation check revealed a lower
correlation of r = .46 for GH versus r = .84 for MC. Only one Marion County
clone grew reasonably well on the soil with pH value of 6.5.
Stem Dry Weight
Seed weight or common environment effect contributed 40% of the vari-
ation in stem dry weight at age 16 weeks (r = 0.63). Similar results have
been reported at age 24 weeks when the correlation coefficient was 0.57
(Waxler and van Buijtenen, 1981).
The analysis showed that stem dry weight was strongly affected by soil
acidity levels. Overall, seedlings grown on the low pH soil were 33% heavier
(Table 3). Fluctuations in the target acidity levels indicated that growth
was severely restricted when the soil acidity rose to approximately pH 7; the
ability of the seedlings to absorb nutrients, even if abundantly supplied, and
function well physiologically was impared at pH 6.5 and above.
Table 3:--Stem dry weight for all from sources by pH levels and combined
over pH Tevels.
Stem Dry Weight
Seed Source pH 4.5 pH 6.5 overal]
-------- GhamS=u— ==) =) —
Gulf Coastal Plain 1.454 1.139 1.297 b 2/
Marion County 1.467 1.062 i265) b
Atlantic Coastal e422 1.028 te2z25ub
Gulf Hammock 1.299 1.009 leet 5 4b
Means 1.41] 1.060 1235
/ Res La
2’ Means within a soil pH level sharing the same superscript are not signifi-
cantly different at the 0.05 level.
As in the other two traits measured, there was a significant family
response. Dry weights for families, averaged over pH levels, ranged from
0.925 grams to 1.569 grams, a difference of 70%. There obviously is suf-
ficient genetic variation to make good gains with selection in a breeding
program.
Sources were not found to be significantly different in their dry
weights, at least at the 0.05 level. The Gulf Coastal Plain and Marion County
Sources were nearly equal, with a slight edge to the GCP material (Table 3).
The Gulf Hammock source was consistently last in dry weight, despite the fact
with GH material exhibited good height growth and average diameter growth.
Seedlings of this source apparently contained more water in the above ground
biomass. The reason for this is not known.
Neither sources nor families within sources were found to vary in per-
formance with respect to soil acidity; no statistically significant evidence
of ecotypic adaptation was found. Both of the Florida sources gave high rank
167.
correlations when Spearman's procedure was used to compare family ranks on the
high and low pH value soils, (re was 0.64 for GH and 0.68 for MC). Again,
clones 22-32, 22-33, and 22-23 of GH source grew very well on the less acidic
soil. These clones appear to be well adapted to a wide range of soil acidity
levels.
CONCLUSIONS
The most reliable growth trait examined, seedling height, gave strong
evidence that the two Florida seed sources, Gulf Hammock and Marion County,
are genetically similar in their response to soil acidity. Combining the
Florida sources into one breeding program seems to be justified. Addi-
tionally, there is evidence for including selections from adjacent geo-
graphical sources in an expanded breeding population for Florida loblolly
pine, particularly selections from the Atlantic Coastal Plain of Georgia and
possibly South Carolina. Use of selections from the Gulf Coastal Plain should
be avoided until more reliable data is available. These results must be
considered tentative until field trails either confirm or dispute the
findings.
Fortunately, field trials using the same open-pollinated seed lots for
each source have been established in a join project between the N. C. State
University - Industry Cooperative Tree Improvement Program and University. of
Florida Cooperative Forest Genetics Program. The tests are located at nine
sites chosen to sample the environment within the provenances and provide
information on the extent of genotype X environmental interaction.
While no sources exhibited edaphic adaptation to either high or low soil
acidity, there was evidence of intra-source G X E within the Gulf Hammock
material. Several GH families consistently grew very well on the pH 6.5 soil
and could represent physiological ecotypes. As proposed by Bridgwater and
Stonecypher (1978), the real opportunity here is for realized genetic gains,
made possible by assigning half-sib families to sites to which they are
specifically adapted.
LITERATURE CITED
Bridgwater, F. E. and R. W. Stonecypher, 1978. Genotype by Environmental
Interaction: Implications in Tree Breeding Programs. Proc. Fifth North
American Forest Biology Workshop. Gainesville, Florida. pp. 46-64.
Burdon, R. D., C. J. A. Shelbourne, and M. D. Wilcox, 1977. Advanced
Selection Strategies. Third World Consultation on Forest Tree Breeding.
FAD/IUFRO, Canberra, SAust., March, 1977. 10 pp.
Draper, L., Jr., 1975. Provenance Study of Five Geographic Sources of
Loblolly Pine, Proc. 13th Southern Forest Tree Improvement Conference,
Raleigh, N. €. pp. 83-88.
Kang, H. 1979. Long-Term Breeding. Proc. 15th Southern Forest Tree Improve-
ment Conference. Starkville, Ms., pp. 66-77.
168
Ladrach, W. E., 1980. Variability in the Growth of Pinus taeda in Columbia
due to Provenance. Carton de Colombia, S. A. Forest Research Report
NoomG2o4 (noe
McGilvray, J. M. and J. P. Barnett, 1981. Relating Seedling Morphology to
Field Performance of Containerized Southern Pines. Proc. Southern
Containerized Forest Tree Seedling Conference. Savannah, Georgia.
Dp 39-40)
McKeand, S. E. and W. F. Beineke, 1979. Sublining For Half-Sib Breeding
Populations of Forest Trees. Silvae Genetica 29 (1): 14-17.
Steel, R. G. D. and J. H. Torrie, 1980. Principles and Procedures of
Statistics. Second Edition. McGraw-Hill, Inc., New York. 633 pp.
Waxler, M. S. and J. P. van Buijtenen, 1981. Early Genetic Evaluation of
Loblolly Pine. Canadian Journal of Forest Research 1]: 351-355.
169
Winner of ihe First TONY SQUILLACE AWARD
DYNAMICS OF IMPROVED LOBLOLLY PINE PLANTATIONS AND THE
IMPLICATIONS FOR MODELING GROWTH OF IMPROVED STANDS
Marilyn A. Buford and Harold E. Burkhart?
Abstract.--A study was initiated to examine the dynamics
of genetically improved loblolly pine plantations and to
develop guidelines for incorporating the effects of genetic
improvement into various types of growth and yield models for
loblolly pine plantations. The limited data base from stands
of improved stock dictated that any modeling effort concentrate
on the synthesis of fragmented information from many types of
studies rather than the usual data fitting procedures.
A series of hypotheses concerning stand dynamics and
growth patterns in stands of improved stock relative to stands
of unimproved stock were developed and tested. Results of
these tests indicate: 1) at the seed source and family
levels, the shape of the height-age curve is dictated by the
site, but the level of the height-age curve is dictated by the
seed source or family; 2) at the seed source and family levels;
the shape of the height-diameter relationship at a given age is
determined by site and initial density while the level of the
height-diameter relationship is determined by the seed source
or family and is directly related to the dominant height of
the seed source or family at that age; and 3) given that
silvicultural treatments are the same and are equally intense
and successful, variances of height and diameter in stands
originating from selected genotypes are not different from
those in genetically unimproved stands.
Implications for modeling growth of stands originating
from selected genotypes ares 1) genetic improvement affects
the rate at which stands develop, but does not fundamentally
alter the pattern of stand development from that of unimproved
stands; 2) changes in genetic material on a given site will
likely affect the level, but not the shape, of such basic
relationships as the height-age and height-diameter relation-
ships; and 3) correctly characterizing the height-age profile
will be very important.
e : Pinus taeda L.,» height-age relationship,
height-diameter relationship, stand-level variance.
1. Champion International Postdoctoral Fellow and Thomas M. Brooks
Professor of Forest Biometrics, Virginia Polytechnic Institute and State
University, Blacksburg, Virginia. The authors acknowledge Champion
International Corporation, the Loblolly Pine Growth and Yield Research
Cooperative, North Carolina State University-Industry Tree Improvement
Cooperative, Crown Zellerbach Corporation, and the Southern Forest Tree
Improvement Committee for support and data release.
170
In modeling the growth and yield of forest stands, we seek to
mathematically interpret the biological relationships that underlie and
drive stand development. The logical way to approach the problem of
modeling the growth and yield of genetically improved stands of loblolly
pine (Pinus taeda L.) is by studying the dynamics of such stands. An
understanding of the growth and dynamics of genetically improved stands
relative to unimproved stands is important for decisions regarding
selection and breeding as well as forest management.
Certain basic relationships are key components of stand dynamics.
They are: 1) the development of dominant height through time or the
height-age relationship; 2) the development of the height and diameter
distributions through time; 3) the relationship of the mean height by
- diameter class across the range of diameters at a given age, or the height-
diameter relationship; and 4) the mortality-time relationship. Hypotheses
concerning the similarities and differences between stands of improved and
unimproved stock were developed and tested for the first three
relationships listed above. This paper details the hypotheses tested, the
test results and the implications of the results for modeling growth of
stands of genetically improved stock.
DATA
The data available to this study were diverse types and not, by study
design or plot structure, the kind usually used for growth and yield
analysis. This dictated that the hypotheses be tested and the implications
be proposed by the synthesis of fragmented information rather than the
usual modeling techniques.
The data base used is comprised of three components: 1) the Loblolly
Phase of the Southwide Pine Seed Source Study up to age 25 (Wells and
Wakeley 1966); 2) a 15 year-old block-plot half-sib progeny test planted
near Bogalusa, Louisiana, belonging to Crown Zellerbach Corporation; and 3)
three 16 year-old, ten-tree row-plot progeny tests located in eastern
Virginia, eastern South Carolina and north-central Alabama released to this
project by the North Carolina State University-Industry Tree Improvement
Cooperative. The row-plot progeny test data were grouped into half-sib
families on the male parent thereby combining rows from different locations
within a rep in an attempt to overcome the environmental artificiality of
the row-plot design. The row-plot tests were then analyzed at the half-sib
level.
The block-plot progeny test is replicated 4 times at the same location
with each rep containing twelve 121 tree plots (11 x 11 trees) with the
inner 49 trees as measurement trees. Spacing is 8 x 8 feet. Each rep
contains 11 plots of selected genotypes and 1 plot of local woodsrun origin
as a control.
A complete description of the Southwide Pine Seed Source Study design
is given in Wells and Wakeley (1966).
Analysis of the Seed Source Study allowed hypotheses to be tested
regarding similarity or difference of growth pattern of different seed
171
sources at the same location and at different locations. Analysis of the
half-sib block-plot progeny test allowed hypotheses to be tested regarding
growth patterns of half-sib families at the same location. The row=plot
progeny test data were used to further test hypotheses accepted using the
Seed Source Study and the half-sib block-plot progeny test data.
METHODS AND RESULTS
anes Tat ionet
When modeling stand development, the most important relationship to
understand and correctly characterize is the height-age relationship. The
analysis of the height-age profiles in the data available was therefore of
primary interest. J
Nance and Wells (1981) used the Southwide Pine Seed Source Study and a
fairly inflexible model for height growth to study differences in site
index among different seed sources. One result of their work is that at
any given location of the Seed Source Study, the shape of the height-age
curve is the same for all seed sources, but the level of the curve is
affected by seed source and block. To determine whether or not these
results were an artifact of the model used by Nance and Wells (1981) the
analysis was repeated using the very flexible Richards’ function:
H = A(l-exp(=-b*age) )**c » (1)
where: H = height at any given age
A = asymptotic or maximum height
b = rate parameter
c = shape parameter.
Equation (1) was fitted to the tallest seven trees at each age (roughly
_ analogous to the tallest 100 trees per acre) for each seed source x block x
- location combination. Analysis of variance for As bs, and c and graphical
analysis of the resulting curves for each location showed significant seed
source and block effects on A, the asymptotic height, but generally no seed
source or block effects on b and cy, the rate and shape parameters,
respectively, Differences did occur among all the parameters from location
to location.” These results support the findings of Nance and Wells (1981)
. and also indicate that an extremely flexible function is not necessary for
this analysis.
Extending the conclusions about height-age patterns in the Seed Source
Study to a general hypothesis, patterns of the height-age profiles in the
block=plot progeny test were examined. The equation
2. In the interest of space, exhaustive tables of test and fit
statistics are not given in this paper. Such tables can be obtained from
the senior author on request. All tests of significance were carried out
with ® = 0.05.
172
+
log (H) = a + b(l/age), (2)
wheres H = height at any given age
a = level or intercept parameter
b = slope or shape parameter
log = logarithm base e,
was fitted to the tallest seven trees at each age for each plot (family x
block combination). An analysis of variance for the estimates of the slope
parameter, bs was done following the form of Table 1. Results of this
analysis showed that there are no family and no rep effects on the shape
parameter. That iss the height-age curves are the same shape for all the
families across all the reps. Equation (2) was fitted to the data from
each plot again while maintaining a common slope parameter, b» for all
plots. An analysis of variance (Table 1) was performed on the estimates of
the intercept and the results showed significant family effects on the
intercept, but no rep effects on the intercept.
Table 1.--Form of analysis of variance for block (rep) and seed source
(family) effects on the parameters of the height-age and height-
Source of Degrees of Mean
Variation Freedom Square F-ratio
Block (Rep) b-1 MSB F .=MSB/MSE
Seed Source (Family) s-l1 MSS F o=MSS/MSE
Error (b-1)(s-1) MSE
Extending this analysis to the row-plot progeny tests, equation (2)
was fit to the tallest 15% of the trees at each age for each plot (half-sib
family x rep combination) for the three progeny tests. An analysis of
variance on the estimated slopes was done following Table 1. Results of
this analysis showed that there are no family and no rep effects on the
slope or shape parameter, b. Equation (2) was fitted to the data from each
plot again while maintaining a common slope parameter, b, for all plots
within a test. An analysis of variance (Table 1) was performed on the
estimated. intercepts and the results showed significant family and rep
effects on the intercept term in all three tests.
The conclusion of this phase of the study is that at a given location,
the shape of the height-age profile is the same among seed sources or
families, but the level of the height-age curve differs by seed source or
family. This result is consistent from the Southwide Pine Seed Source
Study to the block-plot half-sib progeny test to the row-plot progeny
tests.
Height-diameter relationships are used in many growth and yield
models to predict the mean height for a given diameter or diameter class.
173
The diameter and predicted height values are then used in stand volume and
value calculations.
The wide use of this relationship dictated its
investigation. Only data from unthinned studies or unthinned parts of
studies were suitable for analysis of the height-diameter relationship as
thinning destroys stand structure and, with it, the base line development
of the height-diameter relationship.
Study unthinned by age 15 were used (see Table 2).
test was thinned after age 8.
Nine locations of the Seed Source
The block-plot progeny
Given that the dominant heights ranged from
38 to 50 feet at age 8 in these plots, the height-diameter relationship was
well-developed at age 8 and these data were appropriate for the analysis.
The three row-plot progeny tests were unthinned and the age 16 data from
these studies were used in the analysis.
Table 2. = ocaelen of the nine Southwide Pine Seed seule ss i plantations
Location
C AL (2)
E MD (1)
N MS (2)
S MS (1)
S MS (2)
SE LA (1)
SE LA (2)
SW GA (1)
W.SC (2)
The function used
analysis was:
County or
Parish
Coosa
Worcester
Winston
Pearl River
Pearl River
Washington
Washington
Dooly
Newberry
State
Alabama
Maryland
Mississippi
Mississippi
Mississippi
Louisiana
Louisiana
Georgia
South Carolina
Series
NRMNFMNFEN EF PD
to model the height-diameter relationship in this
log(H) = a + b(1/D)
wheres: H
a
b
log
Equation (3) has been
mean height for a given diameter, D
level or intercept parameter
slope or shape parameter
logarithm base e.
(3)
found to work well for loblolly pine and examination
of residual plots showed that it fit the data from all the studies well.
Equation (3) was fitted to all the height-diameter pairs for each seed
source x block combination (plot) for each of the nine locations of the
Seed Source Study used.
estimated slope parameters, bs, at each location.
seed source did not affect the shape of the height-diameter curve.
An analysis of variance (Table 1) was done on the
At 8 of the 9 locations,
At 8 of
the 9 locations, blocks did not affect the shape of the height-diameter
curve. Given these results, equation (3) was refitted to each plot while
maintaining a common slope within each location.
The analysis of variance
indicated in Table 1 was performed on the estimates of the intercept, a,
within each location. At 2 of the 9 locations, blocks significantly
affected the intercept. At 3 of the 9 locations, seed source significantly
affected the intercept, and at 3 of the 9 locations, block and seed source
significantly affected the intercept. Recall that there were significant
block and seed source effects on the intercept of the height-age
relationship. The estimated slopes were significantly different across
locations. Simple linear regressions were calculated for the estimated
intercepts on dominant height at age 15 of the appropriate plot. Eight of
the 9 regressions had r- values greater than 0.79. Using r- simply as a
measure of association, the intercept, or level, parameter of the height-
diameter relationship at age 15 is strongly related to the dominant height
at age 15 of that source on that site. The conclusion from this portion of
the analysis is that the height-diameter relationships are the same shape
across seed sources at any given location at age 15 and the level of the
relationship is directly and strongly related to the dominant height of the
source at age 15.
Extending this conclusion to a general hypothesis, the same type of
analysis was performed on the block-plot progeny test data at age 8.
Equation (3) was fitted to the height-diameter pairs for each plot (family
x block combination). The analysis of variance indicated in Table 1 was
performed on the estimates of the slope, b. The results of the analysis of
variance showed no block or family effects on the slope parameter. Using a
common slope, bs, equation (3) was refitted to the data from each plot. An
analysis of variance of the form in Table 1 was carried out on the
estimated intercepts. There were no rep effects, but there were
significant family effects on the intercept parameter. Recall that there
were no rep effects but there were significant family effects on the
intercept of the height-age equation for these plots. The simple linear
regressign of the estimated intercepts on the dominant height of the plots
had an r- of 0.93. The conclusion from this portion of the analysis is
that the height-diameter relationships are the same shape across families
at age 8 and the level of the relationship is directly and strongly related
to the dominant height of the family at age 8.
Performing the same type of analysis with the row-plot progeny tests,
equation (3) was fitted to the height-diameter pairs for each plot (half-
sib family x rep) for the three separate tests. For each of the three
tests, an analysis of variance (Table 1) was conducted on the estimates of
the slope, b. There were no significant rep or family effects on the slope
parameter, b, in any of the three row-plot progeny tests. Maintaining a
common slope for each test, equation (3) was refitted to the height-
diameter pairs for the three tests. An analysis of variance (Table 1) was
performed on the resulting estimates of the intercept, a. There were
“significant family and rep effects on the intercept in all three row-plot
progeny tests. Recall the significant rep and family effects on the
estimates of the intercept for the height-age relationship for these data.
A simple linear regression of the estimated intercepts of the height-
diameter relationship on the plot dominant height was calculated for each
of the three progeny tests. All had r° values greater than 0.88. The
slope parameters were different in the three tests. From this analysis of
the three row-plot progeny tests, the conclusion is that the height-
diameter relationships are the same shape across families within a test at
age 16 and the level of the relationship is directly and strongly related
175
to the dominant height of the family at age 16.
The conclusion of this phase of the study is that at a given location,
the shape of the height-diameter relationship at any age is the same among
seed sources or families, but the level of the height-diameter curve at
that age differs by seed source or family. In additions, the level of the
height-diameter curve is directly and strongly related to the dominant
height of that seed source or family at that age. These results are
consistent for the Southwide Pine Seed Source Study, the block-plot half-
sib progeny test, and the row-plot progeny tests.
Stand-jevel variance
The shape or spread of the diameter and height distributions, as well
as their relative location, determine the ultimate value of a stand.
Accordingly, the variances of the height and diameter distributions are of
interest in modeling the growth and yield of loblolly pine stands. It has
been conjectured that while increasing the mean tree size, genetic
selection could reduce the stand value by reduction of the stand-level
variance and possible production of fewer large trees (Nance and Bey 1979,
Thurmes 1980).
To test this proposal, data from unthinned family block plots were
needed. Because the block-planted half-sib progeny test had been thinned
after age 8, the data at age 8 were used, since given the initial spacing
of 8 x 8 feet and the dominant heights ranging from 38 to 50 feet, the
distributions of height and diameter were well developed.
Simple variances of height and diameter were calculated for each plot
(family x rep combination) of the block=plot half-sib progeny test.
Bartlett's test for homogeneity of variance was performed for both diameter
and height in each rep; that is, the variances of the 12 families within a
rep were compared. The hypothesis of homogeneity of variance was accepted
for the variance of diameter in each of the four reps. The hypothesis of
homogeneity of variance was accepted for the variance of height in three of
the four reps. These results indicate that the variances of height and
diameter generally did not differ among the families within a rep.
An observational analysis was done to determine the relative sizes and
ranks of the variances of height and diameter of the twelve families within
each rep. For each rep, the variances of height and diameter were ordered
from largest to smallest. There were no consistent family rankings across
the reps for diameter or height. The variance of diameter for the woodsrun
control family ranked largest in one rep, third largest in one rep, and
seventh largest in two reps. The variance of height for the woodsrun
control family ranked fourth largest in one rep, sixth largest in one rep,
seventh largest in one reps and tenth largest in one rep. It is clear that
the height and diameter variances of the woodsrun control family are not
consistently larger than those of the eleven selected genotypes.
The general conclusion to be drawn is that given that silvicultural
treatments are equally intense and successful, the variances of height and
diameter in stands originating from selected genotypes are not different or
consistently smaller than those in stands originating from genetically
176
unimproved stock.
CONCLUSIONS AND IMPLICATIONS
The major conclusions to be drawn from the work presented here are:
1) given that silvicultural treatments are the same and are equally intense
and successful, variances of diameter and height in stands originating from
selected genotypes are not different or consistently smaller than those in
stands originating from genetically unimproved stock; 2) at the seed source
and family levels, the shape of the height-diameter relationship at a given
age is determined by the site and initial density while the level of the
height-diameter relationship is determined by the seed source or family and
is directly related to the dominant height of the seed source or family at
that age; and 3) at the seed source and family levels the shape of the
height-age curve is dictated by the site, but the level of the height-age
curve is dictated by the seed source or family. A very important result of
this work is the consistency of the results concerning the height-diameter
relationship and height-age relationship from the Southwide Pine Seed
Source Study to the block-planted half-sib progeny test to the row-plot
progeny test. This consistency suggests implications for growth and yield
modeling.
Implications for modeling growth of stands originating from selected
genotypes are: 1) genetic improvement affects the rate at which stands
develop, but does not fundamentally alter the pattern of stand development
from that of stands of unimproved stock; 2) changes in genetic material on
a given site will likely affect the level, but not the shapes, of such basic
relationships as the height-age and height-diameter relationships; 3) from
the analysis of the height-diameter relationship, it is likely that the
development of the height and diameter distributions will follow from the
development of dominant height; and 4) correctly characterizing the height-
age profile for a given site will be of primary importance.
LITERATURE CITED
Nance, W. L. and C. F. Bey. 1979. Incorporating genetic information in
growth and yield models. Jn Proceedings 15th Southern Forest Tree
Improvement Conference, p. 140-148.
Nance, W. L. and 0. 0. Wells. 1981. Site index models for height growth
of planted loblolly pine (Pinus taeda L.) seed sources. In Proceedings
16th Southern Forest Tree Improvement Conference, p. 86-96.
Thurmes, J. F. 1980. Predicting the effects of intensive cultural
practices on optimal management strategies for loblolly pine
plantations. Unpub. M.S. Thesis, Virginia Polytechnic Institute and
State University, Blacksburg, Virginia, 90 p.
‘Wells, O. O. and P. C. Wakeley. 1966. Geographic variation in survival,
growth, and fusiform-rust infection of planted loblolly pine. Forest
Science Monograph 11; 40 p.
177
2
: nah ek 3
Mr .
™ i
+f iy
i
a Oe ey pis
ye
’ if o
hn,’ ro
Tan eo & a;
4 eo
Va J ee
4, 4
i 2%
‘ ae
% ¥/ aie.
st
{
fi 4
i
4
Ni
pn tees
En Ry
i a
was
ri : 5 ate
ry 4 ae ay
~ as Ye
u
“e F.
f
i y
SEED ORCHARD MANAGEMENT
MODERATED BY DR. RAY GODDARD
University of Florida
ti
\
Vtiise
ees,
i al
Pay!
wnt
Hy RN Me AE MA init
j vey i ‘ Anh na! ‘ wr . i i'l
ESTIMATING POLLEN CONTAMINATION IN
LOBLOLLY PINE SEED ORCHARDS BY POLLEN TRAPPING
1
Michael Crecnmood =. and Terry Rucker a.
Abstract.--The construction of an inexpensive, easy-to-build pollen
trap is described, which rotates to keep the trap oriented into the
wind at all times. These traps have been deployed for several years,
both within and surrounding four loblolly pine (Pinus taeda L.) seed
orchards ranging in age from 7 to 25y. A comparison of pollen
caught on both the orchard and background traps permits an estimate
of how much pollen the orchard is producing relative to background.
The resulting estimates of background contamination are very similar
to those obtained with other methods.
Keywords: pollen contamination, pollination, seed orchards, Pinus
taeda.
Loblolly pine seed orchards of the North Carolina State and Western
.Gulf Cooperative Tree Improvement Programs produced over 60 tons of improved
loblolly pine seed in 1982, enough to grow about 1 billion seedlings (Talbert
et al., 1985; Byram et al., 1985). After genetic roguing of the orchards, the
predicted volume gain at rotation from these improved seedlings is about 122.
However, these gains assume that there is no pollen contamination from outside
the seed orchards. Squillace and Long (1981), citing several types of
estimates of background contamination, suggest that contamination may range
from 30 to over 80%, even in mature orchards surrounded by isolation zones.
Friedman and Adams (1981), estimate that outside contamination in a 16-year
old loblolly pine seed orchard was 282%, based on detection of several allozyme
gene markers via electrophoresis. Pollen contamination of 30 or 80% will
reduce a projected 12% gain to 10.2 or 8.22 respectively. Given the amount of
seed now being produced by seed orchards, minimizing the impact of background
pollen should be a major concern. Therefore, measuring orchard polien
production and the contribution of background pollen should be a routine part
of quality control.
The most accurate way of assessing background contamination is by using
genetic markers for such traits as allozymes or monoterpenes. But these
methods, although potentially precise, require relatively sophisticated
equipment and methods. A simple and possibly quite accurate method of
determining background contamination is to compare pollen production within
the orchard to background pollen trapped nearby. Koski (1970) used a similar
method for estimating background contribution to total pollination in Scotch
pine plantations in Finland. Here we describe a simple method of measuring
the relative amount of pollen contributed by both the orchard and background.
|!
Peo fessor, College of Forest Resources, University of Maine, Orono, ME.
— Lead Technician, Southern Forestry Research Dept., Weyerhaeuser Co., Hot
Springs, AR.
179
Contamination is then estimated as a simple ratio of background production to
total orchard production. Estimates of background contamination by pollen
trapping are being verified by comparison with estimates made by analysis of
allozymes.
METHODS
Pollen flight in four Weyerhaeuser Co. loblolly seed orchards was
assessed by means of pollen traps changed at 24-hour intervals throughout the
period of pollen flight. The daily peak in pollen flight was detected by
changing traps at hourly intervals on 4/11/80 (see Fig. 1). Each trap
consists of.a glass microscope slide bearing a piece of double coated tape~
(about lcm’), held at a 45° angle by a clothespin mounted on a vane which
keeps the trap oriented into the wind at all times (see Fig. 1, inset).
Tn
Pollen Flux During 4/11/80 - 4/12/80
2000
Seed Orchard
No. Pollen Grains/ eas pimtua
cm2/h a
| CLOTHES PIN
(EPOXIED TO
1/2" DOWEL)
1000
A \
. “\ WOODEN BEARING
(1/2* DOWEL TAPERED
6* stance J h On TOP)
—
yi sa
AM PM AM
Time
Figure 1.--Diurnal pattern of seed orchard and background pollen flight,
diagram of pollen trap.
Materials required to construct the vane-type pollen trap are: 1) 0.5
in. diameter wooden dowels; 2) wooden clothespins; 3) aluminum nails (2.5 in.
long x .12 in. dia.); 4) aluminum sheets (very thin aluminum plates used for
the printing of newspapers, when cleaned, work very well and are cheap); 5)
180
2.5x7.6 cm (1x3 in.) glass microscope slides with frosted end; 6) Scotch brand
#666 double coated tape with liner (0.5 in. width). they. 0; Seam wooden
dowels are cut into l-ft. lengths. Ome end of the dowel is then cut at a 45°
angle to serve as the attachment point for the wooden clothespin. A thin
slit, approximately 2 inches long, is cut in the other end of the dowel for
attachment of the aluminum vane. The clothespin is glued to the dowel using
contact cement or some other waterproof glue. The clothespin is oriented so
that the glass slide, when held by the clothespin, is angled toward the vane
end of the dowel. At this point, the dowel with clothespin attached can be
painted. The aluminum vane, when cut to a desired size and shape, is inserted
into the slit and attached to the dowel with aluminum nails. To avoid
splitting the dowels, the heads should be cut from the nails to allow them to
be mounted into a. drill for insertion. Once the aluminum vane is attached,
the excess nail should be cut off and filed flush with the surface of the
dowel. To determine where to drill the hole in the dowel for attachment to a
wooden stake, a slide is mounted on the clothespin and the balance point for
the pollen trap is located. A hole large enough to allow rotation of the trap
about the aluminum nail is drilled in the dowel at the balance point. A 1.25-
inch segment of dowel with one end rounded serves as a bearing for pollen trap
rotation (see Fig. 1).
Keeping the traps oriented into the wind resulted in up to a /-fold
increase in pollen catch over a trap fixed in a SW direction. Whenever
possible, the traps were changed before rainfall, since rain will wash off
significant quantities of pollen (0.6 cm of rain washed off over 60% of pollen
trapped previously). After drying, however, the adhesive properties of the
tape were restored.
In order to. monitor pollen flight within the seed orchard, the traps were
placed in open areas where trees had been removed, to minimize the impact of
individual trees. Although Koski (1970) reports maximum pollen catch at crown
level in Scotch pine, mounting traps in tree crowns is very inconvenient.
During heavy pollen flight in the J. 1D Weyerhaeuser seed orchard, we
compared pollen catch on traps within the crowns of 5 different trees at a
height of about 12M, with a trap located near each tree approximately 1M from
the ground. The mean catch over a 24-hour period for the crown traps was
about .3100 grains/cm compared with 2,800 for the ground traps. Since the
difference was small, we adopted the easier option of deploying traps on 1.8M
tall stakes so they could be easily reached.
Background pollen was trapped in open areas at least 300M from the edge
of the orchard or any non-orchard sources of loblolly pine pollen. At one
seed orchard, wind direction and velocity were recorded with a vane
anemometer, while at the other three, the wind direction during orchard pollen
flight (which usually occurs in the morning) was noted visually and recorded.
At least four background traps were located approximately north, south, east,
and west of the orchard. Traps located downwind during heavy production of
orchard pollen could then be excluded from background averages. Ten traps
were located throughout the seed orchard in open areas where trees had been
removed. ;
Several transects, consisting of 4 traps 30M apart, were also placed
downwind from the edge of the orchard to assess how pollen catch changes with
181
distance from the orchard. At Lyons, background traps were also located in
the center of a 300-acre field about a mile from the orchard site (the Cato
traps - see Fig. 2).
GRAMS / CM* / 24 HOURS
2400 6
Lyons (1984 data)
ae Kap Ss in .onenand
2000 O——] Background traps 2, ¥ ¥ <)
near orchard
1600 e—— Oa GAT On (naps p ul
g—@ Female stage 2
=
op)
| 200 ei,
|
=
800 2 Ww
LL
400 |
O O
3/10
Figure 2.--Seasonal pattern of pollen flight for orchard located in Lyons, GA.
Background trapped near the orchard and in a 300 acre field about
1 mile from the orchard (Cato) is also shown.
The traps were collected at the same time every day (either early morning
or late afternoon, depending on the location), and the tape was covered with a
plastic cover slip to prevent contamination. Number of pollen grains trapped
per cm super cestimated by counting 10 separate fields of view at 100x or
430x using a compound microscope. Thee \aréad oft’ theses iteldy tote vaecwr was
calculated after measuring its diameter ae a stage micrometer. The results
are presented as number of pollen grains/cm*/24h.
Stage of female development was estimated for 10 clones, representing
early, late, and average occurrence of female receptivity. Fifty or more
strobili were observed from a lift truck on each of 3 ramets/clone every one
or two days, and stage of development (Bramlett and O'Gwynn, 1980) was
recorded for each.
182
RESULTS and DISCUSSION
After several years of observation, the orchard pollen almost always
flies in mid to late morning after the catkins have dried out and when the
wind is sufficiently strong to move the branch (see Fig. I A similar
diurnal pattern is observed with other wind pollinated species (Ogden et al.,
1969). Occasional afternoon pollen flights have been seen if morning rain was
followed by a period of clearing and drying. Virtually no pollen is shed on
rainy days. A comparison of hourly pollen catch for both an orchard and
nearby background trap (upwind from the orchard) during heavy orchard pollen
flaiche sdisha shown) janie Figurey 1: Note the great bulk of orchard pollen flew
between 9 and 11 a.m., while background catch was fairly uniform, showing a
slight peak between 3 and 4 p.m. Little pollen was trapped between 3 p.m.
and 12 midnight, or 12 midnight to 6 a.m. the next day.
Since thé background catch shown in Fig. 1 did not peak at the same time
of day as the orchard, we can be reasonably certain that the background trap,
located upwind from the orchard, was not receiving residual pollen from the
seed orchard. Nonetheless, on a daily basis, background peaks, whether
adjacent to the orchard or a mile away, do occur at roughly the same time as
orchard peaks. (see Figure 2). This suggests that the background pollen
trapped was probably shed the same day as the orchard pollen. If we assume 1)
that background pollen was shed at the same time as the orchard pollen; 2)
that the wind velocity averaged from 10-15 mph on 04/11/80; and 3) that
background catch peaks at 3 p.m. (about 6 hours after peak shed in the
orchard see Fig. 1), then some of the pollen trapped late in the day could
have traveled 60-90 miles.
To assess the possible impact of orchard pollen on background traps
located downwind from the orchard, transects were located in open areas
normally downwind from the orchards. Table 1 shows the catch averaged over
several days versus distance from the edge of the orchard (Table 1).
Table 1.--Impact of orchard pollen on background estimates - pollen catch vs.
distance downwind from orchard.
Location Orchard 30M 60M 90M 120M Background
eee ceriaeusen”” PZ mes59) 7a23720 9 19790) 2037 ith
Comfort! 2464 2440 1846 1947 1428 909
iene! | | G080 CATE 593) Sala. 407 351
a/ Results from 2 transects over 8 days.
— Results from 2 transects over 4 days.
=" Results from 4 transects over 12 days.
At 120M downwind from the orchard, the pollen catch diminished
considerably from that in the orchard but was still 13 to 45% higher than the
mean of background traps not located downwind. Wang et al. (1960) showed that
while pollen catch from a single tree declines logarithmically with distance,
183
the decline from a stand is much less steep and appears somewhat linear.
Background traps that are downwind from the orchard on a given date should be
excluded from background estimates. We recommend that background traps be
located at least 300M from the periphery of the orchard or any local source of
pollen.
A comparison of pollen trapped throughout the pollination season both
inside and outside four seed orchards, over two successive years, is shown in
Table 2. Note that the older seed orchards produce the most pollen, but there
is considerable variation by year. Also, total background catch is not
correlated well with orchard production (total orchard minus background).
Variation in orchard production only explains 15% of th variation in
-background catch across all seed orchards and years (r°=0.15), so the
magnitude of background catches shown here are not significantly related to
seed orchard pollen production.
The patterns of pollen flight and female receptivity shown in Fig. 2 are
representative of all orchards studied here. Both background and orchard
pollen flies during the receptive period of most female strobili within the
orchard.
Pollination in loblolly pine is a two-step process, the first being
accumulation of pollen on the micropylar horns, the second being transfer of
the grains to the nucellus by rain or the pollen drop (Brown, 1984; Greenwood,
1985). Any pollen grain reaching the micropylar horns, whether it arrives
early or late, has an equal chance of reaching the nucellus and presumably
germinating (Greenwood, 1985). An estimate of the contribution of background
pollen to total pollination (in the orchard) should, therefore, be equal to
the ratio of total background to total orchard (the latter includes both
background and _ orchard). Estimates of contamination expressed as % total
orchard pollination contributed by background are shown in the last column of
Table 2.
Table 2.--Background pollen catch as percent total catch in 4 seed _ orchards,
for two yeas each. Both background and orchard pollen were trapped
throughout pollination period.
TOTAL TRAPPED Background as
Orchard Year Background Total Orchard % Total Orchard
Washington, NC. 1983 10,868 | 34,362 322
Est. 1959. 1984 N35 2)3)5) 50,968 31%
Comfort, NC. 1983 16,342 27,086 602%
Est. 1974 1984 Tpoy? LON299 39%
Lyons, GA. 1983 (Shy OZ 11,209 UWE
Eisites97is 1984 4,782 Wk 5 S)sK0) 422
Magnolia, AR. 1982 TIES SAS) 12,838 882%
Ise G | LYYZ 1983 16,130 Dye i813) 68%
184
Estimates of contamination range from 3l to 88%, similar to the range
presented by Squillace and Long (1981). As expected, the oldest orchard
showed the least contamination. However, the Magnolia orchard has sustained
very high contamination because background there was very high both years. On
the other hand, background was consistently low for both years at _ Lyons.
Clearly, orchard location can significantly affect background load, especially
when the orchard is young.
We are currently verifying the estimation of background contamination at
the Washington, N.C., seed orchard with allozyme markers, and our first
results show close similarity to those presented here (N. C. Wheeler and M. S.
Greenwood, unpublished data). As mentioned earlier, other workers, also using
biochemical genetic markers, have obtained a comparable range of estimates for
southern pine seed orchards.
LITERATURE CITED
Bramlett, Die anduC. wh. Ol Gwyn) 1980). Recognizing developmental stages
‘ in. southern pine flowers. USDA Forest Service General Technical Report
SE-18.
Brown, Sends 1984. Pollination mechanisms in Pinus taeda L. M.S. Thesis,
Dept. of Forestry, No. Carolina State Univ., Raleigh, N.C. 30 pp.
Bynai sae DasetWiqnGre rn LOWwes (GC. Roi McKinley, J. K.. Robinson, A..F. Stounder, , and
J. Ps van Buijtenen.,* 1985. 32nd Progress Report at the Cooperative
Forest Tree Improvement Program, circular 269. Texas Forest Service,
College Station, TX.
Friedman, S. T., and W. T. Adams. 1981. Genetic efficiency in loblolly pine
Seeds Onenacds 0 broe., LOth Sos, Fore \lree “ imp.) Cont 4, Blacksburg. VA: ;
ppe 213—-220'.
Greenwood, M. S. 1985. Gene exchange in loblolly pine: the relation between
pollination mechanisms, female receptivity, and pollen availability in
seed orchards. Ms. in preparation.
Kosiiere 1970. A study of pollen dispersal as a mechanism of gene flow in
Condfers. Commuin. Inst.) Fenn’. 70, 738 p:
Ogden, E. C.,.J. V. Hayes, and G. S. Raynor. 1969. Diurnal patterns of pollen
emission in Ambrosia, Phlenm, Zea, and Rieinus. Amer. J. Bot. 56:16-21.
Squaiitaice rw ivAL Ea yowand (EB. M. Long. 1981. Proportion) hol spollen, fron
non-orchard sources. In Franklin, E. C., ed.’ Pollen mgt. handbook. Agr.
Handbook 587, Washington, D.C. USDA, pp. 15-19.
Maikenstre dic mineisnne Jie Wemicm maniduk (De sAtrnolids . 985. mCositswands bene tats) of, ‘a
mature first generation loblolly pine tree improvement program. J. For.
83: 162-166.
185
Wang, C. Wes). 0. yRerny, and A. G. Johnson. 1960. Pollen dispersal of slash
pine (Pinus elliottii Engelm) with special reference to seed orchard
management.. Silvae Genetica 9: 78-86.
186
SUPPLEMENTAL MASS POLLINATION OF SINGLE CLONE ORCHARDS
FOR THE PRODUCTION OF SOUTHERN PINE HYBRIDS
Donald R. Knezick, John E. Kuserl/, and Peter W. Garrett2/
Abstract.--Supplemental mass pollination, "mistblowing", of a
multiclone pitch pine orchard with loblolly pine pollen produced
an average of 11% hybrid seed. Results of a controlled
pollination study indicate that mistblowing isolated single clone
. pitch pine orchards would produce much higher percentages of pitch
x loblolly hybrids. This technique could also be used to mass
produce other commercially valuable southern pine hybrids.
Additional keywords: Pinus rigida, P. taeda, inbreeding
depression, seed orchard, supplemental mass pollination.
The northern range of loblolly pine, Pinus taeda L., extends into
Maryland, Delaware, and even southern New Jersey (Little, 1971).
Unfortunately when loblolly is planted in colder areas to the north and west,
it is susceptible to snow and ice damage and winter desiccation. Pitch pine,
P. rigida Mill., although cold hardy, is noted for slow growth and poor form.
Having observed fast growing hybrids between the two species growing around
loblolly plantations in Maryland and New Jersey, Dr. Silas Little of the U.S.
Forest Service saw the potential of using the hybrid for reforestation in the
northeastern United States.
In the early 1960’s, the U.S. Forest Service and Westvaco signed a
cooperative agreement to breed and field test pitch x loblolly pine hybrids.
Under the direction of Dr. Little and Fred Trew of Westvaco, the cooperative
intensively selected 33 loblolly pines in Maryland and Delaware, and 32 pitch
pines from Virginia, West Virginia, Maryland, New Jersey, Pennsylvania, New
York, Massachusetts, New Hampshire, and Maine. In 1964, a clonal breeding
orchard was established at the Northeastern Forest Experiment Station field
office in New Lisbon, NJ. By 1968, enough female strobili were present to
initiate controlled pollinations. The first test plantations were established
in 1971 (Little and Trew, 1979). To date, there are over 50 hybrid
plantations in several northeastern and midwestern states.
Although there currently is a strong demand for hybrid seedlings, the
controlled pollination technique used to produce the hybrid seed is far too
costly for mass production. Reforestation with hybrids on a large scale is
dependent upon development of economical mass production techniques.
Supplemental mass pollination is one of several techniques under
consideration.
1/ Instructor and Assistant Professor, Rutgers University, Department of
Horticulture and Forestry, New Brunswick, NJ 08903.
2/ Geneticist, Northeastern Forest Experiment Station, USDA Forest Service,
Durham, NH 03824. Funding for this study was provided, in part, by a
fellowship grant from the Jessie Smith Noyes Foundation, through Rutgers
University’s Center for Coastal and Environmental Studies, Division of
Pinelands Research. Special thanks are extended to Silas Little, Fred Trew,
Francis Roesch, and Douglas Eberhardt.
187
Mistblowing experiments have been conducted at the multiclone New Lisbon
orchard for several years. Allozyme analysis of seed collected from the 1976
mistblowing. indicate that an average of 11% of the seed was hybrid (Joly and
Adams, 1983). The remaining seed was either self pollinated or was outcrossed
to other clones of pitch pine in the orchard or wild pitch pines from the
surrounding vicinity. Much higher percentages of hybrids must be produced in
order for mistblowing to become operational.
A single clone pitch pine orchard, isolated from all outside sources of
compatible pollen, mistblown with loblolly pollen can yield only two types of
seed: self pollinated and hybrid. If the pitch clone selected exhibits a
significant degree of inbreeding depression, most of the seed will be hybrid.
MATERIALS AND METHODS
All breeding work was conducted at the U.S. Forest Service field office
in New Lisbon, NJ. Controlled crosses were made in loblolly pine and pitch
pine seed orchards established in 1963-64 by the Northeastern Forest
Experiment Station and Westvaco. Each clonal orchard consists of
phenotypically superior selections planted in rows of 8-16 ramets to
facilitate controlled pollination.
Four clones of pitch pine were selected as female parents for the study.
The clones were chosen on the basis of precocious flowering and their
demonstrated ability to produce good pitch x loblolly hybrids. Two clones
each of loblolly pine and pitch pine were chosen as pollen parents. All
pollen used was fresh and tested for viability. Ortet data on the female and
male clones used are listed below.
clone # county state height age
Pitch Pine Females
62 Tompkins New York 95° 160
71 Plymouth Massachusetts 85% 114
76 Carroll New Hampshire 68° 63
79 Oxford Maine Sy? 110
Pitch Pine Males
15-54 Rabun Georgia 90° 51
16-269 Burke North Carolina 64° 40
Loblolly Pine Males
4-32 Worcester Virginia 92° 42
7=56 Williamsburg South Carolina 90° 36
Pollination bags were mounted on twenty branch tips per ramet, each with a
minimum of two female strobili (from hereon referred to as conelets). Four
branch tips per ramet were marked as open pollinated controls. Only one ramet
per female clone was used in order to avoid possible variation between ramets.
Six pollen treatments were applied to each clone. Each treatment was
replicated in four pollination bags. The pollen treatments were:
1) No Pollen: unpollinated to test for complete conelet isolation.
188
2) Self: pollen from the same clone to test for self compatibility
versus inbreeding depression.
3) Self + Loblolly: pollen from the same clone plus loblolly pollen
in a 1:1 mix to simulate the conditions in a mistblown single
clone orchard.
4) Loblolly: a mixture of two loblolly pollens to test the ability
of the pitch clone to hybridize with loblolly.
5) Outcross Pitch: a mixture of two pitch pollens to test the effec-
tiveness of the controlled pollination technique with presumably
highly compatible pollen.
6) Open Pollinated: unbagged wind-pollinated control.
All crosses were made using standard control pollination techniques.
Beginning in early May 1982, sausage casing style pollination bags were
mounted over branch tips with the aid of aluminum rings for added support.
Conelets were bagged while in stages I and II (Bramlett and O’Gwynn, 1980).
When they reached stage V, 0.50cc - 0.75ce of fresh pollen was injected by
hypodermic syringe into each bag. The conelets were treated twice at two day
intervals to bracket the period of maximum receptivity. The bags were removed
when the conelets reached stage VI.
In September 1983, the cones were harvested, They were kept separate
according to treatment and bag number. Conelet abortion was determined by
subtracting the number of cones harvested from the number of conelets
pollinated. Cones were placed in used paper pollination bags with clear
plastic on the upper side in an unheated greenhouse for drying. Cones from
clones 71, 76, and 79 opened by January 3, 1984. Cones from clone 62 were
serotinous and were opened by heating in an oven at 400 - 450 C. Seeds were
extracted from each cone by hand and tallied for each bag separately. The
number of seeds per cone was determined by dividing the number of seeds per
bag by the number of cones per bag.
As fresh pitch pine seed does not require stratification (USDA, 1974),
the seed was ready for germination. All seed was surface sterilized with 0.7
molar NaOCl for 15 seconds and placed directly on moist blotter paper in
germination trays. Two trays of approximately 60 seeds each were prepared for
the four replicates of each treatment. Seed was germinated under eight hours
of light at 300 C and 16 hours of darkness at 20° C for 14 days.
At the end of the germination period, the numbers of normal and abnormal
germinants were counted. The remaining seeds were opened to determine whether
or not they were filled. The percentage of filled seed was determined by
dividing the total number of germinants plus the number of ungerminated filled
seed by the initial number of seeds placed in each germination tray. Numbers
of filled seeds per cone were determined by multiplying the number of seeds
per cone by the percentage of filled seed. Numbers of germinated seedlings
per cone were determined by multiplying the number of seeds per cone by the
percent germination of all seed (filled and empty).
In February of 1984, 28 self pollinated, 28 loblolly pollinated, and 28
self + loblolly pollinated seedlings from clone 76 were planted in Ray Leach
super cells. (Only clone 76 was used because there were not enough seedlings
from the other clones). The seedlings were grown in a heated greenhouse under ~
supplemental lighting. In September of 1984 the seedlings were transplanted
into one gallon containers.
189
RESULTS
Clones 71, 76, and 79 did not produce any cones when unpollinated (table
1), but clone 62 developed eight cones from the initial 13 conelets bagged.
The cones were significantly smaller than those resulting from other pollen
treatments. Although normal wings developed, the seeds were small and
rudimentary, and are not counted as seeds in table 1. Abortion of self
pollinated conelets varied by clone. While all self pollinated conelets of
clones 71 and 76 developed into mature cones, in clone 79 only one of fifteen
developed. In clone 62, self pollinated conelets aborted less often than
those from the outcross pitch treatment.
The total number of seeds per cone was not affected by inbreeding
depression. In clones 62 and 71, the number of seeds per cone in the self
pollination treatment was actually greater than the number of seeds per cone
in outcross pitch pollination. (Data from clone 79 are difficult to assess
due to low sample size in the outcross pitch treatment because of damage to
three of the four pollination bags during the breeding season).
The number of filled seeds per cone resulting from outcross pitch and
loblolly pollination of clones 62, 71, and 76 was far greater than those in
the self pollination treatment. Clones 71 and 76 produced more filled seed
from outcross pitch pollination than from loblolly pollination, but clone 62
produced more with loblolly pollination. In clones 62, 71, and 76 the self +
loblolly mix treatment is intermediate between self pollination and loblolly
pollination. .
Inbreeding depression effect on germination was determined by comparing
percent germination of filled seed. In all three clones, percent germination
follows the same pattern: outcross pitch > loblolly > self + loblolly > self.
In all cases, self pollination yielded fewer seedlings per cone than the other
pollen treatments. The self + loblolly pollen treatment always produced more
seedlings per cone than self pollination but fewer than loblolly pollination.
As of June 1985, seedling survival is 93% for the hybrids, 54% for the
selfs and 61% for the self + loblolly pollinated seedlings. The hybrid
seedlings are an average of three times taller than the selfs. There is no
overlap between the two groups as the shortest hybrid is still taller than the
tallest self. The self + loblolly pollinated seedlings have segregated into
two populations, fast growing and slow growing. The average height of the
fast growing seedlings is equivalent to that of the hybrids and the average
height of the slower growing seedlings is equivalent to the of the selfs.
DISCUSS ION
Use of inbreeding depression to facilitate production of hybrids is
certainly not a new idea. It is widely used in crop breeding and its
application to forest genetics was discussed by Wright in 1976. The effects
of inbreeding depression can be used to increase the percentage of pitch x
loblolly hybrids in various ways. In the four clones of pitch pine tested,
self pollinated cones always had fewer filled seeds that loblolly pollinated
cones. Although southern pine pollen is capable of self pollination and
fertilization (Bramlett, 1981), embryo collapse is more likely to occur when a
pitch pine ovule is self pollinated than when loblolly pollinated. Thus there
190
¢°66 G*L6 T2cOr © :SGt
7-9 ¢°96 G(s T° 06
9°SS 0°46 €eLS OSS iet
We) LEGS CL 0°OOT
8°OL 0°26 ONEL € 18
TES €°S6 6° 6S ‘se
8° LC 8°L8 LEALG c°8L
5 =GG TaGes OMG eGy)
¢*8T O°CL L£°S?. Ea35
Qeavey 1°S6 OES L°C8
0-59 6°96 ee9 Gales
LQG 6°76 SEG ¢°L8
C* 71 GO Oeil tse
73.6 1°69 9°€1 G*c6
PS G28 G7? O°9TT
€°8z 7° 76 90Gs ee C168
O°€S 8°28 7209 OF 041
b> Sue 0°78 LESES 5 SOK
eG Gs WAS 8°76
= = = ey 0°O
auoj ited Pees PeTTtTA jo euoj ited euo) aod
squeutwies uoT}euWwIEeD) ¥Y sSpees peTita spoes
91 LT
I 7)
L el
0 ZI
I SI
0 1
a ZI
ZI ZI
ZI ZI
1 GT
81 91
0 81
Sr a
ET LT
91 91
02 02
va 41
0 SI
LT QT
Ge at ZI
9 Z1
9 eT
Tet 91
8 €1
peanjweQ, pe ieuUrT[od
SjaTouo)
-sauo)
peweUTT[oOg uedo
YO4Td Ssoird4noC
ATTOTQOT
ATTOTGOT + F19S
3T9S
u2eTTOd ON
poqweur{T {tog usdo
yo4tq ssoi93ng
ATTOTQOT
ATTOTQOT + JTS
3T9S
ueTTOg ON
poqeurtT Tog usedo
YdITqd Ssoi93NnO
ATTOTQOT
ATTOTQOT + FT2S
3T°S
UeTTOd ON
poqeurtpTog uedgo
YoqtTg ssoi93ng
ATTOTQOT
ATTOTGOT + F12S
Sees
usxsTTOd ON
Juew esi WaT Tod
6L
Me
Wee
c9
# 9Uu0TO
SUOTIEUTTTOd peTpToaquos jo AxzewuNSs--‘], eT qe]
191
should be more hybrids than selfs in mistblown seedlots (assuming equal pollen
volumes).
All four clones of pitch tested had lower seed germination upon selfing
than with loblolly pollination. Reduced germination of selfed seeds in the
nursery seedbed will further increase the percentage of hybrids.
Inbreeding depression in survival and growth of selfed seedlings, as seen
in clone 76, indicates yet another means of increasing the percentage of
hybrids. As the seedlings grow in the seedbed, there is competition for
light, nutrients, water, and space. Many of the smaller self pollinated
seedlings will suffer and die as a result. In addition, many of the surviving
selfs can be rogued out of harvested mistblown seedlings during standard
nursery grading operations. In our greenhouse experiment, it appears that all
of the selfs could be rogued out of the self + loblolly seedling group.
It is impossible to accurately estimate percentages of hybrids which
could be produced by mistblowing single clone orchards based solely on data
from this experiment. The actual percentage of hybrids produced will depend
upon the degree of inbreeding depression expressed by the orchard clone,
compatibility of the orchard clone to the pollens used, the amount of pollen
mistblown, and the timing of the application. Even so, it seems safe to
predict that in clone 62, where 53.0 seedlings per cone developed when
loblolly pollinated compared to only 2.8 per cone when self pollinated, the
percentage of hybrids in mistblown seedlots would be very high.
CONCLUS ION
Results of this experiment indicate that mistblowing isolated
single-clone pitch pine seed orchards with loblolly pine pollen may be an
effective technique for mass producing pitch x loblolly hybrids.
Establishing several single clone orchards and mistblowing each with a
variety of compatible loblolly pollens would reduce problems associated with
low genetic variability. Ease of access must be weighed against adequate
pollen isolation when deciding upon the number of orchards to be located in
any particular area.
In order for this technique to become accepted, it must be cost
effective. Currently, mistblowing is being carried out operationally in
several southern pine seed orchards. Mistblowing in seed orchards may become
commonplace. as a means of alleviating pollen shortages and to circumvent
inbreeding, with increases in seed yield and variability justifying the
additional expense (Kellison, 1971) (Bridgewater and Trew, 1981). Mistblowing
single clone orchards should be equally cost effective.
While this study was designed to test the feasibility of single clone
orchard production of pitch x loblolly hybrids, the technique appears to be
readily adaptable for the production of other commercially valuable southern
pine hybrids.
192
LITERATURE CITED
Bramlett, D. L. 1981. Effectiveness of wind pollination in seed orchards.
In E. C. Franklin (ed.) Pollen management handbook, p. 10-14. USDA Forest
Serv. Handb. # 587.
Bramlett, D. L. and C. H. O°Gwynn. 1981. Controlled pollination. In E. C.
Franklin (ed.) Pollen management Bandbook, p.- 44-51. USDA Forest Serv.
Handb. # 587.
Bridgewater, F. E. and I. F. Trew. 1981. Supplemental mass pollination. In
E. C. Franklin (ed.) Pollen management handbook, p. 52-57. USDA Forest
Serv. Handb. # 587.
Joly, R. J. and W. T. Adams. 1983. Allozyme analysis of pitch x loblolly
pine hybrids produced a supplemental mass pollination. Forest Sci.
29:423-432.
Little, E. L. 1971. Atlas of United States trees. Volume 1: Conifers and
important hardwoods. USDA Forest Serv. Misc. Pub. # 1146, 304 p.
Little, S. and I. F. Trew. 1979. Pitch x loblolly hybrids: loblollies
for the north? J. of Horectty 77:709-716.
USDA. 1974. Seeds of woody plants in the United States. USDA Forest Serv.
Handb. #450, 883 p.
Wright, J. W. 1976. Introduction to forest genetics. 463 p. Academic Press
Inc., New York.
193
THEORETICAL IMPACT OF POLLEN VIABILITY AND DISTRIBUTION ON THE NUMBER
OF STROBILI TO USE FOR CONTROLLED POLLINATIONS IN LOBLOLLY PINE
D. L. Bramlett, F. E. Bridgwater, J. B. Jett, and F. R. Matthews '/
Abstract.--The number of female strobili to pollinate per
controlled cross in loblolly pine can be estimated from three
factors: (1) in vitro pollen viability, (2) pollen distribution
frequency within the ovules, and (3) the expected cone survival of
pollinated flowers. In this study empirical data on the number of
pollen grains per ovule were obtained from sampling loblolly pine
conelets 2 weeks after pollination. The frequency distribution for
the number of pollen grains per ovule was then used to develop a
nonlinear model to estimate pollination effectiveness. Adjustments
for less than the maximum number of seeds per cone and for empty
seed losses resulted in estimates of filled seed per cone for
varying levels of pollen viability. From these data, the numbers
of flowers required to produce 300 filled seed per cross are
presented for four levels of cone survival. These guidelines can
improve breeding effectiveness and efficiency for loblolly pine.
Additional Keywords: Pinus taeda, tree breeding, tree
improvement
INTRODUCTION
Controlled pollinations are a vital component of the recurrent
selection and breeding program of southern pines. Production of adequate seeds
for genetic testing requires both an effective and efficient controlled
pollination procedure. An effective program would have a high success rate for
completion of the attempted prescribed crosses. For example, it is important to
have all the cells of the mating design completed before outplanting. However,
an effective program could have a high completion rate but would not necessarily
be efficient in terms of the required resources such as the number of
pollination bags installed, the amount of pollen required, or the labor
employed.
WW The authors are respectively: Plant Physiologist, Southeastern Forest
Experiment Station, Macon, GA; Plant Geneticist, Southeastern Forest Experiment
Station, Raleigh, NC; Associate Director, Cooperative Tree Improvement Progran,
School of Forest Resources, N.C. State University, Raleigh, N.C.; and Plant
Pathologist, Southeastern Forest Experiment Station, Athens, GA.
We wish to acknowledge the assistance of Dr. M. Buford, Virginia Polytechnic
Institute and State University and Mr. W. D. Smith, North Carolina State
University in developing the theoretical model used in this study.
194
An efficient program could provide a high completion rate for attempted
crosses yet should not utilize more than the necessary amount of resources.
Some safety margin is prudent, but consistent overproduction of cones or seeds
is certainly not cost effective. It would be better to refine the breeding
program so that more crosses were completed in a given year rather than to
overproduce seeds from a smaller number of crosses.
In research studies to evaluate the factors affecting the success of
eontrolled pollinations, pollen viability appears to be the single most
important factor relating to high seed yield per cone (Matthews and Bramlett
1985). Correct timing of female receptivity and the delivery of the pollen to
the ovules are important factors but if the pollen is not of high quality, the
seed yield and seed quality may be seriously reduced.
Large-scale breeding programs must routinely deal with the subject of
pollen viability because of the impact this variable has on how aggressively
breeding efforts can be pursued. The North Carolina State Cooperative Tree
Improvement Program maintains a large, centralized pollen storage bank to
expedite the Cooperative's second-generation breeding program. The
Cooperative's program involves the sharing of the control-pollination workload
and the exchange of pollens. In order to meet the biological time constraints
imposed by the pollination season, it is necessary to use stored pollen.
| The Cooperative annually receives and processes for storage some 600 to
800 pollen lots. In the course of processing, several decisions are made that
take into account pollen viability. Moisture content and percentage of
germination in vitro are determined upon receipt of the pollen. When moisture
content exceeds 15 percent, the pollen is dried in an extractory as described by
Sprague and Snyder (1981) to a moisture content of 8 to 10 percent. This
initial drying is done to stabilize pollen quality until final preparations for
long-term storage can be completed. Subsequent drying to approximately 4
percent moisture is done by freeze-drying and the pollen is vacuum sealed in 10
ml vials for long-term storage. Germination tests (0.5 percent agar) are only
done before pollen storage. Both researca and experience indicate that properly
stored pollen suffers little loss in viability for at least several years
(Goddard and Matthews 1981).
Pollen lots with viabilities as low as 10 percent are stored when no
other pollen of higher viabilitiy is available. This approach assumes that even
at viabilities as low as 10 percent, some seed set will be achieved when
appropriate amounts of pollen and proper timing are utilized in making
controlled pollinations. The low-viability pollen will be employed where
required rather than delay breeding for a year. When sufficient pollen is
available with viability greater than 50 percent, 80 ml of each pollen lot are
put into storage (eight vials of 10 ml each). This amount is considered
sufficient to complete the anticipated crosses involving any given pollen even
when pollen viabilities approach 10 percent. When the viability for a pollen
lot is less than 50 percent, additional pollen is requested the following spring
to upgrade the inventory. As long as a cross has not been completed, the plan
is to replace or upgrade pollen inventories when the prestorage viability is
less than 50 percent.
195
The complexities of managing a large pollen bank with constantly
shifting inventories are eased with a computer-based record system. Matching
pollen availability to the availability of female strobili also requires good
records to avoid inefficiencies in breeding efforts and lost time.
This paper presents the impact of pollen viability on the effectiveness
and efficiency of controlled pollinations in loblolly pine. Empirical data were
used to construct a generalized model to generate the pollination effectiveness
at varying viability levels. The pollination effectiveness was then transformed
to the expected seed yields per cone and the number of strobili estimated to
produce 300 filled seed for progeny testing.
EFFECTIVE CONTROLLED POLLINATIONS
The pine reproductive system consists of the megasporangiate strobilus
(female flower), and the microsporangiate strobilus (male catkin). The female
flowers are in an upright position on the tips of the new vegetative growth and
are phenologically synchronized in development to be at maximum receptivity
during the peak of pollen release from the male catkins. As the wind transports .
pollen to the female flower, individual pollen grains adhere to micropyle arms
and are transported to the pollen chamber via a pollination droplet. In
controlled pollination, breeders try to simulate the natural wind pollination
process. Wind pollen is excluded with an isolation bag and the selected pollen
is injected into the bag by the breeder. Key elements of successful breeding
are (1) correct timing of pollen application, (2) providing adequate quantities
and distribution of the pollen to the flowers, and (3) maintaining a high
viability and vigor of the pollen.
Up to 10 million pollen grains may be injected into a single
pollination bag, and the female flower may have a large amount of pollen between
the cone seales, yet only a very small number of pollen grains are found in the
pollen chamber. Because only those pollen grains in the pollen chamber are
capable of producing seed, the effectiveness of the controlled pollination can
be evaluated by examining the pollen chambers and recording the number of pollen
grains present. Matthews and Blalock (1981) have described the procedures for
making the pollen chamber count and this method has been a valuable tool to
quantify the effectiveness of both controlled- and wind-pollinated pines.
By using the pollen count technique, the timing of pollen application has been
found to be more flexible than once thought. It appears that 2 days before or
after maximum receptivity (Stage 5) are nearly as effective as pollination at
stage 5 (Bramlett and Matthews 1983).
The pollinator used to supply the pollen can be an important part .of
the seed yield and many types of pollinators have been successfully used
(Bramlett and O'Gwynn 1981). Any device that effectively delivers the pollen to
the ovules can be used and may include a camel's-hair brush, wash bottle,
syringe, or cyclone pollinator. The quantity of available pollen may influence
the choice of pollinator. For very limited amounts of pollen, the camel's-hair
brush is effective. When pollen is abundant, the cyclone pollinator uses high
volumes of air and pollen to completely distribute pollen to all flowers within
the bag.
196
Best results have been achieved by using the cyclone pollination with 1
ee of pollen applied per bag. Normally, one application at the correct stage of
flower development is adequate, but up to three pollen applications per bag may
be used when flower development within the bag is widely divergent.
FREQUENCY DISTRIBUTION OF POLLEN WITHIN THE OVULES
In several controlled pollination experiments we have quantified the
pollen catch per ovule. From these studies, the pollen distribution approaches
but does not equal the average number of pollen grains per ovule for
wind-pollinated flowers. In the data presented in Figure 1, four quantities of
pollen 0.25, 0.50, 1.00, and 2.00 ce were applied to three separate female
clones in the Georgia Forestry Commission's Arrowhead Seed Orchard. Conelets
were collected 10 to 14 days after pollination and approximately 1,500 ovules
were observed for pollen counts within the pollen chamber. An analysis of
variance of the data set indicated no statistical differences for the mean
pollen count among the 0.25, 0.50, and 1.00 quantities of pollen. These three
quantities were combined in the frequency distributions shown in Figure 1A and
had a mean value of 2.22 pollen grains per ovule. In this distribution 12
percent. of the ovules had zero pollen grains. If an ovule has no pollen grains
or has pollen grains that are not viable, the ovule aborts soon after
pollination and no seeds are formed. If the ovule has at least one viable
pollen grain, however, the ovule continues development and forms a mature seed
coat unless the normal development is disrupted by destructive agents. In
Figure 1B, the pollen counts for ovules pollinated with 2.0 ce of pollen had a
mean value of 2.78 pollen grains per ovule and lower frequencies of ovules with
0, 1, or 2 pollen grains. Thus, the percentage of ovules with at least one
viable pollen grain would be expected to increase with increasing mean values
and consequently a low-frequency distribution of ovules with 0, 1, or 2 pollen
grains per ovule. For example, in Figure 1C, the frequency distribution for
wind-pollinated ovules illustrates a mean value of 3.97 with small numbers of
ovules with 0, 1, or 2 pollen grains per ovule.
Regardless of the distribution frequency of the pollen grains in the
ovules, the probability that some ovules will have no viable pollen grains
increases as the viability decreases. Thus, for each distribution function, the
empirical pollination effectiveness can be calculated by summing the
probabilities of ovules with: at least one viable pollen grain for each pollen
count class.
DEVELOPMENT OF A PREDICTIVE MODEL
If the pollen viability and distribution are known, a probability model
can be developed for varying quantities of pollen. For example, with 50 percent
viable pollen and the pollen distribution as shown in Figure 1A, the combined
probabilities would give 60 percent of the ovules with at least one viable
pollen grain. This ratio of ovules with one or more viable pollen grains to the.
total number of ovules per cone (seed potential) has been termed the pollen
effectiveness (PE) (Bramlett 1981). Therefore, the maximum predicted seed yield
for pollen lots with 50 percent viability and 2.2 pollen grains per ovule would
be PE x SEED POTENTIAL or 0.60 x 160 = 96 developed seeds per cone. Obviously
the actual yield of developed and filled seeds would be lower than the maximum
as will be discussed later in the paper.
197
A
PERCENT OF OVULES
@
= B
a
2 2.00 cc POLLEN/BAG
°
«
°
e
z
L*}
1S)
«
w
a
@
w
a
2 Cc
>
© WIND POLLINATED
wm
°
eK
2
w
o
4
w
a
NUMBER OF POLLEN GRAINS PER OVULE
Figure 1.--Number of pollen grains
per ovule for loblolly pines
following controlled pollinations
and wind pollinations (A)
Controlled pollinations with 0.50
to 1.0 ce of pollen per bag (B)
Controlled pollination with 2.0 cc
of pollen per bag (C) Wind
pollinations.
0.26-1.00 cc POLLEN/BAG
198
160
2.0 POLLEN GRAINS/OVULE
140
120
100
SEEDS
80
60
NUMBER OF
40
20
© 10 20 30 40 50 60 70 80 90 100
POLLEN VIABILITY (%)
Figure 2.--Predicted maximum seed
yield per cone, total developed
seed) pemicone, filled jseed sper
cone, and filled seed per flower
for controlled pollinatians in
loblolly pine. The predieted
values are generated from a model
with 2.0 pollen grains per ovule
and a seed potential of 160 ovules
per cone and are shown for
varying pollen viabilities.
The next development of the general predictive model was derived from
pollen count data collected from a pollination timing study in 1982. In this
study, pollen was applied to loblolly pine (Pinus taeda L.) flowers at varying
female flower development stages (Bramlett and Matthews 1983). Pollen
application before or after the date of maximum receptivitiy reduced the mean
number of pollen grains per ovule. Using this data set a nonlinear predictive
model was derived by fitting the observed data with a function described by
Richards (1959):
Pollination effectiveness = a x (isenae x VIAB) )
Where: a = 8.0402 + (53.4169 x MPC) - (7.7432 x (MPC*))
b = 0.00737 + (0.0087 x MPC)
with VIAB = Viability of applied pollen
and MPC = mean pollen count per ovule
From this predictive equation, values of pollination effectiveness were
generated for the values of the mean number of pollen grains per ovule (MPC)
ranging from 0.50 to 4.0 (Table 1). The range of MPC are realistic values that
have been observed in controlled pollination studies. Expected values in
controlled pollinations average between 2.0 to 3.0 pollen grains per ovules.
Counts approaching 4.0 have been observed with heavy applications of pollen but
would generally be expected to be the upper limit of average number of pollen
grains per ovule. Counts below 1.0 occur when low quantities of pollen are
applied or when the pollen is not applied at the correct stage of flower
development.
Table 1.--Predicted pollination effectiveness and apparent pollination
effectiveness for loblolly pine at varying pollen viabilities.
Mean Pollen Pollen Viability (%)
Grains/Ovule 10 2 0 0
-------- Pollination effectiveness--------
O25 0.036 0.083 0.146 0.192 0.214
1.0 0.0380 0.178 0.297 0.376 0.411
2.0 0.184 0. 388 0.596 0.708 0.749
3.0 0.281 0.559 0.801 0.906 0.938
4.0 O33" 0.638 0.860 0.937 0.956
----Apparent pollination effectiveness----
5 0.027 0.062 0.110 0.144 0.160
0 0.060 0.133 0.223 0.282 0.308
-0 0.138 0.292 0.447 0.531 0.562
0 0.211 0.419 0.601 0.680 0.704
0 05253 0.478 0.645 0.703 0.717
199
PREDICTED SEED YIELD FROM CONTROLLED POLLINATIONS
The PE value can be transformed to seed yields by multiplying PE value
by seed potential. In Figure 2, the predicted number of developed seed per cone
for a mean pollen count of 2.0 and varying pollen viabilities is illustrated.
The highest values are the maximum seed per cone that could be expected with the
given parameters (Maximum Seed/Cone, Fig. 2). However, not all of the ovules
with one or more viable pollen grains per ovule will continue development and
form a seed coat. Numerous causes may exist but one primary problem is insect
damage to the ovule resulting in ovule abortion before seed coat formation.
If the pollen grains have not been counted in the pollen chambers, the
apparent pollination effectiveness (APE) can be measured as the ratio of the
total developed seed to the seed potential (Total Seed/Cone, Fig. 2). APE
approximates PE when insect losses are minimal. From seed yield data when
varying quantities of pollen were applied to three loblolly pine clones, APE was
75 percent of the pollination effectiveness as determined by the pollen count
per ovule data. Thus the APE for a given tree could be expected to average 0.75
x PE (Table: 1).
A second factor in the reduction of seed yield is that not all of the
developed seeds are filled. Again, several causes of empty seed are known but
insect damage and embryonic abortion are principal reasons that only a
percentage of the total seeds are filled and thus capable of germination.
The values in Table 1 for APE can be converted to predicted filled seed
yields per cone by expanding to the seed potential for loblolly pine and then
adjusting for the percentage of total seeds that are filled. Although the
percentage of filled seeds may vary from year to year depending on the level of
insect protection and other factors, an average of 80 percent filled seed is a
reasonable number to use for seed orchard controlled poliinations. Thus the
predicted filled seeds per cone would equal:
Filled seed (FS) APE x SP x PFS
APE = Apparent pollination effectiveness
SP = Average seed potential for loblolly pine (160)
PFS = Percent filled seed (0.80)
Values of predicted filled seeds per cone for varying viabilities and pollen
distribution functions are shown in Table 2. As expected, the seed yields in
Table 2 indicate that low seed values are associated with low mean pollen counts
per ovule. These low mean values could result from inadequate distribution of
pollen to the ovule or if flowers were not pollinated at the proper time. More
important, low values of filled seed would also be expected when the viability
is reduced. For example, with a mean value of 2.0 pollen grains per ovule,
predicted filled seed yields with 90 percent viable pollen are four times the
yields when using 10 percent viable pollen. In Figure 2, yield of filled seed
per cone can be compared with the maximum seed and total developed seed per
cone.
200
Table 2.--Predicted number of filled seeds per surviving cone for control-
pollinated loblolly pine at varying levels of pollen viability.
Mean Pollen Pollen Viability %
Grains/Ovule ) 2 0 0
wee nnn eee filled seed-------------~----
0.5 3 7 14 18 20
1.0 7 17 28 36 39
250 17 37 57 67 Ti
3.0 Alf 53 76 87 90
4.0 32 61 82 89 91
NUMBER OF FLOWERS TO POLLINATE
The final factor for controlled pollination guidelines to be both
effective and efficient is to know the number of flowers to pollinate based on
the viability of the pollen source. As stored pollen viability can be
determined in vitro before installation of pollination bags, guidelines are
needed to adjust the number of bags to use per cross based on the expected seed
yield. Currently 144 seedlings are required for outplanting each cross in the
N.C. State Tree Improvement program (Talbert et al. 1981). To ensure adequate
seedlings, approximately 300 filled seeds per cross are needed.
The number of flowers required to provide 300 filled seed per cross are
estimated in Table 3. The table gives four levels of cone survival. Rarely
would 100 percent of the flowers pollinated reach cone maturity. Based on
experience, more reasonable survival rates are 50 to 75 percent of the flowers
pollinated. In Figure 2, the number of filled seeds per pollinated flower are
presented for 2.0 pollen grains per ovule at varying pollen viabilities.
Survival rates of 25 percent indicate serious problems involving insects or
other factors including poor pollination techniques. With pollen viability of
75 to 90 percent and two or more pollen grains per ovule, less than 10 flowers
would be required to produce the 300 seeds for progeny testing. This number may
be lower than is currently being used for operational breeding programs.
Obviously, it is important to be sure that adequate seed are available, and the
numbers of flowers given in Table 3 should be considered as minimum
requirements. Also Table 3 illustrates the guidelines to follow for loblolly
pine. Other species could be adjusted based on pollen counts per ovule and
differences in the seed potential.
201
Table 3.--Minimum number of loblolly pine flowers to pollinate to produce 300
filled seed per cross, at varying levels of pollen viability.
Cone Mean Pollen Polle ili
Survival rains/0O e 0 2 0
ae no. of flowers-~--------~--—----
25 percent 0.5 400 172 86 67 60
; 1.0 172 71 43 34 31
20 T1 33 22 17 17
30 45 23 16 17 14
4.0 38 20 15 14 14
50 percent 0.5 200 86 43 34 30
1.0 86 36 22 17 16
20 36 17 11 9 9
3.0 23 11 8 9 7
4.0 19 10 8 tf it
75 percent 0.5 134 Si 29 22 20
1.0 57 24 15 12 11
2.0 24 11 8 6 6
3.0 15 8 5 5 5
4.0 13 7 5 5 5
100 percent 0.5 100 43 22 17 15
1.0 43 18 11 9 8
7250) 18 9 6 5 5
3.0 12 6 4 5 4
4.0 10 5 4 4 4
ADDITIONAL REMARKS
The theoretical model for the effectiveness of controlled pollinations
with varying levels of pollen viability indicates that tree breeders sacrifice a
large amount of efficiency when low pollen viabilities are used. However, our
model indicates that only a very few female flowers per cross are required to
produce adequate seeds when correct timing, adequate distribution of pollen
within the bag, and highly viable pollen are used.
One of the untested assumptions of the model is that the ratio of
developed seeds to the maximum percentage of ovules with at least one viable
pollen grain would be the same regardless of the pollen viability. This
assumption may not be valid. Some preliminary work indicates that when pollen
viability is low, that pollen vigor may also be reduced. Pollen vigor is a
loosely defined term that currently is not measured in vitro. The effects of
low pollen vigor are reduced percentages of filled seed per cone. This
apparently is a result of pollen that germinates on the nucellus but does
complete fertilization. Thus ovules pollinated with low-vigor pollen could have
had relatively high in vitro viability but rather low filled seed yields per
cone.
202
The number of flowers to pollinate per cross does not indicate how many
bags should be used. Frequently large numbers of flowers can be enclosed within
one bag. The danger is that a broken branch would be disastrous. Therefore, a
minimum number of bags should be five or six. Also, when large numbers of
flowers per bag are used, the flower development may extend over several days,
requiring more than one pollination per bag.
Finally, the number of flowers required to complete a cross can be used
in two ways. The first way is increase the effectiveness and efficiency of the
breeding program. The second way is to evaluate current breeding results. If
large numbers of flowers are required to complete each cross, this would appear
to be a warning signal that one or more of the several components of pollination
are not favorable. Pollen viability would be the most likely item but pollen
distribution and flower mortality could also be part of the low efficiency.
Taking corrective action should mean that seed yields increase and the breeding
program becomes both more effective and efficient.
LITERATURE CITED
Bramlett, David L. 1981. Effectiveness of wind pollination in seed orchards.
In; E.C. Franklin (ed.) Pollen Management Handbook. Agric. Handb.
587. Washington, DC: U.S. Dep. of Agric. Forest Serv. p. 10-14.
Bramlett, David L., and Frederick R. Matthews. 1983. Pollination success
in relation to female flower development in loblolly pine. In:
Proc. 17th South. For. Tree Improv. Conf. June 6-9, 1983;
Athens, GA. p. 84-88.
Branmlett, David L., and Claude H. O'Gwynn. 1981. Controlled pollination. In:
E.C. Franklin (ed.) Pollen Management Handbook. Agric. Handb. 587.
Washington, DC: U.S. Dep Agric. For. Serv. p. 44-51.
Goddard, R.E. and F.R. Matthews. 1981. Pollen Testing. In E.C. Franklin
(ed.) Pollen Management Handbook. Agric. Handb. 587.
Washington, DC: U.S. Dep. of Agric. For. Serv. p. 40-43.
Matthews, F. R. and T. E. Blalock. 1981. Loblolly pine pollen grain counts by
ovule dissection. In: Proc. 16th South. For. Tree Improv. Conf.
May 27-28, 1981, Blackburg, VA: p. 276-278.
Matthews, F. R., and D. L. Bramlett. 1985. Pollen quantity and viability
affects seed yields from controlled pollinations of loblolly pine. South.
J. Appl. For. (In Press).
Richards, F.J. 1959. A flexible growth function for empirical use. J.
Exp. Bot. 10(29): 290-300.
Sprague, J.R. and E.B. Snyder. 1981. Extracting and drying pollen. In: E.C.
Franklin (ed.) Pollen Management Handbook. Agric. Handbk. 587.
Washington, DC: U.S. Dep. of Agric. For. Serv. p. 33-36.
Talbert, J. T., F. E. Bridgwater, and C. C. Lambeth. 1981. Genetics Testing
Manual, N.C. State Univ.-Ind. Pine Tree Improv. Coop., Raleigh, N.C., p. 37.
29%
A SEVEN-YEAR-OLD OCALA SAND PINE SEEDLING SEED ORCHARD
Ralph A. Lewis, Timothy LaFarge and James L. McConnell 1/
Abstract.--A 20 acre Ocala sand pine seedling seed orchard
was established in 1978 near Ocala, Florida. First year
survival was 67%. An apparent adaptive mechanism that allows
some trees to survive unfavorable environmental conditions was
noted. After 7 years and 2 thinnings, the tallest trees were
over 26 ft. and the tallest family averaged over 19 ft.
Although a few cones were observed after 3 years, the orchard
is just starting to produce a significant amount of seed.
Additional keywords: Orchard management, seed orchard design,
single-tree plot, Pinus clausa.
The theory for design and management of seedling seed orchards is
well known but few production orchards of this type have been established
in the South. In January of 1978, a production seedling orchard of the
Ocala race of sand pine (Pinus clausa var. clausa Ward) was planted on the
Ocala National Forest in central Florida.
METHODS AND MATERIALS
Orchard Design and Layout
Design.--The orchard is designed for 30,000 seedlings, consisting of
120 families with each family represented by a single tree plot in each of
250 blocks (replications). A block consists of 10 rows of 12 trees
planted on a 5 ft. by 5 ft. spacing. Actual orchard area is slightly less
than 17 acres with another 3 acres devoted to roadways and border strips.
Border rows were planted around the exterior and along both sides of all
interior roadways.
Layout.--The size and complexity of the planting required that the
layout be simple but precise. All blocks and rows within blocks were
tagged. Every planting point was marked with a wire stake flag. In order
to help guide the planters, each block was staked with a single color with
all adjacent blocks staked with different colors. This was accomplished by
alternating red and white blocks next to alternating blue and yellow
blocks. Another color (orange) was used exclusively for border rows.
1/ Forester, Eastern Zone Geneticist, and Regional Geneticist respectively,
USDA Forest Service, Region 8, Atlanta, Georgia.
204
Materials
Site.--The planting site is located in central Florida on the Lake
George Ranger District, Ocala National Forest. Soils are excessively
drained, stongly acidic deep sands of the Astatula series. The area is
part of a "longleaf island" and it has a slightly higher clay content than
the more typical sand pine sites. The topography ranges from flat to
slightly rolling. The site was prepared by removing all woody vegetation
followed by raking and double disking.
Seed.--Wind-pollinated seeds from 131 select trees (all growing in
wild stands on the Ocala N. F.) were collected in 1976. Sufficient seed
from each collection to produce at least 250 healthy seedlings were planted
in April of 1977 in the Chipola Experimental Forest nursery near Marianna,
Florida.
Seedlings.--The early care and culture of the seedlings were routine
but in late summer, both moisture and fertilization were gradually reduced
in order to induce hardening-off. Although this procedure tended to
produce slightly smaller seedlings, it prepared the seedlings to better
cope with planting shock. Lifting began during the first week in
January ,1978. Trees were individually tagged with family identification,
sorted into "block" bundles (one tree each of 120 families), and packed in
kraft bags. After transportation to the vicinity of the planting site,
the bags were stored under refrigeration until planted.
Planting and Mapping
Planting.--Each block was planted by a single two man crew using a
standard dibble. Randomization of families in a block was obtained by
planting the seedlings in the sequence they were removed from the bag
(trees were throughly mixed during packing) and by starting the planting
of each block on a row picked at random.
2 Mapping .--Mapping of each block was done as quickly as possible after
planting. The family identity of each seedling was recorded from an
attached paper label along with the planting location (row and point
within row). These data were double checked in the field and edited by
computer for duplicate family numbers within block.
RESULTS
~ Survival
First year.--Two survival counts were made during the first year.
The first count was done in May on about 26% of the orchard. Overall
’ Survival was estimated to be 66.5%. Both block and family survival
205
appeared to vary greatly. In October, a complete inventory counted 20,047
live trees for a survival rate of 66.8%. Individual families ranged from
27% to 86%. Of the 120 families, 24 had survival rates of 75% or more
while 8 had rates less than 50%. Block survival ranged from 3% to 89%.
The pattern of survival (by block) indicated that some mortality was not
random. Of the 20 blocks in the orchard with survival below 40%, 16 were
located in the southwest corner of the orchard. An exact cause for the
poor survival in this area could not be determined.
Third year.--Additional mortality between the first and third year
was very small. Survival was 64.2% for a net loss of 794 trees.
Growth
_ Third year.--Height growth for individual trees varied from less than
1 foot up to 14 feet. Mean height for the entire plantation was 7 feet
with family means varying from 4.6 to 9.4 feet.
Seventh year.--Height growth continued at a rapid rate for most
remaining trees. Mean height was 19.5 feet with some trees exceeding 26
feet tall..Individual family means ranged from 17.9 to 20.9 feet. Mean
diameter (d.b.h.) was 2.9 inches.
Apparent Survival Mechanism
Description.--During the analysis of the first-year survival checks,
inconsistencies were found in the data that indicated live trees were
growing in spots previously tallied as dead. At first these were thought
to be simple recording errors but a field check indicated the data was
correct. Close examination of the seedlings revealed that they probably
appeared. to be dead during the early survival check because all foliage
had turned brown and dropped off, leaving only the naked stem. At some
time later, a new sprout appeared near the top of the old stem and took
over as the terminal shoot. By the time of the second survival check,
many of these "dead" trees appeared almost normal.
Frequency of Trait.--Since the first survival check only examined
26% of the plantation, it is not possible to determine the full extent of
this trait. Comparison of data between the two first-year checks and the
third-year check indicated that about 1.5% of the trees were involved.
All families except one contained at least one tree that had the trait.
The maximum frequency of occurence within family was 4%.
Growth.--Generally, height growth of the trees exhibiting this trait
was inferior to other trees of the same family. Also, these trees were
usually shorter than their neighbors in the block. There were a few
exceptions where trees displayed outstanding growth and these exceptions
seem to be concentrated in a few families.
206
Cultural Practices
In the first three years after planting, cultural practices were
limited to spot control of competing vegetation and other activities to
maintain the health and safety of the orchard. After the third-year
measurements were analyzed, the first thinning was done. All stunted,
deformed or badly overtopped trees were removed. Trees were also removed
in order to improve spacing but care was taken to retain a good
representation of all families. A subsequent thinning in 1984 removed
more trees for spacing and rogued the four worst (in terms of height
growth) families. This reduced the tree count to 7,400. Another thinning
is currently underway to rogue several more of the poorest families based
on combined orchard and supplemental test data.
Seed Production
Sand pine is well known as an early and prolific seed producer. By
the third year after planting, several cones were observed on scattered
trees. A majority of the trees were producing a few cones at the fifth
year. The 1984 cone crop was estimated at 10 bushels per acre. Any
serious effort to collect these cones has been precluded by the relatively
dense and irregular spacing of the trees.
DISCUSSION
Successful establishment and maintenance of a large seedling seed
orchard involve complex and demanding tasks that must be done at the
proper time. Orchard development will be very rapid for a species such as
sand pine. Frequent inventories and measurements are necessary in order
to plan for thinnings and roguing. These data can also provide additional
insight into the species and the population being grown.
For example, Ocala sand pine is difficult to transplant.
Nevertheless, our data indicate considerable variation in survival by
family. This information should be of value for both future testing and
regeneration.
The apparent survival mechanism displayed by a small number of trees
is interesting but it will probably have little influence in this orchard.
Poor height growth by most of these trees has resulted in their removal
during the first or second thinning. On the other hand, knowledge of the
existence of such traits may be useful when selecting for tolerance to
unfavorable site conditions.
A seedling seed orchard will never be as neat and tidy as its clonal
counterpart. The close spacing during early development makes any
vegetation management practices very difficult. The manager should be
prepared to do much of the early thinning by hand. Mortality and removals
due to thinnings are almost always randomly spaced so mowing and cone
collection may be hampered for several years.
207
MONITORING CONEWORMS WITH PHEROMONE TRAPS: A VALUABLE PEST
DETECTION PROCEDURE FOR USE IN SOUTHERN PINE SEED ORCHARDS
J. C. Weatherby, G L. DeBarr, and Le R. Barber
Abstract.—-Sticky traps baited with synthetic pheromone
were used to detect the presence of the webbing coneworn,
- Dioryetria diseclusa Heinrich, during 1981-1984 in southern
pine seed orchards. The blister coneworm, PD. clartoralis
(Walker), and the loblolly pine coneworm, D. merkelt Mutuura
and Munroe, were also trapped during 1982-1984. More than 80
orchards were surveyed in 1984. Trapping data indicate that
outbreak populations of D. disclusa present during 1981 in
eastern North Carolina and eastern South Carolina declined
each year, while populations remained relatively stable or
increased in orchards in central Georgia, Alabama and
Mississippi during this same period. Trap catches of
D. merkelt remained high (25-49 males/trap/year) or very
high (550 males/trap/year) during 1984 in 57% of the orchards
which detected high or very high catches in 1983. Trap catches
of D. clartoralis remained relatively stable during the 3 year
trapping period. Regional seasonal activities of the insects,
and pest management decisions for individual orchards are
discussed.
Keywords: 1). disclusa, D. merkelt, D. elarioralis, IPM,
Dioryctrta.
Several species of coneworms, Dioryctria sppe, are considered key pests
damaging cone and conelet crops in southern pine seed orchards (Ebel et al.
1980). Prior to 1981 blacklight traps were occasionally used by orchard
managers to detect and monitor coneworm populations (Yates and Ebel 1975).
This procedure was time consuming and accurate identifications of the various
coneworm species were difficult.
The discovery (DeBarr and Berisford 1981) and the identification of the
sex pheromone produced by female D. disclusa moths for the purpose of
attracting males (Meyer et al. 1982) led to the development of a highly
specific bait which could be used to monitor D. diselusa populations. In 1981
traps baited with synthetic sex pheromone of the webbing coneworm, D. disclusa
were installed in 63 pine seed orchards (DeBarr et al. 1982). This survey
demonstrated “the value of pheromone-baited traps as part of an integrated
pest management approach to coneworm control in southern pine seed orchards.”
Results of the 1981 survey and research aimed at identifying the pheromone
produced by D. cGlarioralis females (Meyer et al. 1984) showed that both
D. Clarioralis and D. merkelt males were frequently attracted to bait with the
single chemical component of the D. dtsclusa pheromone. Hanula et ale (1984a)
demonstrated that traps baited with D. ditsclusa pheromone could be used to
detect the presence of 3 coneworm species — D. dtsclusa, D. clartoralts and
D. merkeli. Because of the success of the 1981 survey and the discovery of
cross attractancy, the 1982 survey was expanded to monitor populations of all
three of these insect pests.
208
Survey cooperators in industrial, state and Federal forestry organizations
are compiling historical data files for their orchards which are also being used
to detect regional changes in yearly trap catches. Comparisons of trap catches
from individual orchards are used to locate potential “hot spots” of activity.
This information provides the orchard manager with an early warning system to
indicate years and locations where coneworms are likely to cause substantial
damage. Interpretations of trap catch data are also helping to define the
regional flight periods and the population phenologies of the major coneworm
species. This information is expected to increase the effectiveness of sprays
timed to suppress adults or newly hatched larvae.
METHODS AND PROCEDURES
The coneworm survey is a cooperative effort between the USDA-Forest Service
and cooperating orchards. Entomologists with Region 8, Forest Pest Management
(FPM) and the Southeastern Forest Experiment Station supply cooperators with
instructions, data sheets, and pheromone baits. Cooperators are responsible for
obtaining, installing and checking traps. Orchard managers are also requested to
submit their data to FPM for use in summaries and regional interpretations of
trap catches.
Baits and Traps
Rubber septum dispensers impregnated with 100 pg of 98% pure
(Z)-9-tetradecenyl acetate (Z9-14:AC) dispensed in 5 ul carbon disulfide are used
as baits. A seasonal supply of baits is mailed to each cooperator. Baits are
stored at below-freezing temperatures prior to use in the field. Either of two
commercially available traps, the Pherocon 1C (Zoecon Corp., Palo Alto, CA) and
Sentry wing trap (Albany International Corp., Needham, MA) are effective. In
1981, six traps were used in each orchard (DeBarr et al. 1982). Since then, the
recommended procedure is to install 10 traps in a selected seed source at each
orchard site. The traps are assembled and 1 bait is placed in the center of each
trap bottom.
Trapping Procedures
Trapping locations are randomly selected on a grid drawn to overlay a map of
the entire trapping area. The tallest tree in the general vicinity of each
trapping location is selected (Hanula et al. 1984c). Selected trees are to be at
least 90 feet apart.
Traps are placed in the top 10 feet of the tree crown (Hanula et al. 1984b).
Traps are hung by running a nylon cord from the ground, through a wire loop which
is attached at the designated height to a branch, and back to the ground (DeBarr
et ale 1982). Both ends of the nylon cord are attached to the trap so that the
trap can be raised and lowered from the ground.
The traps are checked once or twice a week. The numbers of moths of each
species caught in each trap are recorded, and the moths are removed. Baits are
changed the first of each month. Trap bottoms are changed as needed.
Traps are deployed in late March or early April and trapping continues
through mid-November.- Trapping is discontinued in late October or November when
no moths are caught for 2 weeks at any trapping location. Data are submitted to
FPM monthly.
209
Reporting and Summary Procedures
The Southern Region was divided into 3 areas based upon average daily
January temperatures (Fig. 1). All cooperating orchards located below the 50° F
isotherm are placed in Area I. Orchards located between the 45-50° F isotherm
are in Area II. The remaining orchards are in Area III.
NY AREA I
AREA Ill
Figure 1.--Pheromone survey areas within the Southern Region.
Each cooperating orchard receives a monthly summary which ranks the total
trap catch of each species for all orchards within each area. Also included in
the monthly summary are 2 sets of 3 graphs. One set compares the orchard trap
catch for the current year with the area trap catch for the previous year on a
weekly basis. The other set of graphs compares the weekly trap catches for the
orchard and the area during the current year. \
RESULTS AND DISCUSSION
Interpretations of Monthly Reports
During the last week of every month each orchard Manager receives a survey
summary for the previous month. A case study of the trap catches at Orchard #101
illustrates how a typical monthly report might be used by a seed orchard manager
in developing a pest management strategy.
Orchard #101 is located in central Alabama above the 50° F isotherm. A
monthly report for Orchard #101 summarizes the trapping data from the orchard
with data from other orchards within Area II. Table 1 lists the orchard rankings
by total moth catch for orchards located in Area Il. During the 1984 trapping
season, Orchard #101 captured 105 D. clarioralis, 17 D. disclusa and 149
D. merkelt. While the absolute relationship between total trap catch and
coneworm impact is unknown, orchard managers are encouraged to compare their
data with data from other similarly managed orchards. Based on these relative
comparisons, a sizeable ). clarioralis population was present in 1984 at
210
Orchard #101 and the potential for significant losses was anticipated. Trap
catches of D. disclusa detected the presence of a minimal population in 1984 and
suggested that damage during 1985 caused by D. disclusa should. be low.
Approximately 40% of the orchards located in Area II caught more D. merkeli
males in 1984 than did Orchard #101. However, the relatively large D. merkelti
trap catch (149 total) indicated that populations were present and this species
should probably be considered in the pest management program.
Table 1.--Ranking by total moth catch for Orchard 101 compared to other
orchards in Area IL (March - November 1984)
No. of No. of No. of
Rank © Orchard DC Orchard DD Orchard DM
i 111 182 130 249 61 402
2 101 105 1S 227 63 380
3 109 We) 115 187 103 375
4 128 63 55 149 123 267
5 13 62 103 97 14 236
6 105 58 116 93 LY 201
7 54 Sif 14 78 LS 198
8 US 48 114 66 114 191
9 55 43 105 58 116 163
10 NW AZ72 40 61 56 138 155
11 130 38 109 44 101 149
12 138 31 63 35 oy) 141
13 115 16 iL7/ 618} 128 137
14 14 14 124 29 13 101
15 116 11 28 27 109 94
16 123 8 ime! 22 122 73
17 103 4 ay 22 28 40
18 HS) 1 101 b7/ 126 5
In addition to the ranking tables a typical monthly report contains 2 sets of
graphs. One set of graphs (Fig. 2) plots the total weekly trap catches of
D. clartoralts from all orchards located in Area II for the 1983 season.
Superimposed upon the area graph is a plot of the total weekly trap catches of
D. clarioralis from Orchard #101 for the 1984 season. This graph summarizes the
seasonal flight patterns for the previous year, and by studying it the orchard
manager can anticipate when current flights might begin, peak, and end. Similar
graphs plotting trap catch data for D. disclusa and D. merkeli would be included
in a typical monthly report.
211
|
!
|
So SLAwmRa SRA ss =>
AREA—198323 vs. OROCOHARD—-1984
| i 20 LEGEND
1 o
2 1983
| -
| area ee nr ee NGS PLES OOM ee NINE TUN SETS sen tea | eR ES ON ae es 1984
aE
| Oe
S
| x
\ iy <
=
o
| si
m
-
eran
f < |
oo
[eae
ex a
x
i 77. 91 105 119 133 147 161 175 189 203 217 231 245 259 273 287 301
a
| JULIAN DATE
Figure 2.--Total trap catch for 1984 from Orchard #101 superimposed upon the
total trap catch for all orchards in Area II which submitted data in 1983.
The other set of graphs (Fig. 3) displays the total weekly catches of
D. clarioralis from Orchard #101 and the total weekly catches for all the
orchards in Area II for the 1984 trapping season. Orchard managers can determine)
if the flights detected in their orchard correspond to the area flights. Similar)
graphs plotting the trap catch data for D. diselusa and D. merkelt would also be |
included in a typical monthly report.
- u
rr EN ee ee EN Ee
| oo A REAT1SSs4 _ vs. ORCHARD—1984
180
160
140
120
100
LEGEND
AREA
Sern ORCHARD
TOTAL. TRAP CATCH
T AXIS—-AREA RT AXIS—-ORCHARD
JAN SAgie
=
|
\ fe
Figure 3.--Total trap catch for 1984 from Orchard #101 superimposed upon the
total trap catch for 1984 from all orchards in Area II.
212 |
Pest Management Decisions Based Upon Trap Catches
Early Warning System
Seed orchard managers are using trap catches to detect the presence of
D. clarioralis, D. disclusa, and D. merkelt. The relative importance of each
member of the coneworm pest complex varies by orchard and season. Although, the
exact relationship between trap catch and potential damage is unknown, population
trends are also evident. Table 2 lists trap catches from 2 orchards with dif-
ferent coneworm complexes. Trapping data from Orchard A indicate that
D. dtselusa and PD. merkelt are the most important pests. At Orchard B, the major
coneworm is D. clartoralis and pest. management efforts should be aimed at
suppressing this species. Based upon our trapping experience and observations
the following arbitrary scale of seasonal trap catches has been developed in
order to rank the relative potential for cone attacks by each species: very
high, >50 moths/trap; high, 25-49 moths/trap; moderate, 10-24 moths/trap; low,
1-9 moths/trap; very low, <l moth/trape Each season, seed orchard managers use
this arbitrary scale to determine which species pose the greatest threat to next
years cone crops Management decisions are modified in order to target control
actions for major pest species at each location.
Table 2.--Total trap catches for 1982-1984 at Orchard A
(Perry Coe, MS) and Orchard B (Geneva Co, AL)
Orchard Year D. clartoralis D. dtsclusa D. merkelt
1982 27 323 230
A 1983 By t Hi 415 1
1984 52 76 1
1982 120 4 155
B 1983 227, 3 Zi
1983 25 0 18
In addition to identifying the coneworm complex found within each orchard,
trapping data indicate population trends over time. Figure 4 illustrates two
different population trends which have been detected with pheromone baited traps.
Orchard C, located in Beaufort Coe, NC had a D. disclusa outbreak which peaked in
1981. Since 1981 the trap catch has steadily declined. Using the previously
described population scale, the 1984 population level can be classified as low
(1-9 moths/trap). Therefore the predicted potential cone loss for 1985 should be
minimal. Pest management actions designed to control D. disclusa in 1985 are
probably not necessary. Orchard D, located in Putnam Co., GA has also had a
population outbreak of D. disclusa; however trap catches have remained high
(25-49 moths/trap) or very high (550 moths/trap) during the last 5 years. Trap
catches indicate a steady increase from 1982-1984. Therefore, pest management
actions targeted for D. disclusa are advisable at Orchard D.
213
00
100
1
ORCHARD C ORCHARD D
90
BEAUFORT CO. NC PUTNAM CO. GA.
80 90
80
70
50
oO’ D.DISCLUSA / TRAP
$0 60 70
oo. DISCLUSA/ TRAP
60
40
40
X NUMBER OF
30
X NUMBER OF
30
20
20
10
10
80 81 82 83 84
Figure 4.--Mean numbers of male DPD. disclusa per trap captured during 1980-84 in
two loblolly pine seed orchards.
Seasonal Activity
Coneworm development and population phenologies are controlled predominantly
by temperature accumulations (Hanula 1984b). Temperature accumulations vary at
each orchard and, as a result, moth flights often begin, peak and end on dif-
ferent dates at different trapping locations. Peak trap catches for D. disclusa
in the more northern orchards are 3 to 4 weeks later than the most southern
orchards (Fig. 5). Figure 6 is “ << te
a graph of weekly total trap
catches of D. disclusa at 3 dif-
ferent orchards. Orchard E is
located in Washington Parish, LA
(Area 1); Orchard F is located in
Webster Parish, LA (Area II); and
Orchard G is located in Murray
Co... GA (Area, DET). bine 19845
flights of D. dtsclusa males
began on Julian dates 141, 162,
and 169 at Orchard E (Area I),
Orchard F (Area II), and Orchard
G (Area III), respectively.
Figure 5.--Peak catch dates
(month-day) for Dtoryetria
dtsclusa males during 1981
(From DeBarr et ale 1982).
214
oO. Or Ssectklousa
LEGEND
epehteliss(: ORCHARD E
Somes ORCHARD F
ORCHARD G
20
TOTAL TRAP CATCH
6s 99 113 127 141 155 169 163 1987 211 225 239 253 267
JULIAN DATE
Figure 6.--Total weekly trap catchs of D. disclusa for 1984 at Orchards E
CArea))s Ff (Area Il), tand .G (Area EID).
The population phenology of D. merkelt is also affected by temperature
accumulations. Unlike D. disclusa which flies in late spring, D. merkeli flies
in late summer and fall. With the onset of cooler temperatures and shorter day
lengths typical of late August, September and October, adult flights of
D. merkelt in more northern orchards occur earlier than flights in southern
orchards (Fig. 7). The earlier flight is a survival mechanism which insures that
adult emergence, mating and egg development occur before low temperatures prevent
normal behavioral activities and development. Figure 8 is a graph of the weekly
total trap catches of D. merkeli at 3 orchards. Orchard H is located in Monroe
‘Co., AL (Area IL); Orchard I is located in Greene Co., AL (Area IIL); and Orchard J
is located in King William, Co., VA (Area III). In 1984 the adult flights of
D. merkelt began on Julian dates 267, 246, and 225 at Orchard H (Area I), Orchard
I (Area IL) and Orchard J (Area IIL), respectively.
Figure /.--First catch dates (month-day) for Dioryctria merkeli males during
1984 (Only orchards which caught 50 or more males/season and trapped through
Julian day 270 are included).
215
TOTAL TRAP CATCH
JULIAN DATE
Figure 8.--Total weekly trap catches of D. merkeli for 1984 at Orchards H
(Area® 139 (AreaY ED), and® J> (Area® Lit).
For the present, pheromone traps provide the orchard manager with local
seasonal activity patterns for the major coneworm species. Research is underway
to develop degree-day prediction models for use with pheromone traps.e In the
future, temperature accumulations will be set to coincide with the beginning of
moth flight and insecticide applications will be timed for maximum effectiveness
against susceptible lifestages.
N
Regional Interpretations
The regional population levels for the 1981-1984 D. disclusa survey are
shown in Figure 9. Each dot on the map represents the total seasonal catch per
trap for each cooperating orcharde Dot size indicates the relative population
classified according to the previously described population scale. In 1981
outbreak populations were detected at several orchards along the east coast in
North Carolina, South Carolina, and Georgiae By 1984, populations in this area
had declined. However, “hot spots" were detected at several orchards scattered
throughout the south. Figure 10 shows regional trends for orchards which
increased or decreased one or more population classes. Comparisons between
catches in 1982 and 1983 indicate increasing population trends in Mississippi and
Alabama. A similar trend was apparent in comparisons between catches in 1983 and
1984. The majority of the orchards which trapped moderate or high populations in
1984 detected increasing population trends from levels detected in 1983. Orchard
managers are encouraged to consider both the relative population size and trend
when developing pest management strategies.
216
“P861-1861 SUTANp uoTRRPDOT pAzeYyoIO YORe Je suOoWwcAsYd xes oTAeyIUAS
YITA petted (OT=N) sdea3z OT uosoreyg 10z deaq aed AYysneod vsn7oOS1p °*qg aTeM JO sisqunu sx8e12xAy—°*6 oaNn3Ty
8
ai9
6b -S%
|
|
|
|
diacer
| °
|
17
2
)
~ St
TRAP CATCH
Oo LOW
@ uP
Oo DOWN
GLD
Figure 10.—-Regional population trends for pheromone trap catches of D. disculsa
from 1982 to 1983 and 1983 to 1984. (Up and down arrows reflect a change of one
or more population classes). :
Similar regional population maps have also been drawn for D. merkelz and
D. elartoralis. Catches of D. merkelt for orchards which trapped through Julian
date 270 indicated high to very high populations in southwestern Alabama, eastern
North and South Carolina, and Virginia during 1984 (Fig. 11). Scattered “hot
spots” also occurred at several other locations. In contrast, only two locations
had high trap catches of D. clarioralis during 1984 (Fig. 12).
oO MOTHS/ TRAP
A EIA
41
1-9
10-24
25-49
6
e
8
a >50
Figure 11.-—-Average numbers of male D. merkelt caught per trap for Pherocon 1C
traps (N=10) baited with synthetic sex pheromone at each orchard location
during 1984.
218
o MOTHS/TRAP
41
1-9
10-24
25-49
>50
Figure. 12.--Average numbers of male D. clartoralis caught per trap for Pherocon
1C traps (N=10) baited with synthetic sex pheromone at each orchard location
during 1984.
CONCLUSIONS
Data from the southwide coneworm survey are providing seed orchard managers,
pest management specialists, and researchers with invaluable information.
Orchard managers are using historical data files to make informed decisions
concerning pest management strategies targeted for specific key pests. In
addition, survey data are helping to define regional population trends and pest
phenologies.
Efforts are being made to improve the quality of the survey data. Species
identification causes confusion and occasionally errors have been detected.
Incomplete data sets are fairly common particularly for fall flights of
D. clartoralis and D. merkeli. Orchard managers are encouraged to continue
trapping throughout the fall emergence period. Research concerning trapping
procedures (Hanula et al. 1984c) indicates that minor deviations from suggested
procedures, particularly trap height, can cause considerable variations in
trapping data. Data interpretation for individual orchards continues to be a
major problem. Efforts are being made to decrease the turn-around time so that
orchard managers can use current data to adjust pest management strategies.
Despite a few minor problems, the southwide coneworm survey has been an
extremely successful and valuable regional effort. In the future the survey will
continue to function as an early warning system, alerting seed orchard managers
to potential problem species. Field tests with a synthetic pheromone for the
Southern pine coneworm, D. azatella (Hulst.), have been completed (Meyer et al.
1985) and baits for this species were deployed at 20 locations throughout the
South. Research efforts are being directed toward gaining a better understanding
of the relationship between relative trap catch and damage potential, as well as
developing timing systems for insecticide applications using trapping data and
temperature accumulations.
219
LITERATURE CITED
DeBarr, Ge Le; Le Re Barber; C. W. Berisford and J. C. Weatherby. 1982.
Pheromone traps detect webbing coneworms in loblolly pine seed orchards.
Soe Je Appl. Fore 6: 122-127.
DeBarr, G Le and C. We Berisford. 1981. Attraction of webbing coneworm males
to female sex pheromonee Environ. Entomol. 10: 119-121.
Ebel, Be He, Thomas H Flavell, Loyd E. Drake, Harry 0. Yates III, and
Gary Le DeBarre 1975. Seed and cone insects of southern pines. USDA For.
Serve Gen. Tech. Rep. SE-8, 40 p.
Hanula, James Le, C. Wayne Berisford, and Gary L. DeBarr. 1984a. Pheromone
cross~attraction and inhibition among four coneworms, Dtoryctria spp.
(Lepidoptera: Pyralidae) in a loblolly pine seed orchard. Environ. Entomol.
13s 1298-1301.
Hanula, James L., Gary L. DeBarr, and C. W. Berisford. 1984b. Oviposition
behavior and temperature effects on egg development of the southern pine
coneworm, Dioryctria anatella (Lepidoptera: Pyralidae). Environ. Entomol.
13:6 1624-1626.
Hanula, James Le, Gary Le DeBarr, William M. Harris, and C. Wayne Berisford
1984c. Factors affecting catches of male coneworms, Dtioryctria spp.
(Lepidoptera: Pyralidae), in pheromone traps in southern pine seed orchards.
J. Econ. Entomol. 77: 1449-1453.
Meyer, Wendy L., Gary Le DeBarr, C. Wayne Berisford, Larry R. Barber, and
Wendell Le Roelofs. 1982. Identification of the sex pheromone of the
webbing coneworm moth, Dioryctria ditsclusae Environ. Entomol. 11: 986-988.
Meyer, Wendy L., Gary Le DeBarr, James H. Hanula, Boris Kovalev, Re Scott
Cameron, C. Wayne Berisford, and Wendell Le Roelofs. 19 .
Z-11 hexadecenylactetate, a sex pheromone for the southern pine conewormn,
Dioryctria anatella (Lepidoptera: Pyralidae). Environ. Entomol. (In press)
Meyer, Wendy Le, Re Scott Cameron, Ashok Tamhankar, Gary DeBarr, C. Wayne
Berisford, and Wendell L. Roelofs. 1984. Sex pheromone of the blister
coneworm moth, Dioryctria clartoralis (Lepidoptera: Pyralidae). Environ.
Entomol. 13: 854-858.
Yates, H» O., ILI, and B. He Ebel. 1975. Light-trapping and identifying
Dioryetria that damage pine cones in northeastern Georgia (Lepidoptera:
Phycitidae). Je Ga. Entomol. Soc. 10(1): 78-86.
220
SELECTION POTENTIAL FOR CONEWORM AND SEED BUG
RESISTANCE IN LOBLOLLY PINE
SEED ORCHARDS
George R. Askew, Roy L. Hedden, and Gary Denar
Abstract.--Current seed orchard practices rely heavily on
pesticide control of cone and seed insects. Advanced genera-
tion orchards will have a greater need for control as seed
becomes more valuable. However, as environmental concerns
about pesticide use change political pressure may force
changes in management practices. Increasing costs of pesti-
cides and the need for new formulations may make the use of
inherent resistance breeding seem more plausible. Intensive
breeding programs utilizing decreased generation lengths can
be adapted to allow for the inclusion of several resistant
parents. Levels of resistance within the orchards can be
increased slowly, a little each generation, and coupled with
selective spraying regimens without great disruption of
current management schemes. Advanced generation selections
made without regard to infestation potential, perhaps because
pesticide use has virtually eliminated the problem in the lst
generation orchard, may result in a 2nd or 3rd generation
orchard with an infestation potential that has increased be-
yond the control capabilities of available pesticides.
- Additional keywords: Pesticide, Dioryctria, breeding
Coneworms (Dioryctria spp.) and seed bugs have been recognized as
serious threats to seed orchard production (Ebel et al. 1981) with damage
estimates as high as 90Z in some untreated southern pine orchards.
Dioryctria amatella, D. disclusa, D. merkeli, Leptoglossus corculus and
Tetyra bipunctata are the main problem pests for southern pines. Coneworms
present a major problem for seed orchard managers because the entire cone is
destroyed. Seed bugs destroy individual seeds and hence even though damage
may be extensive some seed may be salvageable from infested cones. Together
they constitute a management problem that is currently dealt with by spray
applications of insecticides such as fenvalerate. The commercial value of
seed orchard seed, particularly those from advanced generation orchards,
warrants every effort to minimize insect predation.
Even in orchards with great amounts of damage, the pattern of
infestation is not consistent among all trees. Some trees are virtually
eliminated as seed producers while others receive no appreciable damage.
Differentiation of heavily infested and lightly infested trees has been found
to be associated with clonal affiliation in some orchards, Askew et. ail
(1985). They found that among 22 loblolly pine (Pinus taeda) seed orchards
ranging in age from 9 to 13 years, 13 had significant clonal variation in
coneworm damage. We have found 8 of these same 22 orchards with significant
1/
Associate Professor of Forestry, Belle W. Baruch Forest Science Institute,
P. O. Box 596, Georgetown, SC 29442, Professor of Forestry, Clemson
University, Clemson, SC 29631 and Research Entomologist, U.S.D.A. Forest
Service, Athens, GA.
221
levels of clonal variation in seed bug damage. If lower infestation levels
are due to a resistance mechanism as suggested by Merkel et al. (1966) then
the variation among clones suggests that breeding for resistance may enhance
current chemical control measures. This paper will discuss some alternatives
to current pesticide application techniques and their implications for the
breeding program.
SELECTION
Selection for coneworm and seed bug resistance may be effective if the
controlling mechanism is genetic in nature. However, selection for this type
of resistance is much different than selection for other types of pest
resistance that are associated with product degradation in the outplanted
progeny. Such selection places emphasis on a non-commercial product in the
sense of fiber, lumber, etc. Improvement in seed production will not
necessarily improve the potential genetic gain obtainable in the orchard, but
will provide a greater resource for planting improved trees. Consideration
of the increase in base population size necessary for incorporation of an
additional independent criterion is of primary importance. The additional
criterion will not be burdensome if the resistance level is positively
correlated with the major commercial traits being evaluated. However, the
number of trees to be screened may be increased by several fold, perhaps to
an impractical level if a strong negative correlation is found. Askew et al.
(1985) found little or no correlation of seed bug or coneworm infestation
levels with height growth or diameter growth performance values. They found
that the probability of a tree selected on its commercial value having less
than 5Z coneworm infestation was approximately 0.35. Hence, if no conscious
selection was practiced for resistance, 35 out of every 100 select trees
would be expected to be 95% resistant. Only 40% of the clones examined had
less than 5% of their seed damaged by seed bugs. However, if coneworm damage
and seed bug damage are viewed as independent events. Then only 14% of the
trees selected for commercial traits could be expected to suffer less than 5Z
damage from both coneworms and seedbugs without effective pesticide
applications.
In general, it seems to be an undesirable proposition to include
resistance as a selection criterion for every first generation selection.
Commercial value of the wood product far outweighs the potential problem of
seed and cone predation. Even if the resistance criterion was mandatory it
would be difficult to evaluate the potential for infestation of trees growing
in the wild. The opportunity for infestation may not exist at the time of
evaluation or may not be a problem in that particular geographic region. Low
levels of cone production in forest stands may be prohibitive to the
formation of large insect populations and hence infestation levels may be low
even on highly susceptible trees. Indirect selection is also a problem
because at present there is no morphological trait that is known to be highly
correlated with the trees' infestation potentials. A breeder would be
limited to selecting trees on the basis of commercial value and then
estimating the number of resistant trees that will be included by chance.
Thus, it would be difficult to assess the potential for cone and seed insect
problems until the orchard was established and their trees began to produce
substantial cone crops.
222
In light of the problems of developing methods of selecting base trees
for resistance it seems feasible to use selection for resistance when the
orchard receives its first roguing. If cone and seed insects are a major
problem despite spraying, then a substantial gain in seed production and
perhaps a reduction in insecticide costs can be obtained by tailoring the
roguing criteria to include insect problems as well as commercial genetic
value. There are several implications of selecting against insect damage at
the time of roguing. First, you must determine what level of infestation is
tolerated for your orchard. You may have to rogue 65% of your trees if a
level of 5% infestation is the maximum you wish to allow. As you reduce the
stringency of your criterion the number of trees that will meet your needs
will increase. In any case it will be necessary to begin with a larger base
orchard in order to assure an orchard of sufficient size after roguing.
Rather than placing a strict resistance criterion on every selection it
may be possible to develop selection indices that incorporate both commercial
traits and insect resistance with each being weighted relative to its
importance. Another possibility would be to use several "completely"
resistant trees as parents for second generation orchard trees. This would
allow resistance to be bred.into the commercially superior base population.
As the breeding program advance, resistance would be accumulated as a
companion trait. Introduction of resistance in this manner would allow for
basic selection on commercial value for the majority of the trees but would
rely on a highly heritable resistance trait in the few trees to be used as
donors.
SELECTIVE SPRAYING
If breeding for resistance is undesirable, a well structured spraying
program may be a viable alternative. Current orchard spraying regimens
usually involve a series of whole-orchard sprayings. Heavily infested trees
receive the same treatment as lightly infested or noninfested trees. DeBarr
et al. (1972) suggested spraying heavily infested clones on an individualized
basis. Noninfested and lightly infested trees could be ignored or receive an
abridged treatment. This idea bears some merit but several assumptions must
be met if it is to be successful. First, the breeder must be able to
accurately identify the heavily infested trees at a sufficiently early stage
in order to prevent damage. Secondly, the insect population must be choosing
_the trees that they heavily infest because they are more susceptible than the
others. If the selection is by chance, or because they are attracted to
these trees rather than being repelled from others the insect population may
merely shift its host base and the problem will still exist. If these
criteria are met, the selective spraying technique may provide the breeder
with a more economical and more effective treatment program.
Combining the selective spraying ideas with inbred resistance would be
the next logical step. During the early years of the breeding program
resistance levels would be low and spraying would be necessary to minimize
the losses. As the level of resistance is increased with advancing
generations, the spraying requirements should decline and may eventually be
unnecessary.
Long generation lengths may make this combined approach technically
viable but practically inoperative. However, advanced generation breeding
223
techniques which reduce generation length may soon put forest’ tree
improvement efforts on a par with agronomic crop improvement programs. The
main point is: tree improvement programs are already in place for many
corporations and public agencies and a minimal effort would be required to
introduce some highly specific genetic material to attain a gain in pest
resistance. Reliance on pesticides causes a continued need for new
pesticides as they lose their effectiveness. Utilizing a natural resistance
mechanism may provide the orchard manager with an opportunity to produce more
seed for less money.
FUTURE PROBLEMS
A potential problem needs to be considered if cone and seed insect
resistance is strongly heritable. Breeding and production programs that are
currently experiencing complete control of cone and seed pests by pesticide
application may not see the justification for increasing the selection
criteria burden. However, infestation potentials may increase with advancing
generations as genes from highly susceptible clones are unknowingly combined
to produce a new parent generation. Potential infestation levels could
accidentally increase to the point being uncontrollable by pesticides alone.
Corrective measures during 4th or 5th generations would be extremely costly
in money and time.
It is important to remember that tree improvement programs in the South
are in their infancy. One or two generations of breeding and selection are
not satisfactory for determining future successes and patterns. Information
that is currently superfluous may be crucial in several generations. Lessons
learned by crop breeders need to be remembered by tree breeders. Our failure
to prepare for future problems cannot be readily corrected. Our problems may
develop slower than those in agriculture but, in all liklihood, they will
develop. .
NEEDS
Much of this proposed breeding strategy is based on a small data base
and personal observations. Many factors still need to be examined in detail.
Heritability of insect resistance needs to be examined. If the trait is
highly heritable and is an additive effect then the prospects for its
incorporation into the commercial orchards gene pool are good. If dominance
factors are the controlling mechanisms the prospects are poor. Linkage may
also be a factor that needs to be carefully evaluated. If the resistance
mechanism is tightly linked to undesirable characteristics, a more exhaustive
search of the base population may be necessary to find qualified trees. The
material necessary for evaluating the heritability of resistance already
exists in progeny test plantings and second generation orchards. Several
years of cone collections using statistically rigorous sampling schemes
should provide breeders with the necessary information for evaluating the
potential of this technique.
Spraying studies need to be conducted in existing orchards to evaluate
the population dynamics of the undesirable insect species. Correct timing of
pesticide applications will be crucial to the success of a “custom made"
breeding and spraying program. Annual changes in infestation levels will
need to be studied before generalized spraying recommendations can be
224
prepared. Confirmation that uninfested trees remain uninfested when the
insect population is excluded from their prime hosts will require careful
observation and cone analyses.
Finally, a method of evaluating the potential for infestation of trees
in previously unaffected regions or orchards will need to be established.
Identification of defense mechanisms and the identification of morphological
markers or biochemical markers that can be easily and positively identified
in the field or laboratory are of prime importance.
LITERATURE CITED
Askew Go Re. kh. he Hedden, CG. Gb. Debarr. 1985. Clonal variation in
susceptibility to coneworms (Dioryctria spp.) in young loblolly pine
seed orchards. Forest Science. (In press)
Debaseee Gereken Ee eb MerkelamC.s Ha OlGwynn, and Ms “H. “Zoerb, . Jr. 1972.
Differences in insect infestation in slash pine seed orchards due to
phorate treatment and clonal variation. Forest Science 18:56-64.
Phe eBe thas else He have aml he Drake. He Ol. Yates, Fil, and, 'G. L:) Debarr.
1981. Seed and cone insects of southern pines. U. S. Dep. Agric. For.
Serv. Gen. Tech. Rep. SE-8, 40 p. Southeast. For. Exp. Stn., Asheville,
N. C., and Southeast. Area, State and Priv. For., Atlanta, Ga.
Merkel, Hh. 9P., A. E. Squillace, and G. W. Bengston. 1966. Evidence of
inherent resistance to Dioryctria infestation in slash pine. In: Proc.
of the Eighth Southern Conf. on For. Tree Imp. pp. 96-99.
225
PREDICTING LOBLOLLY PINE SEEDLING PERFORMANCE
FROM SEED PROPERTIES
T H Shear and T O Perry !
Abstract.- Seed properties affect seedling performance, and may
confound evaluations of progeny tests and determinations of the best
seed handling practices. The total seed weight, coat weight,
gam/emb weight, lipid content, .calories per gam/emb, days to
seedling establishment, and the weights of the seedlings were
determined for the seeds of 19 clones of loblolly pine. There were
large variations among clones in all seed and seedling properties
measured. Various combinations of seed properties accounted for as
much as three fourths of the variation in seedling weight.
INTRODUCTION
Performance of the same progenies in different field tests is often not
consistent. One source of variation could arise from differences in seed
properties and behavior among clones. This paper reports on clonal variations
in seed properties of loblolly pine (Pinus taeda L.) and their effects on the
initial performance of progenies.
In an investigation of loblolly pine seeds, Perry and Hafley (1981)
examined seed weight, embryo condition, seed coat thickness, stratification
requirements, and the number of days required to shed the seed coat. At best,
these factors only accounted for about 20% of the variability in seedling
size. To explain more of this variation, it is proposed here that differences
in loblolly pine seedlings are associated with differences in energy content
of the gametophyte and embryo of the seeds, as well as other seed properties.
The objectives of this study were to:
1. Identify genetic differences in the following seed properties of loblolly
pine: seed coat, gametophyte and embryo, and total seed weight; total lipid
content; energy content (calories per seed); and number of days to seedling
establishment,
2. Measure the performance of the seedlings produced from the seeds, and
3. Correlate seedling performance with seed properties.
MATERIALS AND METHODS
Seeds were obtained from open pollinated clones of loblolly pine in the
Weyerhaeuser Corporation orchard in Washington, North Carolina. Samples
from the seedlots of 10 clones from the 1981 collection and 9 clones from
the 1982 collection were obtained.
Fifty seeds were chosen randomly from each seedlot. The seed coat was
removed. Each seed coat and each combined gametophyte and embryo (gam/emb)
were weighed and saved for subsequent measurements.
Unemployed Ph.D. Forest Scientist, and Professor, Dept. of Forestry,
North Carolina State University, Raleigh, NC
226
Fifty seeds were chosen randomly from each seedlot. The seed coat was
removed. Each seed coat and each combined gametophyte and embryo (gam/emb)
were weighed and saved for subsequent measurements.
The total lipid of each seed was extracted by a modification of the
procedure outlined by Christie (1982). Each gam/emb was crushed in 2 ml of
chloroform:methanol (2:1) and treated with a sonic disintegrator for about 30
seconds to insure complete homogenization and dissolution of all lipid. The
resulting mixtures were filtered into weighed tubes. Distilled water, equal
to one-quarter of the total volume of the filtrate, was added to each filtrate
(approximately 0.38 ml water to 1.52 ml filtrate). The solvents partitioned
into an aqueous upper phase and an organic lower phase. The lower phases,
containing the extracted lipid, were dried in a vacuum dessicator connected to
a dry-ice/acetone cold trap and a vacuum pump. The amount of lipid in each
dried extract was determined gravimetrically. A correction factor which
accounted for the amount of lipid lost from a gam/emb during the extraction
procedure was determined by performing the entire extraction process on 25
seed lipid aliquots of known weight, as if each were a single seed.
Seven clones representing the entire range of lipid concentration (gm
lipid per gm gam/emb) were chosen for measurement of total energy content
using a bomb calorimeter. One-hundred seeds of each clone were decoated and
divided into two groups of 50 gam/embs each and weighed. Each group was
weighed, placed in a gelatin capsule, and burned. The mean calories per gram
of gam/emb determined for each of these clones were regressed against the
lipid concentrations. This regression equation was then used to estimate the
calories per gram of gam/emb of the seeds of all clones.
Forty-nine stratified seeds and 49 unstratified seeds of each clone were
planted at the same time in "Cone-tainers" (1 seed in each). Seeds were
stratified by placing them in a pipette washer with cold running water for 48
hours and then storing them for 40 days at 7 C. They were arranged in random
groups of 7 in racks in a greenhouse (only stratified or unstratified seeds in
each group), watered when necessary, and fertilized approximately every 10
days. The number of days to seedling establishment (DTE) was recorded for
each seed and the mean DTE was calculated for each seedlot. Seedling
establishment is defined here as the number of days from planting to the day
of hypocotyl straightening.
An establishment value (EV) was calculated for each seedlot by
substituting seedling establishment for germination into the formula for the
germination value proposed by Czabator (1962). The establishment value is a
measure of both the speed and the completeness of seedling establishment.
Calculations were made at 30 days for unstratified seedlots and 20 days for
stratified seedlots.
After 75 days from planting, the seedlings from stratified seeds were
harvested (both roots and shoots) because their roots had almost filled the
containers and there was no further seedling establishment. The seedlings
from unstratified seeds were harvested 97 days after planting because there
was still some seedling establishment after 75 days. All seedlings were
oven-dried at 80°C for 24 hours and then weighed.
227
RESULTS
Seed and Seedling Properties
There were significant differences among clones in all seed properties
measured (Table 1).
The loss of lipid in the extraction procedure was determined to be 10% of
the actual lipid weight and the weight of lipid from each gam/emb was
corrected accordingly. The gam/embs varied in lipid concentration from 36% to
70% (mean = 477%)..
The regression used to egtimate the calories per gram of gam/emb from
lipid concentration had an R of 0.88.
Significant differences were evident among mean DTEs of both stratified
and unstratified seeds of many clones (Table 2). Stratification reduced the
mean DTE and its variance by 59% and 70%, respectively. As a result, there
were fewer differences among clones in DTE for stratified seeds.
Prediction Equations
Unstratified seeds.- Every combination of mean DTE and means of seed
properties was examined by regression analysis for prediction of seedling
weight (Table 3). DTE alone accounted for 63% of the variation in seedling
size. When the calories per gam/emb were added to the relationship, 76% of
the variation could be explained.
The number of calories per gam/emb was closely related to the gam/emb
weight (R = 0.99), probably because the calculation of total calories per
individual gam/emb was very sensitive to even small changes in the gam/emb
weight. When the lipid concentration was,halved, only a 20% increase in the
gam/emb weight was necessary to maintain the same total number of calories.
Because of this large dependence, the gam/emb weight could be substituted for
the calories per gam/emb,in the regression without affecting the ability to
predict seedling size (R = 0.76). These were the only models which accounted
for a large amount of the variation in the size of seedlings from unstratified
seeds.
Model 2 supports the hypothesis that seedling weight is affected by the
energy content of the seed. However, Model 3 is more useful for predictive
purposes. It is easier to weigh the gam/embs than to burn them.
Combinations of variables for regression models were reexamined,
substituting EV for DTE. There were no improvements in the regression
coefficients for the previous models and no new models were found.
Stratified seeds.- There were still significant differences in DTE and
seedling weights among clones when the seeds were stratified. However, there
were no significant correlations between DTE and any other measure of the
stratified seeds.
228
6°7L GSL9 aE) 0S00°0 GO MS c°z79 8010°0 Z810°0 16Z0°O SNVEW
2p2 7°6L TtYysse 1¥99 §©FYys L*1Ty 2eP>2 OS00'O 24 O°'Iy 833 0°6S 33 0600°0 SAS OS Os he [tego ?O 9)
y3 ¢°19 qe OLOL q €°€9 2 7600°0 3329 6°9€ P2q T°€9 3F 1800°0 33 6710°O0 33 9€70°0 |
ep L°9L 8339p 9719 = BSB L°GH ~=S PDP_- ZG00"O ys 6°€€ We 1°99 Y3 1g00°0 J®2 £S10°0 3° 6€720°0 V
2p? 7°18 B GOZL B® Z°OL ® 8/00°0 Ul GerAag eB ¢°19 4 6L00°O0 5°P Z910°0 3° 24Z0°0 W
UFOEES Ty LES9 TE 1296 4 8Z00°0 Lie (ROrAS eB €°/9 4 0800°0 JF°P €910°O $F? Z4zZ0°0 q
2 9°76 2P 6419 ap 7°Sr 4 9900°0 F°9P Z°ZE P20 8°79 = BFF: GBD0°0 32 0910°O ®?P S¢z0°0 d
4U6°7S 33°9P OIZ9 U3F9 6°hY YS CE0D0'O 338 8°9E p2q Z°€9 F2 8600°0 F°P2 8910°0 PP 99Z20°0 )
33 7°99 J2P 7/19. Ss « FPP T°8H PPO /¥00°0 Gla Loren 3 Gu/G ee eps VATOLO Jae LKGIORO) — 1D LOLA 0
oqe 6°/8 P2 S789 P 8°0S q $900°0 339 £°9€ Pp2q €°€9 PP OTTIO'O 2¢ g810°0 2 /46z0°0 a
2p O'LL = =FUBF 6199 [tT €°0% 2P2 9400°0 eB G°OY UGS eB LETO°O 32 0910°0 2 46z0°0 a
2P 1°62 YssJep 4899 TYU3F B°eyY P2 1900°O0 eP2 T°8E F°P 6°19 PP? EI110°O Pd S8I0'O ® g86z0'0 y
J2 8°7L = TYysze 7E99 TU 6°07 J°P GYO0'O P29 €°OF 352 £°6S P24 6TI10'O 2P20 6L10°0 > 86zZ0°0 9)
UGG 2q 4969 2 6°4S J°9P GYO0'O PP? L°8E F29P €°19 P24 9TI0°O P24 S8IO°C 4 I10€0°0 a
ys g°09 §=9339P 6619 J9P H°ly 3329 7YV00'O 332 9°9€ P2qd ¥°€9 PP 60T0°0 qd 1610°O0 24 T0€0°0 q
2p 9°LL TF S0S9 46°7E 833 T700°O 2P2 G°6E€ FeP S°09 Pq 6110°O0 P24 S8I0°O0 4 €0€0'0 il
Jep L°€L B3FePp OSL9 FP? 6°94" PP CS00'0 q 7°€y 3 9°9¢ ® ZE10°O 2P2q O8I0°O0 4 LIE0°O N
33 ¢°C9 24 7769 2 8°9¢ pe? 7S00°0 4 6°TE 2 1°89 p2q 6110°0 BECGZO"0 | & 726040 S
P2q 1°Z8 Ty3 ¢099 [tT 9°6€ P29 6700'O USF E°VE FqGB 4°SG9 Ge 6ZI0'0 Bl/720.0) 5 2 9LG0 20 c
qe 1°06 Ty3 €199 [tT O°O% 2 ¢S00°0 ina) 74S ® 0°89 78 4¥Z10°0 ® 9970°0 ® 06£0°0 H
qa /WVD GNa/WVD dIdITI SV LHOIaM GNa/WVD LVOOdAES LHDIaM LHOTaM IHDIGM FNOTO
Wad WVUD Ud awa /WV9 GIdIl SV dGaaS dO LNAOURd awa /WVD LYOO aaas
SATNO1VO SALYOTVO INaOUdd cicicks TVLOL
“GO: 0 =" eUuGC Te s aseavocuenr
pezrquespnyAs Ss, oyANnL q JUSAIFIFITp TqueoTstTu TS oq oq peurutoqjop 70U 9IOM 197997 vues oud q PeMOTLOF
SUuPOoW “suet ur pesSsesidxs o1e S2y TOM UPD TOILNY *SoTIio oid DVIS JO sueowul fo Souorzqo uowe suosTis1eduoy Salli 2FTIeL
229
qe
oq
oqe
e
q
oqe
169€°0O
£96€°0O
L76€°O
Ney) 0)
SLE7 a0
168¢°O
96S7 70
SS 6720
LSLE“O
MGSO)
966€°0
GL9E*O
898E°O
GES -O
786E°O
L£80¢°0
88c£°O
L8€7°0O
7676 0
GSTS O
Sdadas
CaIHILVaLs
sSuT [pees
°(CO°O = PYydTe) 4seq esueT peztTqjuepnis s,Aoyny Aq JUusTeTITp ATQUeOTITUSTS
9q OF PoeUTWAZJOp JOU 9ANM A9JOT owes oy
‘sAep G/ pue /6§ AOJ UMOI
*uOTIBOTITIeAAS
(dS) SWOTIeTASp przepueIAS
e
poqe
qe
oqe
apoqe
oqe
oqe
apoqge
apoqe
poqe
apoqe
apo
apoq
poqe
a
oy
apoqe
apo
2p
7877 °O
YSGL~O
cS0S°0
cv70L°0
68479 °0
IZ8€°O
cL79°O
SYNE) -O
OO8E “0
OZE7 0
OES 70
LDP
TSGS 0)
EECA
6S67 0
6€71°O
S6G 0
LS AO)
TOLE=0
8661 °O
Sdaqds
GaTdILVaLSNnn
(swei3s) [THSIAM ONT IGHAS
*Souo TO
BONG SQ
OKO G0) 18 pe)
LPG. SG po
HP VIG SLAG po
sy tats OG Pe)
Gay) 0°9 po
CSG OMG po
SSe8C Gac po
O}iL NE |! po
HO WIG “OG po
NMG = G po
GL WL OS po
SIG 1S po
HE GG se ES po
BSEEG 29e8 Ppoq
OW Gi {3° ©) Pers
GW ey qe
ale 29%8 oqe
NO) LS BS
76°6 oe Gil e
Ad as
Gl
8°Ol
(ye iI
eG
en OsL
GI
(6 - Ol
Cela
este
LOK
Gul
Oeeil
Or
Gi
ts I
Sh
Casi
caeSi
up eG |
6°81
CRAG
NVaW
SaaS CHIdILVaLs
q PeMmoTTO}F suey.
219M SpooS PeTJTIeIWS pue Spoos PoTjJrT}eAQsuN wor;
q Ppeonped Sem soUeTIPA OYJ MOY MOYS SYLq Ueew oy. Fo
uowe suostiedwod pue *szy43Tem
O8°¢
10°71
8C°?
8L°9
OS-S
7a
6ligZ
LOE
WEG
EL AC
CIRC
LO AO
9S -0
8L°0
fe \L Ih
6€°0O
0S*0O
07°0
9S 40
GEO)
Ad
6
9
c°
0°
z°
Wy Si
6°
> Wt
Loi
(Se VKG
Galg
Gy
8°1¢
ITEC
G16
E.0G
L0G
SSG
SHE
as
ate
3 6°41
3 9°91
3 9°LT
33 9°8I
332 G°IZ
B32 6°IZ
332P 6°72
BJepo €°/Z
BJOPo 0°62
Jopoq L°ECE
Jopoqd E°vE
epoq 6°SE
2poq 1[°9€
P2q 9°8E
2qe 6°17
qe 6°SY
qe ily
qe 1°67
BGG
qLQq ANOTIO
NVaW
WN
CIM 27) © 15s area el = (OA ea Se :
SaaS GHIAILVaLSNn
Teatjzoedse1r
UTT pees ueSW
UTTpees 02 skep Fo SAoqunu ues “7 OTqGeL
230
Table 3. Summary of important regression models for the dependent variable
seedling weight. Unless indicated, all parameter estimates for
independent variables were significant at alpha = 0.05.
MODEL INDEPENDENT VARIABLES R-SQUARE
For Unstratified Seeds
1 DTE 0.63
2 DTE, calories per gam/emb 0.76
3} DTE, gam/emb weight 0.76
For Stratified Seeds
4 coat weight, gam/emb weight, total weight OFS
5 coat weight, gam/emb weight , total weight,
calories per gam/emb 0.50
6 EV 0.46
7 EV, coat weight, gam/emb weight, total weight 0.72
parameter estimate significant at alpha = 0.10
Although other models had higher Ro values, Model 4 (R? = 0.37) would be
the easiest to use because it requires only weights which are easily obtained.
Calories per gam/emb accounted for an additional 13% of the variation in
seedling size (Model 5). However, the parameter estimate for calories was
Significant only at the 90% level. These were the only models in which DTE
and seed properties accounted for a large amount of the variation in the
weight of seedlings from stratified seeds.
All models were retested, substituting EV for DTE. Model 6, with EV as
- the only independent variable, accounted for 46% of the variation. The
weights of the seed parts and total seed weight to the model accounted for
another 26%.
DISCUSSION
The total seed weight, often proposed as being related to seedling size,
was of no predictive value in this study. The seed coat represented the major
portion of the seed (approximately 63% by weight). However, seed coat weight
was not closely correlated with the gam/emb weight (R = 0.51), but was highly
correlated with the total seed weight (R = 0.95). These correlations
demonstrate that seed coat weight may change among some clones without
231
relative changes or even absolute changes in gam/emb weight. Some seeds were
heavier only because they had heavier seed coats.
It might be expected that larger seed coats would offer greater mechanical
restriction, have more inhibitory chemicals, etc., and result in longer
germination and establishment times (Barnett 1972). However, there was no
correlation between seed coat weight and DTE, which does not support this
view. Gam/emb weight appears to be more important than seed coat weight in
determining seedling size. Total seed weight, primarily determined by the
seed coat weight, is a poor predictor of seedling weight.
The EV accounted for variation in the weights of seedlings from stratified
seeds that the DTE could not explain. The establishment (germination) value
varies directly and proportionally with both speed of establishment and total
establishment and is relatively sensitive to minor differences in either. The
EV is an absolute value while the DTE is a mean. Since there were few
differences between DTEs or seedling weights for stratified seeds, it was not
likely that a relationship between the two could be developed. In contrast,
the EVs spanned a large range and the differences among clones in them were
quite large. For unstratified seed, there were many differences in DTE and it
did not matter if DTE or EV was used to develop relationships.
CONCLUSIONS AND IMPLICATIONS
As expected, there were large differences among clones in all of the seed
and seedling properties examined. There were no simple relationships between
total seed weight and seedling performance. Seed size should be closely
correlated with seed weight. It does not appear that sorting seeds by size is
a useful nursery practice for controlling the size of seedlings. Rather,
seeds should be planted separately by clone, a standard practice in some
nurseries.
Many attempts to reduce ithe time required to test the performance of tree
progenies have been unsuccessful. Seed size has often been thought to be
related to early progeny performance, and has been proposed as a possible
quick indicator of progeny performance. But the correlation between seed size
and progeny performance is usually poor.
In attempts to shorten progeny testing, age:age correlations (i.e.,
juvenile:mature correlations) are used to determine if differences among
clones in progeny characteristics (i.e., height) are consistent throughout the
lives of the plants. The correlations between seedling size and size at older
ages decrease with time, but do not disappear. Despite some arguments to the
contrary, trees that start out big seem to remain big (Bengston 1963, Grigsby
1975, Hatchell et al. 1972, Sluder 1979, Wakely 1969, Zarger 1965). Indeed,
exponential growth models demonstrate that the rate of growth is a function of
the initial size of the organism (Lotka 1956). The effect of initial size can
be amplified in the nursery where large seedlings often suppress smaller ones.
If half to three-quarters of the variation in seedling size is
attributable to nongenetic and nonheritable properties of the seeds, and if
seedling size is correlated to tree size, then estimates of genetic gain may
be overstated. When genetic gain is calculated from progeny tests, the effect
of initial size is not considered. Thus, it becomes included in the estimates
of genetic gain. Along with heritable factors, there are many nonhereditary
232
factors and cultural practices that affect seed properties. Energy content
and the number of days to seedling establishment (DTE) may also be strongly
influenced by these factors.
The small embryo represents less than 10% of the total seed and is the
only part that contains chromosomes from both parents. Maternal
influences on the seed may have many effects on the resulting seedling (Perry
1976). First generation progeny tests may partly select the best seeds rather
than genetically superior progeny. While there is certainly genetic gain as a
result of progeny testing, it is likely that it is not being accurately
estimated. To accurately predict future growth on the basis of early progeny
performance, other genetic and nongenetic factors that regulate germination
and growth must be taken into account.
LITERATURE CITED
Barnett, J. P. 1972. Seedcoat influences dormancy of loblolly pine seeds.
Gane). Hor. Res. 2): 7=—10:
Bengston, G. E. 1963. Slash pine selected from nurserybeds: 8-year
performance record. J. For. 61:422-425.
Christie, W. W. 1982. Lipid Analysis, 2nd edition. Pergammon Press Inc., New
Work se2 07-pps'-
Czabator, F. J. 1962. Germination value: an index combining speed and
completeness of pine seed germination. For. Sci. 8:386-396.
Grigsby, H. C. 1975. Performance of large loblolly and shortleaf pine
seedlings after 9 to 12 years. U.S. For. Ser. Res. Note #196.
Hatchell, G. E., K. W. Dorman, and O. G. Langdon. 1982. Performance of
loblolly and slash pine nursery selections. For. Sci. 18:308-313.
Lotka, A. J. 1956. Elements of Mathematical Biology. Dover Pub., New York,
465 pps.
Perry, T. 0. 1976. Maternal effects on the early performance of tree
progenies, pp. 473-481. InM. G. R. Cannell and F. T. Last (eds.), Tree
Physiology and Yield Improvement. Academic Press, New York.
Perry, T. O., and W. L. Hafley. 1981. Variation in seedling growth rates:
Their genetic and physiological bases. 16th Sou. For. Tree Improvement
Conf., Blacksburg, Virg., Sponsered Publ. #38 of the Sou. For. Tree
Improvement Comm.
Sluder, B. R. .1979. The effects of seed and seedling size on survival and
growth of loblolly pine. Tree Planter's Notes, 30:25-28.
Wakely, P. C. 1969. Results of southern pine planting experiments
established in the middle twenties. J. For. 67:237-241.
Zarger, T. G. 1965. Performance of loblolly, shortleaf, and eastern white
pine superseedlings. Silvae Genetica 14:182-186.
233
AN OCALA SAND PINE PROGENY TEST
COMPARED WITH A SEEDLING SEED ORCHARD
Timothy LaFarge, Ralph Lewis, and James L. McConnell 1/
Abstract.--Sixty-three half-sib families of Ocala sand pine were
grown in a seedling seed orchard and a progeny test in Central Florida.
At age 7, differences among families were significant for height, dbh,
and survival in the seed orchard, for survival in the progeny test,
and for dbh and survival in the combined analysis of variance of both
locations. In the combined analysis the location X family interaction was
only significant for survival. However, inspection of family means of each
trait showed that some of the best and worst families changed rank
dramatically between tests.
Additional keywords: Half-sib family, genotype X environment interaction,
Pinus clausa, seedling seed orchard, progeny test.
Geneticists of the Region 8 Tree Improvement Program are assessing the
performance of 120 half-sib families of sand pine (Pinus clausa (Chapm. )
Vasey) in Central Florida on the Lake George Ranger District north of Silver
Springs. In addition to the seed orchard, 106 of these families, plus three
check lots, were also planted the same year (1978) in a progeny test on State
Highway 19 about 20 miles from the orchard. Both the seedling seed orchard and
the progeny test also serve as progeny tests for most of the ortets
represented in a clonal seed orchard growing contiguous to the seedling seed
orchard.
The purpose of the family assessments is to provide data on which to base
thinnings and roguings in the seedling seed orchard. If family performance is
uniform on the two sites, then genetic gain can be maximized by roguing a
large number of the worst half-sib families from the seedling orchard.
However, if there are strong genotype X environment interactions, then family
selection must be decreased and heavy thinnings accomplished primarily by
means of removal of the worst trees within each family.
METHODS
Both the seedling. seed orchard and the progeny test were established in
January and February 1978. In the seedling orchard 120 families were planted
in 250 blocks aS non-contiguous single-tree plots at a spacing of 5 feet. in
the progeny test 106 families plus 3 check lots were established in 10-tree
row plots in 7 replications in a randomized complete-block design.
The seed orchard has been thinned twice, once in 1981 and once in 1983.
Of the 30,000 trees planted, 7,466 remain. Two more planned thinnings will
reduce this number to less than 2,000. The progeny test has only been thinned
by natural mortality, and survival is currently 45 percent. The progeny test
was included in a RARE II Wilderness area soon after establishment, so that no
competition control or cultural activity of any kind has been possible.
1/ Eastern Zone Geneticist, Forester, and Regional Geneticist respectively,
USDA Forest Service, Region 8, Atlanta, Georgia.
234
Table 1. Degrees of freedom, mean squares and significances of
differences for three traits among 63 sand pine half-sib
families in a seedling seed orchard, a progeny test, anda
combined analysis of variance of both tests.
MEAN SQUARES for:
DEGREES
SOURCE OF of PERCENT
VARIATION FREEDOM SURVIVAL HEIGHT DBH
SEEDLING SEED ORCHARD
Replicate
(R) 3 0.146 6.274 1.433
Family
(F) 62 OR023) ** 12469 ** 0.221 **
RF 186 0.005 0.458 0.030
PROGENY TEST
Replicate
(R) 3 0.257 8.389 0.170
Family
(F) 62 0.088 ** Sols ans 0.177 ns
RF 186 0.045 6.189 0.186
COMBINED ANALYSIS OF VARIANCE
Location
(L) iL Arias % 8,296.046 ** 250.190 **
F 62 0). 060) ** Sols ns Oe 690s *
R(L) 6 0.204 (330 0.796
LF 62 a WQeul bk 3.469 ns 05225 ans
RF (L) 372 0.025 3.324 0.108
* Difference is statistically significant at the .05 level.
** Difference is statistically significant at the .01 level.
ns Difference is not statistically significant.
235
Both the orchard and the test were measured in 1981 at age 3. The orchara
was also measured in 1983 at age 5 and in 1984 at age 6. The progeny test was
measured a second time in March, 1985 at age 7. Both height and dbh were
included in the last measurements at each test, and it is this set of
measurements which we will investigate.
We performed a combined analysis of variance for survival, height, and dbh
from both the orchard and the progeny test. To adapt the seed orchard and the
progeny test to compatible field designs, we had to modify the field design of
each. A road system divides the seed orchard into three large blocks of
roughly equal size. Each family plot in a block is represented by a number of
non-contiguous single-tree plots. Hence, for the purpose of data analysis we
treated each of these large blocks as a replication. In the progeny test we
used data from only the first four of the seven replications. These were, in
fact, the most complete replicates anyway.
A subset of 63 families was then chosen because those families contained
no missing plots in the four replications at each location. Because the
orchard contained no check lots, none were included in the analysis. In
summary the combined analysis of variance comprised two locations, each
containing four replications and 63 half-sib families.
There are some important differences between the seed orchard and the
progeny test. First, the seed orchard is mostly on a longleaf (Pinus
palustris Mill.) island, but the progeny test is on a deep sand, a typical
sand pine site. In the orchard each family plot is a number of noncontiguous
single-tree plots randomly scattered throughout each replication, whereas in
the progeny test each family plot is a single 10-tree row plot in each
replicate.
The anlyses of variance were performed by means of program P8V of the
Biomedical Data Processing Statistical Software system (Jennrich and Sampson
1983). Also, Duncan's multiple range tests were performed by hand.
RESULTS
In table 1 there were significant differences among families in the
orchard for survival, height, and dbh. But survival was the only trait
showing significant family differences in the progeny test. In the combined
analysis of variance, survival and dbh showed significant differences.
The analyses considered here were based on measurements taken at ages 6
and 7 on the seed orchard and progeny test respectively. However, height at
age 3 in the progeny test, with essentially the same subset of families
considered, showed significant differences among families. The family ranking
for height at age 6 differ little from those at age 3. Lack of competition
control probably has allowed environmental noise (replicate X family
interactions and variation among trees within plots) to mask family height
growth differences. In table 1 only survival shows significant location X
family interactions. However, we believe that an examination of the data will
show some strong location X family interactions for height and diameter as
well.
236
Table 2.
FAMILY
I.D.
a/ Means not followed by the same letter are significantly different
Rank comparisons of the best and worst sand pine half-sib
families for three traits in a seedling seed orchard and a
progeny test.
SEEDLING
SEED
ORCHARD
Percent Rank
25.7 22
43.5 1
41.5 3
MAE) 63
Feet Rank
20.7 2
20.8 1
17.6 63
19.7 18
18.5 58
Inches Rank
3.4 1
2.9 25
3.0 9
2.5 63
3.0 9
2.7 59
at the .05 level
SURVIVAL
HEIGHT
p= .05
def
237
PROGENY
TEST
Rank
77.5 1
32.5 47
75.0 2
15.0 63
Rank
14.0 1
11.4 33
11.9 21
8.9 63
8.9 63
Rank
1.6 14
2.0 1
20 1
1.6 14
Heal 63
oil 63
p= .05
a
de-gh
ab
h
p= .05
a
abc
abc
c
c
p = .05
ab
a
a
ab
b
b
In table 2 we have listed some of the rank changes of certain families for
best and worst performance in all three traits. Some families maintain their
rank as best or worst very well in both the orchard and the progeny test. Two
examples are: (1) family 12 ranked 2 and 1 for height in the seed ochard and
progeny test respectively; (2) family 120 ranked 58 and 63 for height in the
orchard and the progeny test respectively. However, other families made
dramatic shifts. Two examples are: (1) family 97 ranked 63 for height in the
seed orchard but 21 for height in the progeny test; (2) family 118 ranked 9
for dbh in the orchard, but 63 for dbh in the progeny test.
Family 97 typifies a short term roguing problem. If the seed orchard were
our only source of information, we would rogue this family. But in the
progeny test on a site representative of the kind of site on which it will
probably be planted, it grows well. Family 118 represents a long term
problem. Without the progeny test we would select its best trees for further
breeding, based on its rank of 9 in the seed orchard, but its rank of 63 in
the progeny test should persuade us to exclude it from most sites on which we
are likely to be planting sand pine.
CONCLUSIONS
The results in the combined analysis seem to bear a warning. Since the
progeny test site is representative of the sites on which we intend to
re-establish sand pine, we should probably pay heed to the progeny test
results in our selection and roguing of families. If we should plan to put
any operational plantings on longleaf islands, then we have some families,
which seem to be better adapted to those sites also. Other families may be
planted on either site without risk.
LITERATURE CITED
Jennrich, Robert and Sampson, Paul 1983. P8V 15.5 General mixed model analysis
of variance -— equal cell sizes. In W.J. Dixon (ed.) BMDP statistical
software - 1983 printing with additions p. 427-435. Berkeley: University of
California Press.
238
HARDWOOD GENETICS
MODERATED BY DR. BRUCE BONGARTEN
University of Georgia
239
ee are
hacer
GEOGRAPHIC PATTERNS OF VARIATION IN
GROWTH OF SWEETGUM IN EAST TEXAS
A. F. Stauder, III, and W. J. Lowel/
Abstract.--Two groups of open-pollinated sweetgum
progeny tests from 202 selections were planted in east
Texas. Study A consisted of three tests and was estab-
lished in northeast Texas and southeast Texas. This
study contained seedlings from 100 families representing
12 counties in east Texas. Significant provenance dif-
ferences were observed for height and diameter but not
for survival and volume. The planting location by family
within provenance interaction was significant for sur-
vival, diameter and volume. Two other tests (Study B)
were established in east-central Texas and southeast
Texas and contained seedlings from 102 families repre-
senting east Texas, south Arkansas and west Louisiana.
The provenance effect was significant for survival and
volume, and the family within provenance term was highly
significant for all traits. There were no important
genotype by environment interactions. The data indicate
that sweetgum seed should be collected from Polk, Tyler,
Newton and Jasper Counties for use in east Texas.
Additional keywords: Liquidambar styraciflua, progeny tests,
heritability.
Sweetgum (Liquidambar styraciflua L.) is a southern bottomland
species and occurs mostly on rich, moist, alluvial soils (Harlow and
Harrar 1969). The range of sweetgum extends from Connecticut southward
throughout the eastern United States to central Florida and east Texas,
as well as scattered locations in Central America (Fowells 1965). Be-
sides being an important commercial hardwood species in the United
States, sweetgum is a desirable ornamental because of its attractive
shape and brilliant autumn leaf coloration. The North Carolina State
University-Industry Hardwood Research Cooperative, the Western Gulf
Forest Tree Improvement Program-Hardwood and the Texas Forest Service
Urban Tree Improvement Program have active tree improvement programs
underway with sweetgum, and selected trees have been accepted for use
in breeding arboretums and seed orchards (North Carolina State Univer-
sity 1984, Byram et al. 1984). Sweetgum is an intolerant species and
large, pure even-aged natural stands are not uncommon. It appears
1/Silviculturist, Texas Forest Service, Alto, Texas; and Associate
Geneticist, Texas Forest Service and Assistant Professor, Texas Agircul-
tural Experiment Station, College Station, Texas. The authors express
appreciation to Bosch Development Company, Champion International and
Temple-Eastex Forests for providing land to establish these tests.
240
z;
Ny
“a
\
A
that sweetgum is relatively slow to establish itself; however, once
established growth is relatively fast on many bottomland sites.
The objectives of this study were to examine the amount of genetic
variation present in stands of sweetgum in east Texas and develop seed
movement guidelines. Estimated genetic gains from family selection are
also described.
METHODS
Study A
Open-pollinated seed from 110 sweetgum selections representing 12
counties and six provenances throughout east Texas were collected in
1969 and 1970. The seedlings from these selections were grown in 1970
and 1971 in three replications at Indian Mound Nursery located near
Alto, Texas. One progeny test containing 100 families was established
in early spring, 1971 in Jasper County, Texas (Figure 1). Two other
progeny tests containing 110 families were planted in 1972 at Harrison
County and Montgomery County. All seedlings were root pruned to eight
inches before planting.
Data from the 100 common families among the three plantings were
used in the analysis. The field design for all tests was a six-replica-
tion, randomized complete block with four-tree family row plots.
Spacing was 10 by 10 feet. A single border row was used at each
location to offset edge effects; however, the border row was partially
destroyed at the Harrison County location.
Study B
Open-pollinated seed were collected in 1972 from 106 sweetgum se-
lections representing 16 counties and six geographic areas of seed col-
lection from east Texas as well as two counties each from south Arkan-
sas and west Louisiana. As in Study A, these seedlings were grown at
Indian Mound Nursery. Two progeny tests containing 104 families each
were established in 1974 in Cherokee County and Montgomery County
(Figure 1). Data from 102 families in common between the two locations
were used for the analysis. The field design was the same as that used
for Study A.
Site Description and Cultural Treatments
The Harrison County planting was cleared, previously forested land
of silty loam soil. The Cherokee County planting site was sandy loam
soil and was also cleared, previously forested land. Both sites in
Montgomery County were cut over areas on clay loam soils that were pre-
viously planted in pine and abandoned one year after planting. The
test in Jasper County was an old clearcut on a sandy clay loam soil.
Weeds and sprout competition were controlled in all tests by
disking during the first three years and by mowing thereafter. The
241
Fe
Panola
Red River
Bowie
ie
fu
(7)
n
oF
)
plantation locations for Study A
Plantation locations for Study B
Figure 1.--Seed collection and plantation locations in east Texas for the
sweetgum progeny tests.
242
test at Harrison County was fertilized in 1973 and the plantations in
Cherokee and Montgomery Counties were fertilized in 1974.
Analysis
Provenance and family within provenance variation were examined by
a least squares regression approach using the General Linear Model
(GLM) procedure of the Statistical Analysis Systems (SAS Institute,
Inc. 1982). Plot means were used in each of the combined analyses.
The provenance effect was considered as fixed, while locations, replica-
tions and families within provenance were considered random effects.
Dead trees were assigned a zero volume to account for survival dif-
ferences. A Satterthwaite-F (pseudo-F) test was used in the absence of
valid tests (Hicks 1973). Family heritability and gain estimates were
calculated where appropriate for survival, height, diameter and volume.
RESULTS AND DISCUSSION
Study A
Survival ranged from 58 to 91 percent and averaged 77 percent for
the three plantings at age 10 (Table 1). Height, diameter and volume
averaged 5.9 m, 7.2 cm and 7.7 dm , respectfully. The test in Jasper
County showed the best growth while the one at Montgomery County had
the slowest.
Table 1.--Plantation means for five ten-year old sweetgum tests in east
Texas.
Plantation Survival Height Diameter Volume
(4) (m) (cm) (dm3)
Study A
Harrison County 83 Do U 6.9 6.9
Montgomery County 58 Sel TSO) 4.9
Jasper County 91 6.8 7.8 1 es2
Study B
Cherokee County FOZ iO 10.5 230
Montgomery County 86 Het 10.9 2512
The analysis of variance indicated significant provenance effects
for height and diameter (Table 2). Trees from provenances four and
five in east Texas were the largest (Table 3). These provenances were
represented by families collected from Polk, Tyler, Jasper and Newton
Counties. There was no detectable planting location by provenance in-
teraction for the observed traits.
243
Table 2.--Combined analysis of variance for survival, height, diameter
and volume of three sweetgum progeny tests (Study A).
Source of Mean squares for
Variation df Sur. Ht. Dia. Vol.
Location 2 75505 .83** 272.03** 76.90 3316.45**
Replication(Loc. ) 15 2269 .04** 22.36** 56.40** 369 .28**
Provenance 5 1672.49 8.78** 20.45* 84.41
Family (Provenance) 94 697.31 1,.44** 3.00 27.76
Loc. x Provenance 10 1045.40 OR 75 4.38 46.79
Loc. x Family(Prov.) 188 840 .82** 0.93 2.58% 26. 94%%
Rep.(Loc.) x Prov. 75 (AIS UT 0.92 2.87* 21.45
Error 1332 374.19 0.79 2.08 19.03
*Significant at 0.05 level of significance
**Significant at 0.01 level of significance
Table 3.--Provenance means for the combined analysis for the
ten-year-old sweetgum progeny tests in east Texas.
Provenance Height Diameter Volume
(m) (cm) (dm3)
Study A
1 5.8 TO YES)
2 5.8 7.0 oD)
3 6.0 foe 8.3
4 6.2 1a) Sry
5 6.1 7.6 Sii7
6 Be 7.0 7.3
Study B ,
1 UBS 10.2 20.8
2 oll 10.9 23 yeal
i) Vor 9.9 7G)
4 Srl alyey/, 29.6
5 8.0 elev 2637
6 Voll 10.8 Dalek
15 (La‘..) Tio 10.9 24.3
18 (Ark. ) oe 10.9 24.0
Family within provenance differences could only be detected for
height. Familiy heritability and gain estimates for height was 0.35
(SE=0.16) and 0.16 m (3%), respectively. The planting location by
family within provenance interaction was significant for survival, diam-
eter and volume. This indicates that family rankings for these traits
differed among the plantations. The results from Study A indicate that
sweetgum seed can be collected from any area within these boundaries
and planted in east Texas without a loss in survival and volume growth.
Height and diameter growth can be significantly increased by collecting
seed from trees in provenances four and five. They also indicate that
any family differences in survival, diameter and volume growth are
masked by the significant interactions.
244
Study B
Survival was good (84 percent) for the average of the two 102-fami-
ly progeny tests at Cherokee County and Montgomery County. Average
volume was 23.1 dm3, which was larger than that for any of the three
tests in Study A (Table 1).
Results from an analysis of variance for these two plantings re-
vealed no differences between the planting locations (Table 4). Sur-
vival and volume varied significantly among provenances. As shown in
Table 3, trees from provenances four and five were clearly the largest.
These same geographic areas produced the tallest trees in Study A.
Table 4.--Combined analysis for survival, height, diameter and volume
at_age ten of two sweetgum progeny tests (Study B).
Source of Mean squares for
Variation df Sur. Ht. Dia. Vol.
Location 1 2637028 2.30 byeets: 123)2).73
Replication(Loc.) 10 2079. 18** 51.96** 124,.03** 3050.05**
Provenance 7 1593.10** 15.45 36.15 1585.87
Family (Provenance ) 94 128.12 3,71 25% 8. 44%* 321.11**
Loc. x Provenance 7 USS) 3.74 9.45 274.48
Loc. x Family(Prov.) 94 364.52 1.62 3). 96 125.41
Rep.(Loc.) x Prov. 70 386.15 2.58 5.105 190.15
Error 894 SOMES 2.04 3.96 164.83
* Significant at 0.05 level of significance
** Significant at 0.01 level of significance
In this study, the family within provenance term was highly signif-
icant for survival, height, diameter and volume. There were no signifi-
cant genotype x environment interactions for these plantings, indicat-
ing consistent family rankings between the tests. Selected families as
well as provenances performed well at both locations. This study in-
dicates that a single breeding population can be used in central and
southeast Texas.
Family heritability and gain estimates were calculated for all
four traits (Table 5). Estimates ranged from h2=0.50 for survival to
h2=0.54 for volume. Genetic gain by selecting the best 15 families out
of 102 for survival, height, diameter and volume were 6.15 percent
(7%), 0.47 m (6%), 0.71 cm (7%), and 4.72 dm3 (20%), respectively.
These gains appear to be sufficient for use in an operational tree im-
provement program.
245
Table 5.--Family heritabilities (h2) and estimated genetic gains for
the combined analysis of two sweetgum tests (Study B).
Variable Family Gain
h2+ SE
Height 0.51+0.14 0.47 m
Diameter 0). 53+0.16 0.71 cm
Volume 0.54+0.14 4.72 dm3
CONCLUSIONS
The results from Study A indicated that the geographic area of
seed collection within east Texas affected height and diameter growth
but not survival or volume. Individual family differences could only
be detected for height (h2=0.35). There was also a significant loca-
tion by family interaction for survival, diameter and volume. The re-
sults from Study B revealed a significant effect of geographic area of
seed collection on survival and volume as well as significant family
differences for all traits. Growth traits appeared to be moderately
inherited in sweetgum (volume h2=0.54). By selecting the 15 best fami-
lies, expected genetic gains were six percent for height, seven percent
for survival and diameter, and 20 percent for volume growth.
Based on these studies, it appears that sweetgum seed should be
collected from Jasper, Newton, Polk and Tyler Counties for use in east
Texas.
LITERATURE CITED
Byram, TasDe, “owe, We J... McKinley. Co Ra. Robinson, (Jiu. as Gaudere
A. F. and van Buijtenen, J. P. 1984. 32nd progress report of the
cooperative forest tree improvement program. Forest Genet. Lab.,
Texas Forest Service Cir. 269. 27 p.
Fowells, H. A., ed. 1965. Silvics of forest trees of the United
States, USDA Agric. Handb., No. 27i, 762 p.
Harlow, W. M. and Harrar, E. S. 1969. Textbook of dendrology.
McGraw-Hill Book Coe. Neo. Dli2 pe
Hicks, C. R. 1973. Fundamental concepts in the design of experiments.
Holt, Rinehart and Winston, N. Y. 349 p.
North Carolina State University. 1984. Twenty-first annual report, N.
C. State University - Industry Cooperative Hardwood Research Pro-
gram Sch. Forest Resor., Raleigh, N. C. 61 p.
246
SAS Institute, Inc. 1982. SAS user's guide: statistics, 1982 ed.
Carys) N- Cr. 584 p.
247
JUVENLLE GROWIH PERFORMANCE IN A PROVENANCE TEST OF SWEETGUM
by Kim C. Steiner, Bruce Bongarten, and Randall J. Rousseau!
Abstract.--In a provenance test of sweetgum planted at four
locations, the tallest trees after four growing seasons in the
field were generally of non-local origin. Family-within-prove-
nance variation was significant at two locations, and in every
plantation was 23 to 44 percent as large as the provenance
component. Sweetgum improvement programs should incorporate
both provenance selection and progeny testing of wild parents.
Additional keywords: Liquidambar styraciflua, height, progeny
test, natural variation.
Prior genetic evaluations of sweetgum (Liquidambar sytraciflua L.) have
focused on that part of the species' natural range in which it is planted
most frequently, the Piedmont and Coastal Plain (Mohn and Schmitt 1973,
Sprague and Weir 1973, Texas Forest Service 1975, Wells et al. 1979). How-
ever, sweetgum is also planted commercially in bottomlands of the central
interior states, and it is a common street and ornamental tree as far as 200
km north of the natural range.
This study was created to fill the need for a provenance test appropri-
ate to the northern portion of the sweetgum commercial region. Its purpose
was primarily to evaluate in northern environments the performance of popula-
tions native to the northern two-thirds of the species' range, although two
plantations in more southern locations provide a useful opportunity to com-
pare performance. All or portions of the collection have been established in
experimental plantations in Georgia, Illinois, lowa, Michigan, New York,
Pennsylvania, South Carolina, Vermont, and West Virginia. Growth performance
after one year in three West Virginia plantations was reported by Prowant et
al. (1983). We are reporting performance after four growing seasons in plan-
tations in Georgia, Illinois, Pennsylvania, and South Carolina.
METHODS
Seed collections were made fall 1975 from 1 to 4 open-pollinated trees
in each of 47 populations of sweetgum distributed broadly, but mostly north
of the Coastal Plain (Figure 1). A "population" was arbitrarily defined as
occupying an area no larger than 25 km?, and parent trees were essentially
unselected as to phenotype. The trees in most populations were presumed to
lRespectively, Associate Professor, School of Forest Resources, Pennsylvania
State University, University Park, PA; Assistant Professor, School of Forest
Resources, University of Georgia, Athens, GA; and Research Geneticist,
Timberlands Division, Westvaco, Wickliffe, KY. This research was supported
in part by the U.S. Department of Agriculture, Cooperative Regional Research
Projects NE-27 and S-23, and in part by Grant No. 23-773 from the U.S. Forest
Service, Consortium for Environmental Forestry Studies. Thanks are due to
O. O. Wells for his comments on an earlier version.
248
TB
Va
N
\
\
\
\
co
\ ee --- | c
) =a 1 313
; 7 j \ \ *PA 285
( : | \ 428)
\ ) l I / 446) 4297
a JY | 1 oe ae = 4204
-~~ \ SAS
ae SS ( | \ os Yea !
y \ y \ ( ie 407
\ r6 ey) ab
— PK a
| ) { 12147 WY Yy if wea 38?
| ¢ 1097 Sai25 4 WY UY ae)
! 377 ys, A070 SAA,
{ 245° A481 LLY Lh
apg Aaa? 123 0452/F, MY tft)
l Ky, iL fs,
/ Lh) Lb pf 7]
= — MM, hhh Ld) £4 | wi”
y, 3 7 Ys
\ i A105 rd, Hey
\ yj p
1014 4557 a 4103 2894 Mai fLY OAL
{ boas 94549 Wl boars /] K/
Vyas? WW ff 7, O77 4/77, Hy KK 406! V4
{ /
A445 \ MK 424)
%
*%GA, A* SC
A033// N
eae
SS eS
Figure 1.--Provenance locations of sweetgum evaluated for growth rate. Plan-
tation locations are also indicated. Not shown are several collections of
undetermined origin and four provenances in Georgia and Mississippi sampled
specifically for the GA and SC plantations.
be native in origin. Identities of seeds and progenies were maintained ac-
cording to female parent.
The seeds were distributed to each cooperator, who grew his own seed-
lings for outplanting. Plantations SC and GA (see below) came from a common
set of nursery stock. The description of each plantation is as follows:
Centre County, Pennsylvania (PA). Planted April 1981 with 1-1 stock in 8
randomized blocks of 33 provenances in 4-tree row plots. Each provenance
plot consists of four l-tree family plots (compact family design). Spacing
is 2.4 x 2.4 meters between trees. Soil is a Hagerstown silty clay loam
(upland); site previously in crops, plowed and disked prior to planting and
cultivated for three years afterward.
Aiken County, South Carolina (SC). Planted January 1980 with 1-0 stock in 6
randomized blocks of 26 provenances in 4-tree row plots as in PA plantation.
Spacing is 1.2 meters between trees in rows and 2.4 meters between rows.
Soil is a Fuquay sandy loam (upland); site previously clearcut of pine and
the slash windrowed and burned, cultivated for two years after planting and
fertilized in second year with 560 kg/ha of 10-10-10.
249
Putnam County, Georgia (GA). Planted May 1980 with 1-0 stock in 4 randomized
blocks of 26 provenances in 4-tree row plots as in PA plantation. Spacing is
2.4 x 2.4 meters between trees. Soil is a Vance sandy loam (upland); site
previously clearcut of mixed pine and hardwood and root-raked prior to
planting, mowed for three years afterward and fertilized in second year with
560 kg/ha of 10-10-10.
Massac County, Illinois (IL). Planted April 1979 with 1-0 stock in 4 random-
ized blocks of 83 families in 8-tree row plots, the families representing 31
provenances. Family plots were aggregated by physiographic region, a group-
ing ignored for the present analysis. Spacing is 3.4 x 3.4 meters between
trees. Soil is a Sciotoville silt loam (bottomland terrace); site clearcut
in 1977 and the slash burned (ash deposits not planted), disked prior to
planting and cultivated for three years afterward.
Each plantation was evaluated for height at the end of its fourth grow-
ing season in the field. Height data from the PA, SC, and GA plantations
were subjected to analysis of variance for provenance effects. Because of
the compact family designs in those plantations, family effects were examined
by separate analysis of variance for each provenance. Sums of squares and
degrees of freedom for family and error terms were then pooled across prove-
ances to get an overall estimate of the significance of family-within-prove-
nance effects.
For the IL plantation, a separate analysis of variance for provenance
and family-within-provenance effects was performed for each physiographic
region by which field plots were grouped, and the sums of squares and degrees
of freedom were pooled across regions. Sums of squares and degrees of free-
dom for regional effects, from an analysis of all the data, were combined
with those for provenance effects.
Block, provenance, and family were treated as random effects in all
analyses. Variance components were calculated for each effect as follows,
using mean squares from the pooled analyses:
Provenance VAR, FR /P + (b) VAR, 1p + (£) VAR, p + (bf) VAR,
Family/Provenance VAR F/P + (b) VAR, 7p
Block x Provenance VAR, ee /P + (£) VAR, -p
Block x Fam./Prov. VAR, F/P
Because provenances in the IL plantation were represented by variable numbers
of families, component coefficients for that plantation were generated using
the VARCOMP procedure of SAS (SAS 1982).
No analysis of variance across all plantations was performed because of
the differences in experimental design. Instead, provenance means in each
plantation were standardized (by subtracting plantation mean and dividing by
standard deviation of provenance means) and provenance contributions to prove-
nance x plantation interaction sum of squares were calculated as follows for
each of the six pairs of plantations:
250
mn
Nn
i}
| mn
=
*
I
~
VY
ined
ww
where x,, = mean of provenance "i" in plantation "j" and x
af dite
nance "i" across both plantations. This is Wricke's "ecovalence" formula
(Shelbourne 1972), but simplified because plantation means using standardized
data are 1.0. The use of standardized provenance means eliminates contribu-
tions to the interaction that are purely a function of scale as a result of
provenances being more variable in some plantations than others. It also
enables the comparison of contributions for a given provenance across planta-
tion-plantation combinations.
= mean of prove-
RESULTS AND DISCUSSION
Mortality after four growing seasons was low in all plantations (SC -
2.64, GA - 5.3%, PA - 7.8%, and IL - 14.6%). Differences among provenances
were generally minor, but there were two apparent trends. Provenances native
within 300 km of the IL plantation site had slightly lower mortality (x =
11.1%) at that location than those from more distant locations (x = 17.2%).
It is obviously not a strong difference, but one which will be worth watching
as the plantations develop. In the PA plantation, trees of southern origin
have been repeatedly winter-injured, and this is beginning to show up in mor-
tality. Provenances native south of latitude 35° had higher mortality Eee
15.3%) than those of more northern origin (x = 4.9%), and the difference will
probably increase with time.
Plantation mean heights varied from 1.4 m in PA to 3.1 m in IL (Table
1). Provenance was a significant source of variation in all plantations
(Table 2), and the best provenance in each grew 10 to 20 percent faster than
the plantation mean. In general, the best provenances at each plantation
were native to locations distant to the plantation site. The tallest 10 per-
cent of the provenances at PA were native to Illinois and Indiana; and at IL
the tallest trees were native to North Carolina, Illinois, and Georgia (Table
1). For these two plantations, there was a definite growth advantage in prov-
enances of somewhat more southern origin than the plantation site. In the
case of the PA plantation at least, this advantage was associated with no sac-
rifice in hardiness, since southern Indiana and Illinois trees appear at this
time to be as hardy as those from coastal New Jersey and Pennsylvania.
The tallest 10 percent of the provenances at SC were native to
Mississippi and Alabama; and at GA, to Mississippi and North Carolina (Table
1). Two of the three provenances involved in each case came from milder, more
coastal locations than the respective plantation sites. Mississippi and
Alabama sources were consistently superior at GA and SC and showed a 7 or 15
percent average height advantage over Georgia and South Carolina provenances.
Except at GA and SC, there was little positive correspondence between
provenance means at different plantations (Table 3). IL means showed only a
very weak positive correspondence with those at all three other locations.
PA means were significantly and negatively correlated with those at the two
southernmost plantations, probably as a result of the interrelationships
251
TABLE 1.--Relative heights (percentage of plantation mean) of 51 provenances
after four growing seasons in four plantations.
a
Provenance Plantation
Number State PA SC GA IL
001 GA -- 107.9 106.8 --
006 GA -- 105.8 99.2 --
011 GA -- 109.4 102.7 --
Olo MS = 16 120.1 --
033 GA 87.5 101.3 98.8 --
053 NC G2 107.4 115.0 =—
061 SC 85 ye03) -- -- --
070 KY ~— -- -- OB 2:
077 MD 95:6 -- -- --
081 AL 80.2 116.5 109.8 --
085 MS 9055 102.0 Wie Sr ==
097 IL 114.0 Od 99.4 --
101 TN 100.8 109.7 97.8 --
105 TN 97.8 -- -- --
109 TN 100.8 -- -- --
117 TN 90.5 D769) 96.2 104.0
121 IN 119.1 92.5 92.0 NOES 7/
125 IN 116.9 -- -- 103.4
129 KY 102.2 96.7 101.4 --
133 OH -- 96.3 Bishaill 100.2
137 MO -- -- -- 94.4
229 MS 89.7 NGS 112.4 --
241 sc 86.0 50 95.2 --
245 MO 98.5 -- -- --
257 GA, 96.3 104.0 95.7 108.2
261 cv -- -- -- 96.7
281 PA -- 98.2 100.4 --
284 NJ -- -- -- 107.8
285 PA LSS) deat URS) --
289 TN -- -- -- 106.6
297 NJ 109.6 -- -- 93.8
313 NJ 106.6 853 Oia 395
325 NJ -- -- -- 86.8
329 TN -- -- ~- 101.8
333 TN -- -- -- 95.1
Si, IL 119.9 -- -- 108.2
381 VA 90.5 103.6 85.8 98.3
385 VA » 90.5 100.5 107.7 103.7
420 DE -- -- -- 90.6
433 NC -- -- -~ 110.4
445 AL 100.8 -- -- 96.3
452 KY -- -- -- 100.2
457 AR 86.8 -- -- 89.0
461 PA, 111.8 -- -- 94,8
473 WGN, 91.2 81.5 83.7 102.1
481 TL 116.9 -- -- 100.2
489 MO 114.0 -- -- 103.4
545 TN 102.2 91.8 99.4 102.1
549 TN 100.0 103.6 104.6 104.3
553 TN 100.8 102.0 98.3 104.3
557 TN -- -- -- as)
Plantation mean: 1.36 m 2.63 m 1.91 m Sous fa
‘Cultivated origin.
252
TABLE 2,.--Variance components for age 4 height at four plantations.
Component (% of total less "block") for:
Source PA SC GA IL
Provenance 11. 1**% 10, 4%**% 8.2% 12,3**
Family/Provenance 4 .0** 3.5% 19 5.4
Error! 84.9 86.0 89.8 823
1Pooled over "block x provenance" and "block x family-within-
provenance".
*, **k, ***k Effect statistically significant at P < 0.05, P < 0.01,
and P < 0.001, respectively.
between growth potential, cold tolerance, and latitude of origin. For age l
performance of the same material in plantations in West Virginia and Maryland,
Prowant et al. (1983) documented a negative relationship between latitude of
origin and annual growth increment, but a positive relationship between lati-
tude and height as a result of winter dieback on southern trees in the nursery.
Provenance x plantation interactions for the six plantation-plantation
combinations are shown more clearly in Table 4. Provenances that contributed
most to interactions involving PA and the two most southern plantations were
of either extreme northern or extreme southern origin, an obvious consequence
of the slow growth of the former in GA and SC and winter injury to the latter
in PA. To a degree, the same situation occurred in the PA and IL comparison.
TABLE 3.--Coefficients of correlation between provenance
mean heights at four plantations.
SC GA IL
PA —0.57** -0.44* +0.20
SC -— +0 .68** +0.38
GA -- -- +0.50
*, **Statistically significant at P < 0.05 and P < 0.01,
respectively.
253
TABLE 4.--Provenance contributions to provenance x plantation interaction
sum-of-squares in six plantation/plantation comparisons.
a
Provenance
No. State
033 GA
053 NC
081 AL
085 MS
097 108
101 TN
ney, TN
121 IN
125 IN
129 KY
E33 OH
229 MS
241 Sc
25) GA
281 PA
285 PA
297 NJ
Suis} NJ
SY 1
381 VA
385 VA
445 AL
457 AR
461 PA
473 CV
481 IL
489 MO
545 TN
549 IN
393 IN
1, "4" indicates provenances that grew relatively better in the first
PAG toe
(-)0.709
(-) 1.067
(-)5.930
(-)0.488
(+) 2.225
(-)0.306
(-)0.137
(+)3.690
(+)0.229
(-)2.056
(-)0.281
(-)0.193
(+)7.522
(+)2.557
(-)0.653
(-)0.347
(+)0.692
(+)0.698
(-)0.027
0.000
plantation listed, a
PAs GA
(-)0.416
(-)2.184
(-)3.508
(-)2.896
(GP) he 222
(+)0.101
(-)0.074
(+)3.766
(+)0.030
(-) 1.967
(-)0.250
G)02025
(+)8.284
(+)0.789
(+)0.181
(-)1.080
(+)0.377
(+)0.097
(-)0.034
(+)0.080
provenances that grew better in the second.
PAY +
(-)1.549
(+)0.379
(+)0.272
(-)2.121
Ge) 13972
(+)0.167
(+)0.001
(-)0.189
(-)1.439
1(+)0.236
(+)0.404
(+) 1.986
(-)0.862
(+)0.942
GCG) Oo 112
(-)0.053
(-)0.468
(-)0.407
SC + GA
(+)0.029
(-)0.221
(+)0.295
(-)1.063
(-)0.173
CHiOanZ
(+)0.005
0.000
(-)0.112
(+)0.740
0.000
0.000
(+)0.330
(-)0.024
(+)0.010
(-)0.555
(+)1.465
(-)0.228
(-)0.064
(-)0.307
(-)0.002
(+)0.068
sc + IL
(-)0.043
(-)0.336
(+)0.454
(-)0.461
(-)0.031
@) 35355
(+)0.024
(-)1.402
(-)0.057
(+)0.074
(+)0.017
Contributions to sum-of-squares! for plantation combinations:
GA + IL
(-)0.035
(-)0.224
(~)0.123
(-) 1.693
(+)0.758
(+)0.138
(+)0.820
(-)0.636
(+)0.338
(+)0.217
(-)0.007
However, interactions involving the IL plantation were also distinguished by
the almost consistently superior performance in IL of provenances from the
Cumberland Plateau and associated highlands near southeastern Tennessee, and
one collection (473) of cultivated origin.
were small.
Interactions between SC and GA
Family-—within-provenance was a significant source of variation in two of
the four plantations (Table 2).
Depending upon plantation, family variance
components were 23 to 44 percent as large as provenance components, with an
254
average of 34 percent. This compares closely with an analogous figure of 27
percent that can be calculated from Sprague and Weir's (1973) ANOVAs for age
four height in ten plantations containing overlapping sets of 10 to 12 stand
collections each represented by five open-pollinated families. Wells et al.
(1979) also found significant within-stand variation in growth rate of proge-
nies from 138 stands predominantly in Mississippi.
To determine whether some provenances were consistently more variable
than others, we performed separate ANOVAs for family effects in each prove-
nance at each plantation (Table 5). No provenance exhibited significant
family variation at more than one location, and in fact there was hardly any
consistency across plantations in the relative size of the family mean
squares for each provenance. In other words, the expression of within-popula-
tion variation was too inconsistent from site to site to permit generaliza-
tion, and family x plantation interactions would probably have been large if
we had analyzed for them. Of course, this has little practical import because
provenance selection would preclude most opportunities for the selection of
identical families for two or more of these plantation locations, except
perhaps GA and SC.
CONCLUSIONS
After four years in field plantings, best growth was generally obtained
on provenances native fairly large distances from the plantation site. For
the respective plantations, the fastest growing trees originated as follows:
PA -- southern Illinois and Indiana
IL -- Cumberland Plateau and associated highlands in Georgia,
North Carolina, and Tennessee, and one provenance each in
New Jersey and Illinois.
sc =
Mississippi, Alabama, and Tennessee
GA -- Mississippi, Alabama, and North Carolina
Whether these patterns will persist, and whether such provenance transfers
would entail some risk in adaptation, will require further study to determine.
The results must be regarded with caution because vigorous height growth is
just beginning to occur in the plantations.
There was little consistency in provenance performance except between SC
and GA. The only interpretable interactions were those attributable to win-
ter injury to southern provenances in contrasts between PA and the two most
southern plantations.
Although family-within-provenance was a significant source of variation
in only two plantations, it consistently accounted for at least 23 percent as
much height variation as provenance. This is especially remarkable consider-
ing the fact that provenance representation was nearly range-wide. Cooper
(cited in Wells 1979) has shown no advantage to plus-tree selection in this
species. Consequently, sweetgum improvement programs should incorporate both
provenance selection and progeny testing of wild parents.
255
TABLE 5.--Mean squares for family-within-provenance effects, by provenance
and plantation.
Provenance
Number State
33 GA
53 NC
61 SC
Taf MD
81 AL
85 MS
N/ IEE
101 TN
105 TN
109 -IN
nL, TN
121 IN
125) IN
129 KY
K33 OH
229 MS
241 SC
245 MO
Dy GA
261 CV
281 PA
285 PA
297 NJ
313 NJ
333 TN
377 IgG,
881 VA
385 VA
445 AL
457 AR
461 PA
473 cv
481 IL
489 MO
ByYAS) TN
549 TN
5)5)5) TN
RK,
respectively.
PA
0.3980
ORS
0.6144*
0.2510
0.4161
0.2412
OS 2739
0.4043
0.7411
O97,
0.7160%
0.2859
0.2505
1,0624**
0.1052
0.3670
0.2634
1 .0633*
0.1680
0.3408
0.0279
0.2336
0.5248
0.4521
OLS Li3
0.7143
0.3220
0.4605
0.5267
0.2564
0.1074
0.0469
0.0587
Plantation
SC GA
0.0175 0.0563
0.8886** 0.1383
0.3164 0.1441
0.3220 0.3886
0.2099 0.2597
0.0853 0.0675
0.1641 0.0020
0.0531 0.0653
0.1262 0.1285
0.1278 0.1557%*
0.2764% 0.1978
0.4867 0.1193
085238 0.0290
0.0195 0.0903
OR0318 0.0518
0.3697 0.1623
0.2140 0.1449
0.2938 0.2827
0.0247 0.1872*
0.2765 0.0639
(OG AAAS) 0.0475
0.0561 0.1338
0.0005
0.1329
0.0048
0.0140
0.1625
0.1051
0.0840
0.4608
0.0009
0.1826***
0.2174
0.1097
0.2270
0.0918
0.1537
0.1617
*** Statistically significant at P < 0.05, P < 0.01, and P < 0-001,
LITERATURE CITED
Mohn, C. and D. Schmitt. 1973. Early development of open pollinated
sweetgum progenies. Proc. South. Forest Tree Improv. Conf. 12: 228-232.
Prowant, J. S., F. C. Cech, R. N. Keys, and W. H. Davidson. 1983. A
comparison of a range wide study of sweetgum planted on three diverse
sites. Proc. Northeast. Forest Tree Improv. Conf. 28: 50-59.
SAS Institute Inc. 1982. SAS user's guide: statistics, 1982 edition. SAS
Institute Inc., Cary, North Carolina. 584 pp.
Shelbourne, C. J. A. 1972. Genotype-environment interaction: its study and
its implications in forest tree improvement. Proc. Joint Symp. of Genetics
Subj. Group, IUFRO, and Sec. 5, Forest Trees, SABRAO, Govt. Forest Exper.
Stay LokyonweBol @)m28 pps
Sprague, J. and R. J. Weir. 1973. Geographic variation of sweetgum. Proc.
South. Forest Tree Improv. Conf. 12: 169-180.
Texas Forest Service. 1975. 23rd progress report of cooperative forest tree
improvement program. Texas Forest Service, Texas A & M Univ., College
Station, 25? pp.
Wells, 0. O., G. L. Switzer, and W. L. Nance. 1979. Genetic variation in
Mississippi sweetgum. Proc. South. Forest Tree Improv. Conf. 15:
22-32 e
257
GENETIC PARAMETERS FOR TWO EASTERN COTTONWOOD
POPULATIONS IN THE LOWER MISSISSIPPI VALLEY
G. Sam FosterL/
Abstract.--Genetic variances and heritabilities were
compared between samples of two populations of eastern
cottonwood tested on adjacent sites. Data for fourth-year
growth and second-year survival yielded little difference
between families in either population with most of the
genetic variation associated with clones-within-families.
Resultant estimates of additive genetic variance were low
with much higher estimates of dominance variance.
Consequently, narrow-sense heritabilities ranged from 0.00 to
0.27, and broad-sense heritabilities ranged from 0.01 to
0.45. A more efficient future test design includes smaller
blocks and noncontiguous family and clonal plots.
Reforestation of eastern cottonwood (Populus deltoides Bartr.) has
generally utilized clonal planting stock. Consequently, tree improvement
efforts have been focused primarily on testing and selecting clones with
little emphasis on recurrent selection programs. Without genetic
recombination, a clonal selection program will eventually reach a plateau
beyond which no further genetic gain can be obtained.
A recurrent selection program provides an opportunity for continuing
advancement in gain over time; but to be efficient, breeders need
estimates of genetic parameters for use in program planning. Estimates
of additive genetic variance (Farmer and Wilcox 1966, Farmer 1970, Cooper
and Randall 1973, and Ying and Bagley 1976), mnarrow-sense heritability
(Farmer and Wilcox 1966), and dominance variance (Cooper and Randall
1973) have been published. However with the exception of one
seven-year-old study (Ying and Bagley 1976), the reports have all
described one and two-year-old data.
In this report, genetic parameter and heritability estimates are
presented using second-year survival and fourth-year growth data from two
populations of eastern cottonwood. Recommendations on future cottonwood
test design are also made.
MATERIALS AND METHODS
Populations
Population 1.--Female parent trees were chosen in stands growing
near Lake Albemarle, just north of Vicksburg, Mississippi. Criteria for
ly Manager-Forestry Research/Research Forester-Tree Improvement, Crown
zellerbach Corp., Bogalusa, LA 70427. The author would like to
acknowledge the work of Dr. Tom Cooper, recently of the USFS, and the
South. For. Exp. Stn. for installing both studies.
258
selection included straightness and general appearance compared to
neighboring trees. Open-pollinated seeds were collected and sown in a
replicated nursery in July, 1971. The seedlings were cut back each
following winter. In the spring of 1977, the best of the surviving trees
were cloned and planted in another nursery. After being cut back in
1978, the clones with good survival and growth were again cloned and
planted in a new nursery.
On February 15, 1980, clones were established in a field trial on
Crown Zellerbach Corporation land near Fitler, Mississippi. The trial
included 15 families with an average of 9.8 cloned individuals per family
(range of 7 to 15; 147 clones total).
Population 2.--Female parent trees were chosen from stands in
western Tennessee, 150 miles north of Stoneville, Mississippi. Using the
same criteria as in Population 1. Open-pollinated seeds were collected,
and seedlings were grown in a nursery in 1978 and cloned into a nursery
in 1979.
The field study was planted on February 15, 1980 on a site adjacent
to the test for Population 1. fThe trial included 17 families with an
average of 6.2 cloned individuals per family (range of 1 to 17; 105
clones total).
Experimental Design
The experimental design consisted of five replications of two-tree
plots planted at a 12xl2 ft. spacing. The design employed a compact
family block configuration in which families were arranged as randomized
complete blocks with cloned individuals randomized within their
respective families.
Two unrooted, 20 inch cuttings were planted at each planting spot.
During the second growing season, survival was recorded; and if more than
one tree survived per spot, then the second one was cut.
The test received standard cultural maintenance with several
diskings during the first growing season (personal communication, Pat
Weber, Fitler Managed Forest, Crown Zellerbach Corp.).
Analyses
Three traits were measured in each test including: second-year
survival of the two cuttings planted at each planting spot, total height
(ft) at age four, and d.b.h. (inch) at age four. Merchantable tree
volume (to a three inch top) was calculated using equation 4 from Mohn
and Krinard (1971).
The analysis of variance (Table 1) utilized plot mean data and
employed a least-squares procedure due to imbalance of
clones-in-families. Variance components were calculated by equating mean
Squares with expected mean squares, and coefficients of variance
components were adjusted for the data imbalance (Searle 1971).
The calculation of narrow-sense and broad-sense heritabilities
259
utilized standard formulas’ (Sorensen and Campbell 1980, Foster et al.
1984). Estimates of additive and dominance variance were derived by
equating variance components to their genetic expectations (Bohren et al.
1965). Nonadditive variance was assumed to be due solely to dominance
variance. The ratio of dominance variance to phenotypic variance
provided a measure of its relative importance.
Table 1.--Form of the analysis of variance for growth and survival traits
for two populations of eastern cottonwood
Source Dab i) Expected mean squares?/
Families(F) (£-1) ao + ra C/F + ca FR + rca F—
2 Des 2
Replications(R) Cr=1) a + ca FR + cfa R
2 2
FXxR GESR Gre) a + ca FR
2 D
Clones(C)/F (c=L)E a + ra C/F
2
Error remainder a
Total faire
&/ Population 1: r = 2.9; c = 9.3; £ = 14.0
Population 2: r = 3.9; c = 5.7, (5.9 for survival); £ = 9.0
b/ synthetic F test (Cochran 1951)
Uniformity of family-mean performance between replications was
estimated by calculating appropriate correlation coefficients.
RESULTS
Population 1
Tree growth in this study was considered good for the test site
conditions for the first three replications, but replications four and
five were inadvertently located on very wet areas and suffered from low
survival. For this reason, further analyses refer only to the first
three replications. Total height, d.b.h., survival and volume averaged
41.3 ft., 5.4 inch, 78.1 percent, and 2.0 cu. ft., respectively.
Analysis of variance results were similar among the four traits. No
significant variation occurred among families for any of the traits
(Table 2), with family variation accounting for a very small proportion
(0.0 to 1.0 percent) of the total variation (Table 3). Replication
effects were significant for all traits (p = 0.10) (Table 2) and
accounted for 2.5 to 12.9 percent of the total variation (Table 3). The
replication x family interaction was surprisingly large and significant
for all traits (Table 2), representing an average of seven percent of the
total variation (Table 3). For survival, d.b.h., and volume, this
interaction equaled or exceeded the replication effect in importance.
Considering height, d.b.h., and volume, the most important effect, except
error, in the analysis arose from clones-in-families (Table 2)
260
representing 27 percent of the total variation (Table 3). No differences
occurred among clones-in-families for survival.
Table 2.—-Mean squares and F tests for the analysis of variance for
Population 1 of eastern cottonwood
Mean Squares
Source Deke Height DeBeHe Survival Volume
Families(F) 14 40.98NS 1.74NS 899.27NS 1.35NS
Replications(R) Dee 315 On 3.491 4789.03" " 57.
FXR 28 2A\.78>" Paldies 836 .63* 0.76%
Clones(C)/F 132 PAE a alien 501.89NS idle
Error Prowl 10.37 0.48 489.85 0.45
1 Significant at p = 0.10
Significant at p
Significant at p = 0.01
NS Nonsignificant at p = 0.10
il
Oo
e
oO
oa)
xk
The large replication x family interaction manifested itself in the
family-mean correlations between replications. While the correlation
between replications one and two was significant (p = 0.05) and positive
(0.65), the correlations between replications one and three (-0.32) and
two and three (-0.25) were nonsignificant.
Given nonsignificant family effects and highly = significant
clones-in-family effects (except for survival), estimates of additive
variance were small and nonsignificant (Table 3); estimates of dominance
variance were significant and apparently accounted for all the genetic
variance (Table 3).
Heritability estimates reflect the importance of genetic variance to
phenotypic variance; therefore the trends for heritabilities followed
previous results for variance components. Narrow-sense heritability for
height was 0.05, for survival was 0.01, and 0.00 for the other two traits
(Table 3). Broad-sense heritability ranged from 0.25 to 0.33 for height,
d.b.h., and volume and equaled 0.01 for survival (Table 3). Broad-sense
heritability based on clone-means ranged from 0.51 to 0.60 for height,
d.b.h., and volume (Table 3).
Dominance variance as a proportion of phenotypic variance equaled
broad-sense heritability for d.b.h. and volume and was only slightly
lower in the case of height (Table 3).
Population 2
Although tree growth in Population 2 did not equal that of
Population 1, it was still acceptable for the site. Average fourth-year
height, d.b.h., survival, and volume reached 38.1 ft., 5.1 inch, 77.2
percent, and 1.5 cu. ft., respectively. Correlations of family-means
261
Table 3.--Variance components and genetic parameter estimates for
Population 1 of eastern cottonwood
ree RE OL
Estimate
Parameter Height D.B.H. Survival Volume
oF O18) ¢ oy 0.00 ( 0.0) 1388)" (063) 000) (C4820)
BR 2536) (29) 0702" 225)= 30.236 ( 5.4)) 0203 Gano)
o7FR 1.55 ( 8.5) 0/07. 4.629). 37 29u 66) O03 mn Graaay
a°C/F 3.88 (21.2) OF 225 eC27/29)) CVE CORI) WEP ee GS Bo))
oa 10.37 (56.4) 0.48 (60.7) 489.85 (87.0) 0.45 (60.0) 4
v,a/ 0.72 0.00 USL 0.00
V_b/ 3.34 0.22 0.00 0.24
ee 0.05 0.00 0.01 0.00
Hw 0.25 0.29 0.01 0.33
H2 c/ 0.51 0.55 0.03 0.60
Vp OF 21: 0.29 0.00 0.33
phen. |
variance
a/ Va = additive genetic variance = 4 (a2F)
b/ vy = dominance genetic variance = o2C/F-(3) (a2F)
c/ y2 = broad-sense heritability of clone means
x
d/ variance components as a percent of total variation
|
between replications (although nonsignificant in most cases) were positive)
except with replication two (Table 4). The importance of this result will be’
discussed fully later, but it was used as rationale to delete replication two
from further analyses.
Table 4.--Correlation coefficients between replications based on family
means of eastern cottonwood
a
| Replications
Replications | 2 3 4 5
|
1 | -0.03NS 0.35NS 0.15NS 0.67""
2 | -0.40NS -0.13N5 -0.17N5
3 | 0.32NS 0.41NS |
4 | 0.17NS |
ee Nonsignificant at p = 0.10
Significant at p = 0.01
262
The analyses of variance for Population 2 followed a somewhat
different pattern than for Population 1. Differences among families
achieved significance (p = 0.10) for height (Table 5) and accounted for
Table 5.—-Mean squares and F tests for the analysis of variance for
Population 2 of eastern cottonwood
Mean Sguares
Source DoF Height D.B.H. Survival Volume
Families(F) 16 49.501 1.68NS 1397.18NS 1.73NS
Replications(R) 3 63.90** 0.40NS 3006.03” 0.991
FxR 48 ks 30" 0.421 903.57** 0.41”
Clones(C)/F 88 242407- 133°" 963.94** wagr*
Error 251a/ 7-20 ee 488.07 0.29
1 Significant at p = 0.10
* Significant at p = 0.05
Significant at p 0.01
NS nonsignificant at p = 0.10
a’ p.F. for survival = 264
w*
six percent of the total variation (Table 6). The family source of
variation for the other three traits was nonsignificant (Table 5) and
accounted for little of the total variation (0.1 to 3.5 percent of the
variation) (Table 6). Replication effects were significant for height
and survival (p = 0.05) as well as volume (p = 0.10) but not for d.b.h.
(Table 5). Variation represented by replication effects ranged from 0.0
to 7.0 percent (Table 6). Family x replication interaction was again
significant for all traits (Table 5) and accounted for an average of 5.4
percent of the total variation (Table 6), a smaller percentage than in
Population 1. Clones-within-families clearly exceeded all other sources
of variation in importance. It was highly significant (Table 5) and
contributed an average of 32.8 percent of the variation (Table 6).
The family x replication interaction appears to be largely due to
the unusual family rankings in replication two (Table 4). Replications
one, three, four, and five are positively correlated (though only
replications one and five were significantly correlated), while
replication two was clearly an outlier. With replication two in the
analyses, the family x replication interaction represented an average of
9.7 percent of the variation.
Though still nonsignificant for three of the four traits, additive
genetic variance estimates were all positive (Table 6). Dominance
variance clearly represented the major proportion of the total genetic
variation for d.b.h., survival, and volume; while additive variance for
height was double the dominance variance.
With one exception, narrow-sense heritabilities were still quite
small, and broad-sense heritabilities considerably exceeded the
263
Table 6.--Variance components and genetic parameter estimates for
Population 2 of eastern cottonwood
Estimate
Parameter8/ Height D.B.H. Survival Volume
o2F HED (GSO Oey CE We) 01.575) LO: ng BOD aS)
a2R Wao (E950) O200. © O50)) 39259 (5-5) | O01 Gales)
a2FR OS70 2 ait51 0) 0/202 arses) 70-42 © 98) "0202" 1@se5)
o2C/F 4@240'° (31.0)" “0s26. (43883) 118-97" “Q6.6), 0223" 4C4043)
o2 7-220" (5120)= —-0e 31 (5127) — 488207" (68-0) = 029 ==(5069)
Va 3.60 0.04 3.00 0.08
V 1.70 0223 116.72 0.17
h 0.27 0.07 0.004 0.14
H2 0.40 0.45 0.18 0.45
HZ 0.73 0.77 0.45 0.76
x
Vp 0.13 0.43 On17 0.30
phen.
variance
a/ Parameter symbols explained in Tables 1 and 4.
narrow-sense. Narrow-sense heritability for height equaled 0.27; it
ranged from 0.004 to 0.14 (Table 6) for the other three traits.
Broad-sense heritabilities averaged 0.37 with a range of 0.18 to 0.45
(Table 6). Clone-mean heritabilities ranged from 0.45 to 0.77 (Table 6).
The ratio of dominance variance to phenotypic variance exceeded the
ratio of additive genetic variance to phenotypic variance (narrow-sense
heritability) for all traits but height (Table 6).
DISCUSSION
Family differences achieved significance only for height in
Population 2, with nonsignificant variation for the other traits in both
populations. The low amount of family variability led to either no
additive genetic variation or a small amount, at best, for the traits.
These results were unexpected based on the significant findings of Farmer
and Wilcox (1966), Farmer (1970), and Ying and Bagley (1976).
Clones-within-families comprised the major portion of the genetic
variation in these two studies. Ying and Bagley's (1976) results also
concurred that, for growth traits, clones-within-families comprised a
larger proportion of the total variation than families. In the present
study, the clonal variation derived mainly from dominance genetic effects
rather than additive genetic effects for all traits except survival in
Population 1 and height in Population 2. Cooper and Randall (1973) found
that additive genetic variance accounted for three times the level of
dominance variance for first-year height and one-fifth the level of
dominance variance for first-year survival.
264
Three possibilities exist for the results of this study compared to
earlier studies. The findings may be real but unique (compared to
earlier studies) for the two sampled populations. Intra-locus genetic
interactions (dominance) may actually be the major cause of genetic
variation in these populations. Selection pressure in these populations
may favor survival of heterozygous individuals with very similar
genotypes. The earlier studies cited above sampled many more populations
and therefore had a greater chance of sampling ones with significantly
different gene frequencies.
The second possibility is that these results are an artifact of
analyzing data only for fourth-year growth traits and second year
survival. Only one of the cited studies in the literature examined data
for a range of ages (Ying and Bagley 1976) while the others examined data
only for first and second-year traits. Ying and Bagley's fourth-year
analysis agreed that a larger proportion of variation was due to
clones-within-families compared to families; but families were still a
significant source of genetic variability.
The last explanation relates to the large microsite variability in
this flood-prone test site and the experimental design. The area is
situated behind a levee and is not subjected to major river flooding but
still receives regular backwater flooding from the Mississippi River.
Undoubtably though, the soil profile originally resulted from alluvial
deposits from the river and is characterized by ribbons of fairly
different soil types (Wynn et al. 1961). The topography is slightly
undulating and water pools up in the low spots following rains or
Flooding. The interaction of these site factors yields a large amount of
microsite variability. Block sizes of 0.7 to 1.0 acre were probably too
large and included too much within-block variability. Incomplete block
designs (i.e., as described by Schutz (1966) and Libby and Cockerham
(1980) hold promise for reducing block size thereby increasing efficiency
of test results. In addition as Lambeth et al. (1983) demonstrated,
contiguous family plot configurations (as compared to noncontiguous
plots) cause larger block-by-family interactions and larger coefficients
of variation for family means. A compact family plot design was used in
this study as well as (contiguous) row plots for clones-within-families
which probably contributed to the high family x replication interaction
and nonsignificant differences among families. The efficiency of future
tests of this type could be increased by using smaller blocks and a
noncontiguous configuration of both clones-within-families and
ramets-within-clones.
Results from this study as well as others (Ying and Bagley 1976)
demonstrate the larger importance of clone-within-family variability
compared to family variability. A tree improvement program should be
designed which, while taking advantage of additive genetic variation
through family selection, lends major emphasis to clone-within-family
selection thereby tapping the large amount of dominance variance. One
alternative includes a main line program emphasizing family and
within-family selection for gains from additive genetic variation while
in each generation utilizing a production population derived mainly from
pure clonal selection.
265
LITERATURE CITED
Bohren, B. B., McKean, H. E., and Friars, G. W. 1965. The expected mean
squares in genetic experiments when only one parent is identified.
Biometrics 21:436-446.
Cochran, W. G. 1951. Testing a linear relation among variances.
Biometrics 7:17-32.
Cooper, D. T., and Randall, W. K. 1973. Genetic differences in height
growth and survival of cottonwood full-sib families. Proc. 12th South.
For. Tree Imp. Conf., p. 206-212.
Farmer, Jr., R. E. 1970. Genetic variation among open-pollinated
progeny of eastern cottonwood. Silvae Genet. 19:149-151.
Farmer, Jr., R. E., and Wilcox, J. R. 1966. Variation in juvenile growth
and wood properties in half-sib cottonwood families. In Joint Proc.,
Second genetics workshop of the Soc. Amer. Forest. and the seventh Lake
States For. Tree Imp. Conf., p. 1-4. USDA For. Serv. Res. Pap. NC-6.
Foster, G. S., Campbell, R. K., and Adams, W. T. 1984. Heritability,
gain, and C effects in rooting of western hemlock. Can. J. For. Res.
14:628-638.
Lambeth, C. C., Gladstone, W. T., and Stonecypher, R. W. 1983.
Statistical efficiency of row and noncontiguous family plots in genetic
tests of loblolly pine. Silvae Genet. 32:24-28.
Libby, W. J., and Cockerham, C. C. 1980. Random noncontiguous plots in
interlocking field layouts. Silvae Genet. 29:183-190.
Mohn, C. A., and Krinard, R. M. 1971. Volume tables for small
cottonwoods in plantations. USDA For. Serv. Res. Note SO-113, 4 p.
South. For. Exp. Stn., New Orleans, LA.
Schutz, W. M., and Cockerham, C. C. 1966. The effect of field blocking
on gain from selection. Biometrics 22:843-863.
Searle, S. R. 1971. Linear models. N. Y.: John Wiley and Sons, Inc.
Sorensen, F. C., and Campbell, R. K. 1980. Genetic variation in
rootability of cuttings from one-year-old western hemlock seedlings.
USDA For. Serv. Res. Note PNW-352, 8 p. Pac. N. W. For. Range Exp.
Stn., Portland, OR.
Wynn, Jr., A. H., Alcorn, D. E., and Parker, R. L. 1961. Soil survey of
Issaquena County, Mississippi. USDA Soil Conser. Serv. Series 1959,
No. 2, 42 p.
Ying, C. C., and Bagley, W. T. 1976. Genetic variation of eastern
cottonwood in an eastern Nebraska provenance study. Silvae Genet.
256i 3s
266
GENETIC VARIATION AMONG OPEN-POLLINATED FAMILIES OF BALDCYPRESS SEEDLINGS
PLANTED ON TWO DIFFERENT SITES
Patricia Faulkner, Furcy Zeringue, and John Toldvert!
Abstract.--After two years of growth on two different sites
in south Louisiana, baldcypress seedlings averaged 126.3 cm in
height and 2.05 cm in diameter. Available soil moisture signi-
ficantly influenced seedling growth between the two sites, with
the wetter site producing the largest seedlings. Geographic
variation was not found. However, family-within-source variation
was significant for both height and diameter.
Additional keywords: crawfish, geographic variation, Procambarus
clarkii, Taxodium distichum.
Baldcypress [Taxodium distichum (L.) Rich.] is an important commercial
species in the swamps and bottomlands of the southern and southeastern United
States. Throughout the South, there is an estimated 5.5 billion feo. Glo5e7
million m3) of baldcypress growing stock on 3 to 5 million acres of commercial
timberlands (Williston et al. 1981). Much of this timber will reach merchant-
able size within the next 30 years. Baldcypress is adapted to permanently or
periodically flooded sites that are difficult to restock by natural regenera-
tion. Before existing stands of cypress are harvested, an alternative method
of regeneration must be investigated to improve cypress resources for future
demands. Planting of baldcypress seedlings is a viable alternative, and nu-
merous successful plantings have been reported (Bull 1949, Foil and Merrifield
1966). Researchers at Louisiana State University are studying the genetic
variation of baldcypress in an effort to enhance knowledge of regeneration and
management techniques.
METHODS
Seedlings from 26 half-sib baldcypress families representing 9 geographic
seed sources (fig. 1) were planted at two locations in southern Louisiana in
early 1983. The 1-0 seedlings had been grown at the Louisiana State
University, School of Forestry, Wildlife, and Fisheries nursery at Baton
Rouge, LA. Field design was a ten-replicate, randomized block design with
five-tree-row family plots. Seedlings were planted on a 3-m x 3-m spacing,
and a single row of border trees was planted around each plantation. All
seedlings were root-pruned to eight inches to facilitate planting and graded
by height and diameter. The largest seedlings of each family were planted in
block 1, and successively smaller seedlings were planted in later blocks.
The first out-planting is located on the Thistlethwaite Wildlife
Management Area in St. Landry Parish, Louisiana. This site had been under
cultivation for approximately 25 years prior to which it was a bottomland
hardwood site. The soils at the Thistlethwaite plantation consist of approxi-
mately 80 percent Baldwin silty clay loam and 20 percent Dundee silty clay
loam (USDA Soil Conservation Service 1976). The second plantation is on a
Lf Graduate Research Assistant, Research Associate, and Associate Professor,
respectively, Louisiana State University, School of Forestry, Wildlife, and
Fisheries, Louisiana Agriculture Experiment Station, LSU Agriculture
Center,Baton Rouge, La. 70803
267
*%- Bursery Locstion
Source locations
1] - Natchez, MS
2 - Alexandria, LA
3 - Bogalusa, LA
‘oD 4 - Gibson, LA
De LayPlaces nbs
fe ooh
7
8
y)
- Stoneville, MS
- Illinois/Kentucky
- Monticello, AR
- Bunkie, LA
es°
Figure 1. Relative location of selected geographic seed sources of
baldcypress (Adapted from Faulkner and Toliver 1983).
bottomland hardwood/swamp site owned by the St. Martin Land Company in
St. Martin Parish, Louisiana. This site was cleared of timber and diked for
management as a crawfish (Procambarus clarkii) pond. The soil here is a
Sharkey clay (Murphy et al. 1977). The St. Martin plantation was flooded with
water to a depth of 20 cm for 2-3 weeks in May 1983 to stock the site with
crawfish and then was inundated again from October 1983 through May 1984 for
crawfish production. Both plantation sites were disked prior to planting, and
weed competition after planting was controlled by a combination of disking,
mowing, and herbicide applications. Soil moisture and climatic factors were
monitored and recorded on a bi-weekly basis at each plantation from April
through September of 1983.
An analysis of variance on the height and diameter of the seedlings taken
at the time of planting indicated several significant differences attributable
to the seedling grading procedure. Therefore, in order to remove the effect
of this initial variation, geographic source and family components of the 3-
year-old seedlings were examined by analysis of covariance using the General
Linear Model (GLM) procedure of the Statistical Analysis System (SAS) (SAS
Institute 1982). Further adjustment of the statistical model was necessary
because of animal damage to seedlings that occurred at both plantations. In
the early summer of 1983, white-tailed deer (Odocoileus virginianus) browsed
47.2 percent of the seedlings at the Thistlethwaite plantation. In the spring
of 1984, crawfish either partially or completely girdled 77.8 percent of the
268
seedlings at the St. Martin site. In both cases the damage rate was nega-
tively correlated to the height and diameter of 1-0 seedlings. In order to
remove the effect of seedling damage on the genetic and site components of the
statistical model, a damage factor was added for analysis of covariance of the
3-year-old seedling data.
RESULTS AND DISCUSSION
Site Variation
After two growing seasons in the field, the combined survival rate of the
two plantations was 96.4 percent. The Thistlethwaite seedlings had a survival
rate of 98.3 percent, while survival at St. Martin was 94.5 percent. Combined
mean height was 126.3 cm and mean diameter was 2.05 cm. Analysis of covari-
ance indicated a highly significant difference in height and diameter of
seedlings between the two plantations. Mean height and diameter at
Thistlethwaite were 125.8 cm and 1./2 cm, respectively, as compared with a
mean height of 127.2 cm and a mean diameter of 2.40 cm for the seedlings at
the St. Martin plantation. Mean growth at Thistlethwaite was 15.2 cm in
height and 0.60 cm in diameter as compared to a growth of 28.1 cm in height
and 1.48 cm in diameter at the St. Martin site. The superior growth rate of
the St. Martin seedlings is attributed to the more favorable soil moisture
conditions at this site brought about by the periodic inundation of the
plantation area for crawfish production. Since baldcypress is naturally
adapted to bottomlands subjected to seasonal flooding, the spring and early
summer flooding of the St. Martin site did not hamper seedling growth.
Instead it appeared to enhance growth by delaying the development of weeds and
summer moisture stress conditions, therefore promoting early height and
diameter growth. Precipitation and temperature were not significantly
different between the two sites.
Genetic Variation
Geographic variation was not significant for either height or diameter of
the 3-year-old seedlings. Faulkner and Toliver (1983) found a similar lack of
geographic variation among 1-0 baldcypress seedlings. It is possible that the
scope of both studies was not large enough to detect geographic variation.
Only localities along the Mississippi River floodplain were sampled, while
testing of provenances from a wider range might have provided more geographic
diversity. This could also be a result of seed dispersal down the Mississippi
River during flood conditions. Flood waters could easily have carried seed
from northern sources southward, resulting in less genetic diversity among
‘provenances along the floodplain.
Family-within-source variation was significant for height (p<.05) and
highly significant for diameter (p<.01) of the 3-year-old seedlings (table 1).
Thus it appears that there is greater genetic diversity among individual trees
within natural baldcypress stands than among provenances. The wide range of
variability among families points to a potential for genetic gain in the
growth of baldcypress through family selection. If the best three families
(top 10 percent) were selected for mean height (table 2), one each would come
from the Stoneville, MS; Gibson, LA; and La Place, LA sources resulting in a
realized gain of 13.3 percent (16.8 cm). A gain of 17.6 percent (0.36 cm) in
diameter can be obtained by selecting the best three families, two families
from the Stoneville source and one from the Gibson source (table 2). These
269
Table 1.--Analysis of covariance of baldcypress seedling heights and diameters
after two growing seasons on two different sites in Louisiana.
Source of variation Degrees of Mean squares
freedom Heights Diameters
Damage = D 3} 81739.89 41.07
Plantation = P 1. 86166.832/ 128.942/
Block-within-plantation = B(P) 18 2155.908/ 1.332/ :
Dix BG) 227) 566.35 0.46
Source = S 8 1086.49 0.42
Dysas 24 693.04 0.28
PrechShayy 8 783.17 0.642/
G2) ss j 144 485.32 0.26
Family-within-source = F(S) 17 906.572! 0.522/
Pe xeukiCS) 7 285.94 0.16
Error 597 407.36 . 0.22
af Significantly different at the .01.level of probability.
— Significantly different at the .05 level of probability.
gains could be extremely important to the successful establishment of
baldcypress plantations. Planting of larger seedlings could overcome the
problems of periodic high water levels, weed competition, and animal damage,
and thus result in higher survival rates. It should be remembered, however,
that rapid early growth in either height or diameter of a particular family
does not necessarily indicate that gains will continue through an entire
rotatione Further testing is essential to determine if the magnitude of these
family gains and rankings will remain consistent.
Genotype x Environment Interaction
Plantation-by-source interaction was significant (p<.05) for seedling
diameter (table 1). This genotype x environment interaction indicates that
some geographic sources of baldcypress may be site specific. Certain sources
performed very well on one site and poorly at the other in comparison to the
other sources. In particular, the Stoneville source ranked first at the |
St. Martin site with a mean diameter of 2.91 cm, but dropped to seventh place |
at the Thistlethwaite site with a mean diameter of 1.67 cme The Natchez, MS
source was second at the Thistlethwaite plantation (mean diameter = 1.79 cm)
and ranked ninth at St. Martin (mean diameter = 2.07 cm). If this interaction
270
continues to exist after further testing, then future plantings should be made
by matching provenances to the proper site to obtain maximum tree growth.
Table 2.--Ranking of families by height and diameter across both plantations.
Seed Source Family Code Mean height Mean diameter
SSS SSS (Si)
a/
Stoneville, MS 6-4 WAT LST IS) Ded
Gibson, LA 4-2 140.84 2.32
La Place, LA 5=—2 140.68 2.26
Stoneville, MS 6-2 140.47 2239
Alexandria, LA 2—3 139.88 2.26
Gibson, LA 4—] 135.46 Zio l'9,
Stoneville, MS 6-3 L214 2-30
Alexandria, LA Die 131.66 PPA
La Place, LA 3)! 13 Foro 2.09
Stoneville, MS (ofall 131.06 Pls)
La-Ptace, LA 5=5 127.31 1.86
Bogalusa, LA a=) 125.94 2.02
Bunkie, LA 9=2 125.14 2.08
Bunkie, LA 93 125.07 1.89
Bunkie, LA 9-4 124.58 ees)
Monticello, AR 8-2 124.03 2.19
Monticello, AR 8-6 123.09 Dreil2
La Place, LA 53) WAZA S74 1.93
La Place, LA 5-4 121.68 1.84
Bunkie, LA 95 116.26 1.80
Natchez, MS 2, 116.19 1.79
Tllinois/Kentucky 7-2 115.86 1.78
Illinois/Kentucky 7-3 INE GSi7/ 1.82
Monticello, AR sill LENS YASYS) ZeOil
Tllinois/Kentucky 7-1 107.45 1.66
Tllinois/Kentucky 7-4 105.58 1.86
a/ The first number refers to geographic seed source (see fig. 1), and the
second number refers to family-within-source.
CONCLUSIONS
Site conditions are important to height and diameter growth of
baldcypress seedlings. Adequate available soil moisture is of particular
importance in this respect and should be considered in site selection.
Significant family variation for both height and diameter and potential early
growth gains through family selection warrant the further testing of
baldcypress families. Also, the existence of a genotype x environment
interaction indicates the need for progeny testing of baldcypress over a wider
range of sites to increase the potential for gains through consideration of
Site characteristics. Finally, the planting of larger seedlings (taller than
1.0 m in height and larger than 1.25 cm in diameter) should reduce the
incidence of crawfish damage and deer browse and improve early survival and
growth.
271
LITERATURE CITED
Bull, H. 1949. Cypress planting in southern Louisiana. South. Lumberman
179( 2249) :227-230.
Faulkner, S., and J. Toliver. 1983. Genetic variation of cones, seeds and
nursery-grown seedlings of baldcypress provenances. Proc. 17th South. For.
Tree Improv. Conf. 39:281-288.
Boil; Reo Re, andy ReniGemMertsisinctellids 1966. Planted forests of Louisiana.
North La. Hill Farm Exp. Stn. Bull. No. 6115 47 p-
Murphy, Ke E., Be A. Touchet, A. G. White, J. J. Daigle, and H. L. Clark.
1977. Soil survey of St. Martin Parish, Louisiana. USDA Soil Conserv.
Serv. and LA Agric. Exp. Stn. 73 p-, 74 maps.
SAS Institute, Ince. 19820" SAS User's Guide: “Statistics L982 editvone Sas
Institute, Inc, Cary, NC. 584 p.
USDA Soil Conservation Service. 1976. Soil Survey of St. Landry Parish,
Louisiana. Soil Survey Field Sheet No. 21.
Williston, H. L., F. W. Shropshire, and W. E. Balmer. 1981. Cypress is
promising species for management in southern wetlands. For. Farmer
40(10):6-8,17-18.
272
BIOMASS CHARACTERISTICS OF SYCAMORE COPPICE
INFLUENCED BY
PARENTAGE AND TYPE OF PIANTING STOCK
Seb Lands Jie and Ee Bi. Schultz!
Abstract.--Three years after clearcutting a six-year-old
sycamore progeny test in northeast Mississippi, stem dry weight of
coppice averaged 2.27 Mg/ha and represented 63% of the above-stump
dry biomass. Stumps of trees established from unrooted cuttings
produced fewer coppice sprouts, smaller sprouts, and 40% less
coppice stem dry weight than stumps. of trees established from
seedlings. There were no differences between stumps originating
from top-pruned and unpruned seedlings. Progeny families differed
in survival, dry weight yield of five-year-old trees before the
clearcut, and number and maximum diameter of coppice sprouts per
stump. Small, positive correlations between stump diameter before
clearcutting and the resulting coppice characteristics were found,
and these relationships may differ among families.
Additional keywords: Coppice growth, genetic differences,
Pilatanus occidentalis.
Growing American sycamore (Platanus occidentalis L.) in plantations under
short coppice rotations for fiber has received much publicity as the "silage
sycamore concept" (Georgia Forest Research Council 1973). The concept was
originally advocated for pulp and paper production, but now has applicability
to energy plantations. Age of tree, spacing, and season of cutting can
influence coppicing ability and coppice yields (Kennedy 1975 and 1980;
Kormanik et al. 197.3) re Objectives of the present study are to elucidate
effects of parentage and type of planting stock on coppicing ability in a
young sycamore progeny test and to compare coppice yields three years after
clearcutting with stand yields prior to the clearcut.
MATERIALS AND METHODS
A nine-year-old open-pollinated sycamore progeny test in Oktibbeha
County, Mississippi (33°18' North, 88°55' West) that was clearcut at age six
to produce coppice was used for the study. Three types of planting stock were
utilized to establish each of ten families in the test in June, 1974:
(i) unrooted top cuttings from 1-0 seedlings, (ii) top-pruned 1-0 barerooted
seedlings, and (iii) whole (unpruned) 1-0 barerooted seedlings. The first two
types were obtained by clipping a seedling at 2.5 cm above the root collar to
provide a complete top cutting and a detopped (top—pruned) root system. The
ten families came from eight geographic seed sources in the Gulf South (Figure
1), with two families per source being used from sources "A" and "F" and one
family per source from the other six. The test was arranged as a split-plot
1
a; Professor, Department of Forestry, and Research Associate, School of Forest
Resources, Mississippi Agricultural and Forestry Experiment Station (MAFES),
Mississippi State, MS 39762. MAFES contribution No. 6134.
273
Mississippi
Louisiana
e°
Figure 1 .--Geographic locations of seed sources and test site for a sycamore
progeny test used to study effects of parentage and type of planting stock on
biomass characteristics of three-year-old coppice.
\
design with families as whole units and with four replications. Spacing was
3 x 3 meters, and each rep-by-family-by-stock-type plot contained five trees.
Stump diameters at 15 cm above ground, stem diameters at breast height
(DBH), and tree heights were measured at plantation ages three and five.
During the sixth growing season the test was clearcut, and at stump age nine
the three-year-old coppice was measured for number of sprouts per stump,
sprout diameter at 7.5 cm above stump, and sprout height above stump. Results
of the fifth-year measurements before the clearcut have been reported by Land
et al. (1981), but are expressed in metric terms in this paper for comparison
with the coppice results.
A sample of 257 coppice sprouts from 36 stumps was measured for diameter
and height during the fourth growing season of the coppice, and these sprouts
were then destructively sampled for green and dry weights of stems, limbs, and
leaves separately. Ratios of dry weight to green weight were used to obtain
total dry weight of each component of each sprout, and the 257 records were
utilized in developing dry-weight prediction equations for individual sprouts
from regression on sprout diameter and height.
The above equations were applied to the three-year-old-coppice data to
predict the dry weight of each sprout. Totals for all sprouts on a stump
(kg/stump) were obtained and used as the basic records in analyses of variance
for effects of families and types of planting stock. Survival times number of
274
trees planted per hectare times dry weight per stump provided plot values for
coppice yields in kg of dry weight per hectare.
RESULTS AND DISCUSSION
The 257 destructively-sampled coppice sprouts averaged 3.1 cm for
diameter at 7.5 cm above stump and 3.5 m for height above stump.
Minimum-maximum values were 1.3 cm to 8.9 cm and 1.2 m to 7.6 m, respectively.
Dry weight/green weight ratios were 0.374 for leaves, 0.455 for limbs, and
0.451 for stems. The resulting means and ranges in dry weight per sprout for
each component were 0.222 kg for leaves (0.004 to 1.874 kg), 0.229 kg for
limbs (0.000 to 3.030 kg), and 0.833 kg for the stem (0.037 to 6.325 kg). The
prediction equations derived from the samples (Table 1) should only be used on
coppice within these size ranges.
Diameter squared times height (DH) was a better predictor of sprout dry
weight than diameter squared (D ), as indicated by smaller standard errors of
estimate (S ) and larger coefficients of determination (R ) for DH than for
D in Tablé°T. The no-intercept models with D H were chosen for predicting
dry weights in the three-year-old coppice, since these modeis did not greatly
increase the S$ values over those of the corresponding intercept models and
since the no-intercept estimates for stem, limb, and leaf components can be
added together to equal predicted total values.
The additive feature of the no-intercept predictions allows’ the
calculation of predicted weights for other components if the predicted weight
for one component is given. This feature also means that identical
analysis-of-variance results and rankings of treatment means will be obtained
for the predicted weights of the different components. Therefore, only the
stem dry weights per tree (or per stump for coppice) and per hectare will be
presented, and the other components’ green and dry weights can be calculated
from the conversion equations in Table 2 if desired.
Stem dry weight yield of the three-year-old coppice on nine-year-old
stumps averaged 2.27 Mg/ha (1.0 tons/ac) and represented 63 percent of the
above-stump dry biomass (Tables 2 and 3). These yield and percentage values
are lower than those reported for other sycamore coppice studies (Kennedy 1975
and 1980; Kormanik et al. 1973), because of the wider spacing used here than
in those studies. However, the three years of coppice growth produced
approximately as much biomass as the first four years of tree growth in the
progeny test, as indicated by a stem dry weight yield of 0.66 Mg/ha for the
three-year-old progeny test and 5.75 Mg/ha for the five-year-old test. The
average stump at nine years of age contained 3.9 coppice sprouts that were
greater than 1.8-cm diameter at 7.5 cm above the top of the stump, and the
largest sprout per stump averaged 4.42-cm diameter and 4.35 m height. No
mortality of stumps occurred following clearcutting, so that the 80.5-percent
survival of the nine-year-old stumps was the same as the survival of the
three-year and five-year-old trees before clearcutting.
This investigation of effects of parentage and type of planting stock on
coppice dry weight yields used the initial working hypothesis that stump
survival, number of sprouts per stump, and size of the largest sprout per
stump were the primary determinants of yield. It was also hypothesized that
diameter of the stump when the trees were clearcut would influence these
275
Table 1. Equations for prediction of dry weights of stems, limbs, and
leaves for five-year-old trees and three-year-old coppice in a
sycamore progeny test in northeast Mississippi
a
Dependent Variable (= Y) Prediction Pquat tone. S R2
eee VT —-- rrr
Five-Yr.-Old Trees
Oven Dry Wt. (kg/tree)
Stem Y = 0.1579(DBH) 2 0.9480
Limbs Y = 0.0808 (DBH) 2 1.0206
Leaves Y = 0.0459 (DBH) 2 0.8800 %
Total Y = 0.2846(DBH) 2 2.1410
Three-Yr. Coppice on 9-Yr. Stump
Oven Dry Wt. (kg/sprout)
Sprout Stem Y = -0.1345 + 0.00736(D2) 0.2680 0.957
Y = 0.0810 + 0.0108 (DH) 0.1939 0.977
Y = 0.0698 (D2) 0.2880
Y = 0.0111 (D2H) 0.2058
Sprout Limbs Y = -0.1094 + 0.0258 (D2) 0.1908 0.843
Y = -0.0343 + 0.0038 (DH) 0.1778 0.8@e
Y = 0.0227 (D2) 0.2093
Y = 0.0037 (D2H) 0.1799
Sprout Leaves Y = -0.0348 + 0.0195 (D2) 0.0977 0.92%
Y = 0.0239 + 0.0028 (D2H) 0.0945 0.927
Y = 0.0185(D*) 0.1014
\ Y = 0.0029(D2H) 0.0965
Total Sprout Y = -0.2787 + 0.1189(D2) 0.4920 0.945
Y = 0.0705 + 0.0174 (D2H) 0.4012 0.963
Y = 0.1111 (D2) 0.5387
Y = 0.0177 (D2H) 0.4050
a/
— DBH = diameter at breast height in cm; D = diameter of sprout in ecm at 7.5
cm above stump; H = height of sprout in m above stump.
determinants of yield, and this diameter would be subject to effects of
parentage and type of planting stock.
Trees originating from unrooted cuttings produced significantly fewer
coppice sprouts per stump, smaller diameters and heights for the largest
sprout per stump, and less coppice stem dry weight per stump than did trees
coming from seedlings, but there were no differences between pruned and
unpruned seedlings (Table 3). The trees from cuttings also averaged smaller
stump diameters at age five prior to the clearcut. Since stump diameter of
the young trees may be an index of the size of the root system and thus of the
absorption and storage capacity of the plant, analysis of covariance was used
to adjust coppice means to equivalent stump diameters for each of the types of
planting stock. The smaller size of the largest sprout per stump for cuttings
than for seedlings was related to the effect of smaller stumps from cuttings,
since covariance adjustment removed the significant differences in sprout
276
Table 2. Distribution of green and dry weights among stems, limbs, and leaves
of young sycamore trees and coppice sprouts, and conversion equations
for calculating these components from stem dry weight
Percent
Trait Distribution Conversion Equations
in Tree (Wts. in kg/tree, or Mg/ha)
or Sprout
Plantation Trees (3 & 5 Years Old)
Stem Dry Wt. (= StDW) 56 StDW = given in Tables 3&5
Limb Dry Wt. (= ImDW) 28 ImDW = 0.5117 (StDW)
Leaf Dry Wt. (= LvyDW) 16 IvDW = 0.2907 (StDW)
Total Dry Wt. (= TotDW) 100 TotDW = 1.8024(StDW)
Coppice Sprouts (3 Yrs. on 9-Yr. Stump)
Stem Dry Wt. (= CStDW) 63 CStDW = given in Tables 365
Limb Dry Wt. (= ClmDW) 2h ClmDW = 0.3295 (CStDW)
Leaf Dry Wt. (= ClvDW) 16 CLvDW = 0.2639(CStDW)
Total Dry Wt. (= CTotDW) 100 CTotDW = 1.5934(CStDW)
Stem Green Wt. (= CStGW) 61 CStGW = 2.2188(CStDW)
Limb Green Wt. (= ClmGWW) 20 ClmGW = 0.7214(CStDW)
leaf Green Wt. (= ClvGW) 19 CLvGW = 0.7034 (CStDW)
Total Green Wt. (= CTotGW) 100 CTotGW = 3.6436(CStDW)
sizes. Additional factors--such as stump vigor, root system arrangement, or
response to clearcutting--that are not highly related to stump diameter may be
more influential in determining number of sprouts per stump. Covariance
analysis did not remove the significant difference between cuttings and
seedlings for number of sprouts per stump.
Survival was lower for trees established from cuttings than for trees
coming from seedlings (Table 3). This effect of planting stock occurred prior
to plantation age three, and there was no stump mortality after clearcutting.
Since coppice dry weight per stump was also lower for cuttings than seedlings,
stumps from cuttings produced 40 percent less coppice stem dry weight per
hectare (1.2 Mg/ha) than did stumps from seedlings (2.85 Mg/ha). In the
five-year-old trees prior to clearcutting there was also a significant
difference in stem dry weight yields between pruned and unpruned seedlings,
but the difference was not present for three-year-old trees or three-year-old
coppice.
The primary effect of parentage (family) on coppice characteristics was
in number of sprouts per stump, although the maximum and minimum families for
diameter of largest sprout per stump were also significantly different (Table
4). Stump diameter at age five differed significantly among some families,
and when used as a covariate it removed the one significant family difference
in diameter of largest sprout. Adjustment for stump diameter did not remove
family differences in numbers of sprouts per stump, however. There were no
significant family differences in coppice stem dry weight per stump or per
277
Table 3. Means and tests of significance for effects of type of planting
stock on traits of 3-year-old trees, 5-year-old trees, and
3-year-old coppice in a 9-year-old sycamore progeny test
Trait Overall Type of Planting Stock! F-tese |
Study Cut tings Seedlings Prob.
Mean Top Whole
Pruned
Stump Diam. (cm) at Age 5 8.51 V3 8.7 8.9 -O001%**
No. Sprout s/St mp:
Unad jus ted— 349) 343 4.0 4.2 -0001**
Adjusted 359) 3.4 39 4.2 -008**
Diam. (cm) of Max. Sprout/Stump:
Unad jus ted 4.42 4.1 4.5 4.5 -001**
Adjusted 4.42 4.2 4.5 4.4 6 PKS}
Ht. (m) of Max. Sprout /Stump:
Unadjusted 4,35 4.1 4.4 4.4 -001**
Adjusted 4.35 4.2 4.3 433 159
Survival (%) at Stump Age 9 80.5 56 90 96 -0001**
Coppice Stem Dry Weight:
kg/Stump Zeo2 2.0 Doll 2.8 -0001**
Me/ha Deal leee2 28 2.9 -0001**
Tree Stem Dry Weight:
Age 3 (Mg/ha) 0.66 0.3 0.9 0.9 -0001**
Age 5 (Mg/ha) (4 ew 5/5 350) 6.6 Wei) -O001**
Teens underlined by same line are not significantly different at the .05
probability level.
BiTetesk probability level for "Types of Pianting Stock" in analysis of
variance.
ey
— Means unadjusted and adjusted for differences in age 5 stump diameter.
hectare, even though family differences in survival and dry weight yield of
the five-year-old trees were detected prior to clearcutting (Table 5).
Relationships were examined between stump diameter and dry weight yields,
number of sprouts per stump, and diameter of largest sprout per stump to
elucidate the nature (or lack) of family effects on coppice production (Figure
2). Coppice stem dry weight, number of sprouts, and diameter of the largest
sprout per stump all increased as stymp diameter increased, but the percent of
variation explained by regression (R x 100%) was low. Other factors, such as
families, types of planting stock, and many non-measured variables (stump
height, date cut, etc.), all contributed to the unexplained variation. The
effects of families are illustrated by the plotted family means. Although
family differences were not significant for stem dry weight at coppice age
three, several interesting trends can be observed by comparing graphs. Family
"A2' is consistently a poor performer from tree age five through coppice age
278
*za}0uRTp dunqjs ¢ o3e UT SeoUeteyFTp 10y poysnf{pe pue peysnf[peun sueoy —
jo
*soueTIeA jo sTsk{eue ut ,SseT{TWeg, 10JF {eae}, AIL{[TGeqoid ea
*yaae, A1{1TqGeqoad co* ey ye JodtezzTp Aj }UPOTFJTUSTS jou sie oUT{T owes Aq poUyT{s1epun See
EE a ep ne es ee
6 € 0°# Bey) GY, Cay, ee, c*? Fey) 9° Ea) (°fpe) () °3H
794° IN 1d 1d ad a 19 IL TO Tl IV Ay yueg
45 0°? O°7 ay Gay, Pd Say) LY, Ly, 8°? (*fpeun) (WW) °34
I8€* IN cV 1d Ta cd 10 1D IL iat IV Ay yweg
:dungg/jnoids *xew °*34
0° ey bay) [hy les) 7°H yey, yo 8°47 Gis? (°fpe) (Wo) swetd
70S° cv IL Id IN ca IV 14 LO=—s10 Tl AY yuleg
B°e ONY. (hey) Gey) Co7 Va) Gey, G*¥ Ly, 0°S (°fpeun) (wo) swetg
€6%7° cv 1d IN (Gal TL IV Td 19 10 la Aj yueg
:dunjsg/inoidg *xew *wetg
eS UPAS VES LAS Bre B°E 6°€ Oi, GAY LY (°fpe) tequny
**%700° IN ZV 1d IV 10 TI [Dee cal 1d LL Aj pued
Gace 7° G’e Ee Bre Ome 0°# OF” GY) 6° ose neuD) Toqunn
¥*¥*€00° cv IN 14 10 IV Tl 19 (Gul 1d IL Aj ywed
:dunjzg/sjnoidg °on
Grle WEL 0°8 0°8 S°8 9°8 8°8 6°8 6°8 0°6 (wo) *wetd
G71 CV 10 IN 1d IV (al TI IL ld 19 = Appa
2(¢ o8y) *wetg duinis
Nae ee EEE TEETER EERE
poe ole See a eee Ci) at ania NERY bad eels
#se4-4 ee eee ee 1yeLL
2 UL SO1}STdo}oOeAeYD Bo1ddoo plo-deek-¢ uo AjTWeF JO SJOaTJa 1OJ BoURSTFLUSTS JO s}]Se} pue suPaW “Hh 91 qPL
279
*QouUeTIPA JO SpSsA[eue Uy ,SeTipWweq, 103 jedey, Ai}z{Tqeqoad yseq-q —
/4
*TaAeyT AT{tqeqoad ¢o* eu} 3e WotesjTp AL WeITJTUsTS JOU ee oUt], sues Aq peuT{sepun EON
Gm, 9° L°4 0°S 0°S o°9 G°9 9°9 Gal G°Z ey/3y
790° ZV lia, 1d IN 10 IV TI ld 19 al AY rue gq
2(¢ e8y) °3mM Adg Weis sel]
G°0 G°0 9°0 9°0 L°0 {Lz (0) 8°0 8°0 8°O 8°0 ey /3y
899° CoV 1d TO IL Td TI ite) CA IN IV AT qyueg
2(¢ o8y) *3mM Aaq Wais aoiy
Tal Gale sea Oo ad See ORC = Ca: Oe ee ey/3y
OLY’ (al IN Td Td 1O (al LL IV Tl 1) Ay wey
eT Sra eG [Lae Ld ype GH (G2 WPS, Sas duinz s/3y
71S° cv IN ld 1d 7Cle LO IV ite) LL Tl AjqTweg
2°3m Aaqg Weg votddog
OL OL GL Bl SL SL 08 €8 €6 86 4s
ES: Te 1d IN ZV IV 10 Tl ld 6) al Ay tueg
2(6 e8y duinjs) [eAtTAAns
—"qolg one; 6 Si Lif Qi Fi; VE Ci Che a
/4saq-4 jerrrs Ay weq FEB O EF
So] Au aso id
aioueohS e ut sdunqjs pjo-1eak=-6 uo sotddoo pio-iees—-¢ pue s9ai1} ploO-1eek-¢ pue ieek—-¢
jo sjzy3tem AIp Woes pue TeATAANS uo Aj{TWeJ JO S}JOeTjJa AOJ BOUPOTITUSIS Jo SjZSej pue suPOW
“¢ SL9PL
280
CDW= -0,68+0.39D aie SNUM=1.91 +.0,23D
R°=0,12** 7 By! R°=0,12**
TI
a
S)
a
mp
on
(e)
Coppice (kg /stump)
nm
aay
No. sprouts /stu
# A
(S) oO
|
|
|
|
|
|
|
|
|
|
20
= 35
3 6
= L2 3.0
i 70 0S 8.0 8.5 Oy F140) vs 8.0 8.5 9.0
5 aul
S 72- TOW = -4.69 + 1.33D Al E5.0r SDIA= 2.78 +0.19D Bi
WW Ds *% Ba = = x
= R“=0.80 a R°=0.07
Oo a (BLS $48
i<D) oO
| Bivge 5 a
| SS
| om
(= 60
| 8
Bere
! 5.2
48
7.0 75 8.0 8.5 90 7.0 75 8.0 8.5 3.0
D= STUMP DIAMETER (cm) AT TREE AGE 5
Figure 2.--Regressions and plotted family means for relationships between
stump diameter at age five and (i) stem dry weight of tree at age five and
(ii) characteristics of three-year-old coppice on nine-year-old stumps.
Enneeweurnamiives: Uh Ohl’ tlt. and k2) illustrate contrasts, in| coppice
dry weight production. Alli had large stump diameters at age five, and all but
'Tl' had high stem dry weights at that age. Four years later the stem dry
weight yields of the three-year-old coppice for 'Fl’ and 'F2' were well below
their predicted values based on stump diameter, whereas 'Ll" and 'Tl' were
well above the predicted values. The reasons differ between the two sets.
"Ll' is high because of large diameter sprouts and average number of sprouts
per stump, whereas 'Tl' is high because of large numbers of sprouts and
average sprout diameter. 'Fl' is low because of low numbers of sprouts
and average sprout diameter, whereas 'F2' is low because of small diameter
sprouts and average number per stump. If these trends continue for older
coppice, where family differences in coppice dry weight may be significant,
the desirable families to breed for coppice production would be typified by
"LI1': high biomass on a few, large sprouts per stump.
SUMMARY AND CONCLUSIONS
Dry and green weights of sycamore coppice sprouts ranging in size from
one to nine cm in basal diameter and one to eight meters in height were
reliably predicted from no-intercept linear regression equations presented in
this paper with basal diameter squared times height as the independent
variable. Using these equations, stem dry weight yield of three-year-old
coppice on nine-year-old stumps in a sycamore progeny test planted at 3 x 3m
281
spacing in northeast Mississippi averaged 2.27 Mg/ha. This yield fell between
the yields for the three-year-old and five-year-old trees in that test prior
to clearcutting. Sixty-three percent of the above-stump dry coppice biomass
was located in the stems. There was no mortality of stumps during the three
years following clearcutting, probably because of the wide spacing used.
Trees established from unrooted cuttings of seedlings’ tops had poorer
early survival and grew slower than trees established from pruned or unpruned
bareroot seedlings. When the trees were clearcut, those from cuttings had
smaller stump diameters and subsequently produced fewer and smaller coppice
sprouts than those from seedlings. The end result was 40 percent less coppice
stem dry weight produced on stumps of trees from cuttings than on stumps of
trees from seedlings. Top pruning of bare-root seedlings before planting had
no effect on coppice characteristics when compared with unpruned seedlings.
Families significantly affected survival and growth of trees during the
first five years after planting. When the stand was cut at age six and
allowed to coppice for three years, family differences were detected for
number of sprouts per stump, but not for coppice stem dry weight per stump or
per hectare. Smail, positive relationships between stump diameter and coppice
characteristics were observed, however, and family means varied around this
relationship. Selection for families with the fastest growth rates during the
first five years after planting, and thus the largest stump diameters when
clearcut to produce coppice, may not always be the families that produce the
most coppice biomass or the most biomass in the most acceptable form (few
stems).
’ LITERATURE CITED
Georgia Forest Research Council. 197/355 Forest research introduces new
concept to meet newsprint shortage. GA Forestry Comm., 4 p. Macon, GA.
Kennedy, H. E. Jr. 1975. Influence of cutting cycle and spacing on coppice
sycamore yield. USDA Forest Serv. Res. Note S0-193, 3 p. South. Forest
Exp. Stn., New Orleans, LA.
Kennedy, H. E. Jr. 1980. Coppice sycamore yields through 9 years. USDA
Forest Serv. Res. Note S0-254, 4 p. South. Forest Exp. Stn., New
Orleans, IA.
Kormanik,..P.,.P.-,. Tyne, G. “Li; and Belanger QoR. Po?" 1973." “A -casew hi stomy
of two short-rotation coppice plantations of sycamore on _ southern
Piedmont bottomlands, p. 351-360. In IUFRO biomass studies. S4.01
Mensuration, Growth & Yield, Working Party on Mensuration of Forest
Biomass. Univ. Maine, Coll. Life Sci. & Agric., Orono, ME.
Land, S. B. Jr., :Dicke, 9S. jG.) and Tuskan, G. A. 1981. Parentage and type
of planting stock influence biomass characteristics in sycamore
plantations, p. 223-232. Im Proc. 2/th N. E. Forest Tree Impr. Conf., D.
H. DeHayes (ed.). USDA Forest Serv., N. E. Forest Exp. Stn., Durham, NH.
282
Predicted Genetic Gains Adjusted for Inbreeding
for an Eucalyptus grandis Seed Orchard
K. V. Reddy, D. L. Rockwood,,G. W. Comer
and G. F. Meskimen—
Abstract.--One-half of a genetic base population (GPOP77) of
Eucalyptus grandis Hill ex Maiden containing 529 families
representing four generations of selection was harvested in
August 1978. Family and individual tree heritabilities for
64-month coppice height, diameter at breast height (DBH) and
volume were 0.65, 0.65 and 0.59, and 0.31, 0.32 and 0.27,.
respectively. Predicted gains were adjusted for inbreeding
due to mating of related individuals as well as for natural
selfing. The predicted genetic gains through different
selection strategies ranged from 41% to 90%. Impressive gains
were also predicted for clonal propagation of selected indivi-
duals.
Additional keywords: heritability, coppice, inbreeding,
selfing.
In southern Florida Eucalyptus grandis has been the focus of
intensive genetic improvement (Franklin 1978). It is recommended on the
palmetto prairie (low fertility) and acid flatwoods (Geary et al. 1983).
Trees planted in southwest Florida consitute a landrace developed
through three generations of selection and progeny testing. Each
generation of selection enhances the landrace's adaptation to local
conditions (Meskimen 1983).
In south Florida, substantial genetic gain for juvenile height and
‘volume growth at 2.5 years of age was found in 33 E. grandis progenies
tested (Rockwood and Meskimen 1981).
Won sdaate Research Assistant, Associate Professor, Assistant Research
Scientist and Adjunct Assistant Professor, respectively, Department of
Forestry, University of Florida, Gainesville, FL. Research reported
here is supported by Oak Ridge National Laboratory under subcontract no.
19X-09050C and a cooperative program between the Institute of Food and
Agricultural Sciences of the University of Florida and the Gas Research
Institute, entitled "Methane from Biomass and Waste". The USDA For.
Serv., Southeastern Forest Experimental Station, installed and main-
tained the experimental materials through September, 1984.
283
In Florida the current E. grandis base population consists of 529
families (Geary et al. 1983). These families were all derived from 228
original introductions, mainly from Australia and South Africa, that
went through up to four generations of selection. Consequently, some of
the 529 families in the current base population were related due to
common ancestry leading to a certain degree of inbreeding.
In a tree improvement program, care should be taken to avoid the
deleterious effects of inbreeding. These effects may be of greater
importance in eucalypts than has previously been considered. The most
significant consequence of inbreeding depression is the reduction of the
mean phenotypic value manifested by characters connected with reproduc-
tive capacity, physiological efficiency and general vigor of the off-
spring. Selfing, the most severe form of inbreeding has been shown to
adversely affect most characteristics in E. grandis (Van Wyk 1981).
Estimation of inbreeding consequences is of practical importance in
continuing development of E. grandis seed orchards in southern Florida.
This paper reports the effect of inbreeding due to mating of relatives
as well as to natural selfing on predicted gains from a 50-family seed
orchard.
MATERIALS AND METHODS
The E. grandis genetic base population (GPOP77), located at
LaBelle, Florida, was planted on July 1977 and includes 529 open-
pollinated families representing four generations of selection (Table
1). Each family was represented 60 times in a completely randomized
single-tree plot design on 17.3 hectares. Spacing was 1.8 m between
trees on paired beds spaced 2.3 m within pairs and 3.5 m between pairs,
for a density of 1,916 trees per hectare. The southern half of the
plantation was harvested in August, 1978. Coppice height and DBH at,64
months were used to calculate the volume according to the formula D H,
where H and D are height and DBH, respectively. Genetic gains were
predicted for different selection strategies for both individual and
combined selection, using selection intensities obtained from tables
provided by Namkoong and Snyder (1969).
The effect of inbreeding on predicted gains was studied for a
50-family seed orchard with one tree per family. The study assumed that
the original introductions themselves were not inbred. The inbreeding
coefficients for the progeny of all possible matings among the 50
families were calculated by tracing the pedigree of each mating to the
common ancestor and computing the probabilities at each level (Li 1976)
as shown in Figure l.
Table 1. Proportion of families per generation in GPOP/77.
Generation % Families
1 P27
2 40
3 24
4 9
284
RESULTS AND DISCUSSION
High genetic variation was observed for height, DBH and volume with
the best trees producing 2-3 times more than others. Estimates of
individual and family heritabilities of five traits (seven-month seed-
ling height, three- and 64-month coppice growth, height and DBH) were
not significantly different (Reddy et al. 1985). At 64 months after
harvest, the family and individual heritabilities for coppice height,
DBE eeandsavolume: were, O).05!480.65) and 0.59), “and 0.31590.32 and 0.27,
respectively.
The inbreeding coefficients (F) for possible crosses ranged from 0
to 0.25 (Table 2). The mean F value for all the matings was low (0.5%)
because only 13% of all possible crosses were related. Additionally, in
related individuals two or three generations have passed since direct
relationship by common ancestry.
Hodgson (1976a) reported a reduction in seed yield of 53% to 98%
following self-pollination. In deliberately selfed progenies of E.
grandis, a loss of 8% to 49% in height growth is reported (Eldridge
1978, Hodgson 1976b). Moreover, selfed progeny were found to be more
GENERATION INBREEDING COEFFICIENT (F)
Om25 ORIZ5 OnCOZS: 0.05025 0.0156 0,0078
Res 295 ! 294 ! 295 ! este
1 ef | | : |
836 ao SOG ONS AD) a SLO):
§36
2 856
3 101 101 8, SO S26, , 943 l
4 TOE fp 1002 I x | 1028 Ooh en Oaals
LA us
Figure 1. Representative pedigrees and inbreeding coefficients of
different degrees of relationships among families in
GPOP77.
285
Table 2. Inbreeding coefficients (F) and number of all matings in a
50-family seed orchard.
F Number of Matings
0) 1064
OF25 2
OL125 13
0.063 6
0.031 93
0.016 19
0.008 28
Mean 0.005 Total WARS)
crooked and have 30% abnormal seedlings. The amount of natural selfing
that occurs in eucalypts is higher than reported for pines. An average
of 7% selfing is reported in many pines (Wright 1976). Published
estimates for the degree of natural selfing in eucalypts vary with
species: 24% in E. obliqua (Brown et al. 1975), 37% in E. pauciflora
(Phillips and Brown 1977), 23% in E. delegatensis (Moran and Brown 1980)
and 18% in E. stoatei (Hopper and Moran 1981). Eldridge (1978) reported
20% to 40% selfing occurring in E. grandis and Van Wyk (1981) estimated
an average of 30%. Using a conservative approach for the range of
inbreeding depression reported for E. grandis, a 50% loss in growth for
selfed progenies (F=0.5) amounts to 10% loss for every 10% of inbreeding
coefficient.
Using Eldridge's (1978) estimate of 30% natural selfing in seed
orchards of E. grandis, the mean F value for the offspring of all
possible matings will increase due to selfing. Thirty percent of the
offspring will have an F of 0.5 and 70% will have an F of 0.005. Thus
the estimated inbreeding coefficient of the offspring will be 15.4%
(F=0.154). Therefore, the predicted genetic gains from the seed orchard
should be decreased by 15.4% to account for the loss due to selfing and
mating of relatives.
Based on the results from the 50-family seed orchard, similar
adjustments for inbreeding were made for the predicted genetic gains
from different selection strategies. This assumed that the mean in-
breeding coefficient for the progenies of the families in the seed
orchard is the same (F=0.005) for any selection strategy that has a
variable number of families and number of trees per family.
The predicted genetic gains that could be achieved through differ-
ent selection strategies are given in Table 3. Higher genetic gains
through family and combined selection than mass selection agrees with
the results reported in E. robusta (Dvorak et al. 1981). The greatest
gain in volume (90%) over the whole population can be achieved when
combined selection is performed, selecting the top 100 families with
three trees per family. Similar gains can be achieved by selecting one
tree from the top 300 families (96%). A gain of 80% can be realized
through mass selection of the top 200 trees. Genetic gain through
286
Table 3. Predicted genetic gains in 64-month coppice volume for alter-
native improvement strategies with Eucalyptus grandis.
Selection Strategy Genetic Gain (%)
200 best trees 80
300 best trees 69
10 top families 41
(30 trees per family)
30 top families 61
(10 trees per family)
100 top families 90
(3 trees per family)
300 top families 86
(1 tree per family)
Clonal propagation 44]
of 200 best trees
clonal propagation of the best trees shows considerable promise. A gain
of 441% is predicted through the clonal propagation of the top 200
trees. This estimate, however, is biased upward by the fact that it
assumes 100% survival, which is unrealistic. Nevertheless, it does
indicate the potential that exists in clonal propagation.
CONCLUSIONS
Natural selfing is an important factor to be considered in predict-
ing genetic gains. The estimates should be adjusted to account for the
inbreeding depression that occurs due to selfing as well as due to
mating of related families which share a common ancestor. In this study
the mean relationship among families was low (0.5). However, the
predicted gains were decreased by 15.4% to account for natural selfing
which is reported to be 30% in E. grandis.
High genetic variation for growth was observed in E. grandis with
the best trees producing 2-3 times more than the others. This suggested
the obvious potential for improvement through clonal propagation of the
best individuals to capture full genetic superiority. Impressive gains
are also predicted through combined selection of various selection
intensities. A gain of 90% could be achieved by selecting three trees
from the top 100 families.
287
LITERATURE CITED
Brown, A. H. D., A. ©: Matheson and K./G. Eldridge. | U9/5..5 EStamates! of
the mating system of Eucalyptus obliqua using allozyme polymor-
phisms. Aust. J. Botany 23:931-949.
Dvorak, We S25 HESeeG. “Franklin, Jand) iG.2 ih. 9Meskimen 1981. Breeding
strategy) for E. robusta in sowthern) Blorida. | Proc.) woth mor
For. Tree Improv. Conf., Blacksburg, VA. pp. 116-122.
Eldridge, K. G. 1978. Genetic improvement of eucalypts. Silv. Gen.
27: 205-209.
Franklin, E. C. 1978. Exotics for hardwood timber production in the
southeastern United States. Proc. 2nd Sym. Southeast. Area S
& PE, Atlanta, GAs Sppe 7-79
Geary, iep Fy, (G. (Fa) MeskimensWand E.G. ss Eranklink 1983. Growing
eucalypts in Florida for industrial wood production. Gen.
Tech. Rep. SE=23. USDA For. Serv., Southeast. For. Exp. Sta,
43 pp.
Hodgson, L. M. 1976a. Some aspects of flowering and reproductive
behavior in Eucalyptus grandis Hill ex Maiden at J. M. D. Keet
Forestry Research Station. 2. The fruit, seed, seedlings,
self-fertility, selfing and inbreeding effects. S.A. For. J.
98:32-43. ‘
Hodgson, L. M. 1976b. Some aspects of flowering and reporductive
behavior in Eucalyptus grandis Hill ex Maiden at J. M. D. Keet
Forestry Research Station. SNe Relative yield, breeding
systems, barriers to selfing and general conclusions. S. A.
FOE Jem 99s D3 — 5 Ol.
Hopper, S. D. and G. F. Moran. 1981. Bird pollination and the mating
system of Eucalyptus stoatei. Aust. J. Botany 29:625-638.
Li, C. C. 1976., First Course in Population .,Genetics.: .Boxwood.Press:.
Pacific Grove, CA.
Meskimen, G. F. 1983. Realized gain from breeding Eucalyptus grandis
in Florida. Gen. Tech. Rep. PSW-69. Pacific Southwest For.
and Range Exp. Sta., USDA For. Serv., Berkeley, (CA.
pp. 121-128.
Moran, G. F. and A. H. D.. Brown. 1980. Temporal heterogeniety of
out-crossing rates in alpine ash (Eucalyptus delegatensis).
Theor. Appl. Genet. 57:101-105.
Namkoong, G. and E. B. Snyder. 1969. Accurate values for selection
intensities. Silv. Gen. 18:172-173.
Phillips, M. A. and A. H. D. Brown. 1977. Mating system and hybridity
in E..; pauciflora. Aust. J. Biol. isei. 30:337-334;
288
Reddy, K.
Rockwood,
V., D. L. Rockwood, C. W. Comer and G. F. Meskimen. 1985.
Genetic improvement of Eucalyptus grandis for biomass pro-
duction in Florida. Proc. S. Biomass Energy Res. Conf.,
Gainesville, FL. (in press).
D. L. and G. F. Meskimen. 1981. Genetic, spacing, and
genotype x spacing influences on growth of Eucalyptus grandis
in’ “South “Ploridas” “Proc. 16th “Sie-*For. Tree “Improv. Conf .’;
Blacksburg, VA. pp. 77-85.
Van Wyk, G. 1981. Inbreeding effects in Eucalyptus grandis in relation
to degree of relatedness. S. A. For. J. 116:60-63.
Wright, J. W. 1976. Introduction to Forest Genetics. Academic Press,
New York. 463 pp.
289
CAN LATERAL ROOT CHARACTERISTICS BE A MAJOR
FACTOR IN ASSESSING SEEDLING QUALITY
Paul P. Kormanik
Abstract.--A grading standard for tree seedlings should be an easily
observed or measured characteristic that is strongly indicative of
performance after outplanting. Lateral root morphology may be such
a characteristic. Recent work with sweetgum indicates that at least
four strong lateral roots may be needed to make a seedling competi-
tive for artificial regeneration. Tests with scores of half-sib
seedlots indicate that at least 40 percent of the nursery seedlings
produced may not have this number of lateral roots developed when
they are lifted for planting.
Additional keywords: Root grading, root morphology, sweetgum.
Seedling Quality: What is it? Although foresters readily agree on the
need for quality seedlings for outplanting, few would have enough confidence
to write a prescription for judging seedling quality in the nursery. The need
for production of quality seedlings is certainly well known and equally well
documented (SIFRC 1984, Duryea and Landis 1984, Wakeley 1954). There are,
unfortunately, no reliable criteria for assessing seedling quality for any
Species of forest tree. This lack of agreement on what constitutes seedling
quality has both land and nursery managers in a serious dilemma. Development
of the technology to assess Seedling quality is among the highest priorities
throughout the United States and it is particularly important in the South
because of the large acreages of trees being artificially regenerated (SIFRC
1984).
In the early 1920's and 30's morphological grading of southern pines
seemed most promising, and it seemed to be working (Wakeley 1954). As the
grading procedure became well accepted and more universally applied during the
1930's, erratic performance of graded seedlings began to appear in many field
locations. It soon became apparent that nursery locations, different soil and
management practices were altering seedling development enough that morpholo-
gical grades from different nurseries were no longer comparable and uniform
(Wakeley 1954).
By the early 1960's as the costs of nursery stock began to increase and
when seed from genetically improved stock became more widely used, erratic
plantation performance became a major concern of land managers. Improved
techniques in nursery management did little to improve field performance of
planted seedlings, even those grown from improved seedlots. Over a period of
years, different criteria for judging seedling quality were offered but these
proved to be of limited value and now seedlings are sold predominantly by
weignt.
1 Silviculturist, Institute for Mycorrhizal Research and Development,
Southeastern Forest Experiment Station, USDA Forest Service, Athens, GA.
290
No attempt will be made here to thoroughly cover the seedling grading
standards that have been proposed, but they generally can be classified as
assessments of either morphological traits or physiological attributes. These
seedling characteristics may carry the classification of Material Performance
Attributes (Ritchie 1984), Stock Type, or Physiological Condition (Duryea
1984). Morphological criteria--something that is readily visible to the naked
eye or easily measured or assessed--for grading all planting stock is more
desirable than expensively obtained physiological parameters from a few
seedlings. Wakeley (1954), however, pointed out that any nursery practice which
improves the physiological status of nursery stock will also materially alter
the morphological grade of the seedlings and he felt the best future for
grading seedlings would have to rely heavily on some physiological criteria.
Land managers generally use morphological grades and many feel that root
collar diameter (RCD) is the best measure of seedling quality. However, Webb
(1969) cautioned against using RCD as a grading criteria for sweetgum
(Liquidambar styraciflua L.) seedlings because seedlings grown at different
nurseries or even from the same nursery but at different seedbed densities
varied considerably in RCD.
The questionable reliability of seedling physical measurements as a
grading criteria may be responsible for Burdon and Sweet's (1976) comment that
land managers aren't really interested in nursery performance of seedlings but
desire some measurable attribute on seedlings from the nursery that is well
correlated with later field performance of planted seedlings. They concluded
that possibly genotypic differences might be found within populations of
seedlings that could be used to improve field performance. Any morphological
trait that varies by nursery location or is readily altered with fertility
practices would be of questionable value.
During the past / years working with sweetgum at the Institute of
Mycorrhizal Research and Development (IMRD), Athens, Georgia, we feel that a
relatively stable morphological root relationship has been found that may be
of value in assessing seedling quality and which may be suitable for use in
grading sweetgum nursery stock. It can be of value in realistically assessing
the percentage of plantable seedlings one could expect from a given nursery
and might be helpful in judging mother trees before they are established in
orchards. Preliminary data suggests that comparable lateral root assessments
may be equally common and readily recognized on seedlings from most forest
tree species.
Early Nursery and Field Experiments
Beginning in 1973 and through 1978, numerous nursery experiments were run
on sweetgum and other hardwoods at the U.S. Forest Service's experimental nur-
sery maintained at the University of Georgia's Whitehall Experimental Forest.
The purpose of these experiments was to determine the effects of different
vesicular-arbuscular mycorrhizal (VAM) fungi and fertility regimes on seedling
development. In general, these experiments showed that hardwood seedlings did
not develop normally in the absence of VAM fungi when concentrations of
available soil phosphorus were low (12 to 25 ppm, Bray II) (Kormanik and
others 1977, 1982, Schultz and others 1981). When available soil P exceeded a
291
level near 75 ppm, however, seedling growth was not adversely affected by the
absence of mycorrhizae. If available soil P was in the range of 50 to 75 ppm,
about 95 percent of the mycorrhizal sweetgum seedlings exceeded the minimally
acceptable RCD limits of from 0.64 cm for outplanting. A comparable percen-
tage of plantable nonmycorrhizal sweetgum seedlings was produced when available
soil P was in excess of 100 ppm.
On the average early performance of mycorrhizal and high P nonmycorrhizal
seedlings were about the same--rather poor. To find out why, additional
sweetgum plantations were established on the Savannah River Forest Station,
Aiken, South Carolina, in 1977 and 1978. Fifty to sixty percent of the
planting locations in these plantations were occupied by two seedlings planted
about 30 cm apart. The plan was to excavate one of each pair without
excessive damage to the roots of the other. The original purpose of these
excavations was to follow vesicular-arbuscular mycorrhizal development in
seedlings after outplanting and to correlate stem growth with degree of
mycorrhizal development observed. Within 6 to 8 weeks after plantation
establishment, we observed that all seedling roots, regardless of original
nursery treatments, had comparable mycorrhizal development. This com-
parability in mycorrhizal development was not accompanied by uniform growth or
survival of seedlings from within the different nursery treatments. We found
that a seedling's development appeared to be correlated with the number of
lateral roots on the excavated seedlings. Unfortunately, when these early
plantations were established our primary concern was the presence or absence
of mycorrhizae on the seedlings when they were lifted from the nursery. It
was later that I thought of assessing lateral root development.
From 1978 through 1981, 8,000 to 10,000 seedlings a year from four to six
different half-sib sweetgum seedlots were grown and lifted separately at the
IMRD Experimental Nursery at the Whitehall Experimental Forest. All seedlings
were grown in the same 8 to 12 nursery beds each year at a seedbed density of
62/m2 (ca 6/ft2). Fertility and mycorrhizal variables differed somewhat among
years.
The purpose of these early experiments was to develop a preliminary
prescription for grading sweetgum seedlings based primarily on number of per-
manent lateral roots. Over the years, we had come to recognize at least three
distinct types of lateral roots occurring along the taproot of sweetgum. The
first are small, thin feeder roots seldom exceeding 2.5 cm in length which are
uniformly distributed along the entire taproot. A second type has a similar
Spacial distribution, but lacks rigidity, are threadlike and can attain
lengths of up to 12 cm. Some of these roots have diameters exceeding 1 mm and
have attached many small feeder roots of varying lengths up to about 1.0 cm.
The third type, which we consider to be a part of the permanent lateral root
system, develops primarily within 20 cm of the root collar. These roots are
rigid and have diameters from 1 to 5 mm, lengths exceeding 35 cm, and higher
orders of branching upon which abundant terminal feeder roots develop. Only
the permanent lateral roots generally withstand the rigors of lifting and
packaging in the nursery.
292
Based on data collected from field excavations, we developed a preliminary
prescription for grading sweetgum nursery stock based primarily on the number
of recognizable permanent lateral roots present on a given seedling. The
poorest (grade 3) seedlings were those with three or fewer permanent lateral
roots. The intermediate (grade 2) seedlings had from four to six permanent
lateral roots while the best (grade 1) seedlings have seven or more permanent
lateral roots. In these early nursery studies where percentage of seedlings
in each grade was being evaluated, seedlings were destructively sampled and no
plantations were established. We simply wanted to determine how nursery
fertility and different mycorrhizal symbionts affected lateral root development
for half-sib seedlots. At this time we suspected that nursery management prac-
tices would alter root morphology as clearly as it did stem morphology.
Results from Early Nursery Experiments
In 4 years of nursery testing of 18 different half-sib seedlots, the
number of grade 3 seedlings ranged from about 35 to 60 percent for different
seedlots. The average number of grade 3 seedlings annually approached about
50 percent. Specific half-sib seedlots were tested annually in up to 10 dif-
ferent nursery treatment combinations. We found that the distribution of
seedlings by root grade was comparable across all treatments for a given half-
sib seedlot, even when VAM and fertility treatments resulted in seedlings with
ranges of mean heights of 0.7/5 m to 1.0 m and mean RCD of 0.25 to 1.1 cm.
Nursery practices significantly increased seedling size, but the increases
obtained in RCDs were not normally accompanied by the development of more per-
manent lateral roots. More important, however, even the best mother trees
produced a high percentage of "“carrot-rooted" progeny--those with fewer than
three strong lateral roots. This information may be of considerable importance
for it suggests that regardless of improvement in nursery seedling stem charac-
teristics, a significant percentage of seedlings may not be genetically capable
of being competitive in nature because of limitations in root development.
During this early testing, seedlings from a mixed seedlot in a state nur-
sery and one industrial nursery were also evaluated. In both nurseries,
approximately 45 percent of the seedlings graded had three or fewer permanent
lateral roots--figures comparable to those observed in our nursery testing.
Current Testing
In each of the 1982 and 1983 growing seasons, seedlings from four half-sib
seedlots were grown in nursery beds receiving eight different treatment com-
binations. Seventy-eight seedlings were randomly selected from each
seedlot/treatment combination in both years to provide data on root grade
distribution. Four thousand seedlings from the 1982 nursery test were
outplanted in a field study with a split plot design in which the effects of
both nursery treatments, seedlots and root grades, could be evaluated. From
the four 1982 seedlots, the grade 3 seedlings represented 53, 58, 48, and 50
percent of the seedlings produced. There were no biologically significant
differences among treatments. Table 1 contains nursery seedling information
for each half-sib seedlot for all nursery treatment combinations as well as
first year growth and survival data for seedlings from all three grades.
293
Table 1.--Height and root collar diameter (RCD) of lifted seedlings, and
height, RCD, and survival at the end of the first growng season
in the field, by half-sib seedlot and root grade
Treatment and ng titel sssazZe End of growing season
root grade Height RCD Height RCD Survival
m cm m cm percent
80-58
Grade 1 1.04a 1l.3la 0.87a e380 84a
Grade 2 1.02a ealiOib 0.65b 0.96b 68a
Grade 3 1.00b 0.80c 0.40c 0.58c 52c
81-12B
Grade 1 1.04a 1.34a 0./3a 1.18a 74a
Grade 2 1.04a 1.16b 0.58b 0.89b 64b
Grade 3 1.02a 0.85c 0.29¢c OR52¢ 49¢
81-3U
Grade l 1.04a L.o5a 0.82a 1.26a 80a
Grade 2 1.05a We tilo) OF53b 0.81b 72b
Grade 3 1e02b 0.80c 0.33c OTS5e DYAG
81-5
Grade l 1.10a \1.4la 0.82a 1.32a 78a
Grade 2 l.lla lib OE S//15) 0.9Ub 66b
Grade 3 1.09a 0.82c OS32E OFSVc 50c
Within columns and treatments, values followed by the same letter do not
differ significantly (P = 0.05) according to Duncan's New Multiple Range Test.
Even though grade 3 seedlings would qualify as plantable stock by current
Standards for sweetgum, their survival percentage was not acceptable by any
standard. Drought was severe through most of the spring and summer of 1983 in
the Piedmont of South Carolina and Georgia. Survival in this plantation was
poor as it had been in all plantations in those areas. The drought impact,
however, was far more severe on seedlings with few lateral roots. Second-year
data have not been statistically analyzed, but it appears that the relatively
poor performance of root grade 3 trees is unchanged from the previous year.
Seedlings from the two better root grades are growing well and, with some half-
sibs, it is difficult to distinguish between them. Some half-sib grade 2
seedlings show greater variation in stem development than is apparent in grade
1 half-sib seedlings. This variation within grades may be reduced when new
data are available for determining how many lateral roots are required for a
seedling to be competitive.
294
The results from the 1983 nursery trials were similar to those obtained in
the 1982 tests. Heights averaged ca 0.91 m with no biologically significant
difference among grades. The RCDs for grade 3 seedlings in 1983 averaged 0.90
cm, which was significantly smaller than the 1.17 cm obtained from the root
grade 1 seedlings. However a 0.90 cm diameter would be acceptable under present
sweetgum grading standards. Half-sib seedlot 81-12B was used in both the 1982
and 1983 nursery trials. Among the eight VAM-fertilizer treatments in 1982,
an average of 58 percent of the seedlings from this seedlot were placed in
root grade 3. Comparable treatments used in 1983 resulted in 52 percent of
the seedlings from this seedlot being placed in this group. The percentage of
grade 3 seedlings from the other three 1983 half-sib seedlots were 36, 41, and
47.
Where Are We and Where Do We Go From Here
Our research to date indicates that the percentage of grade 3 seedlings in a
seedlot is quite stable and predictable. It appears that the number of lateral
roots that develop on individual seedlings may not be significantly altered by
fertility practices. At low fertility levels, although the seedlings are
smaller and the diameters of their permanent lateral roots are smaller, we
still were able to separate seedlings into different morphological root gra-
des. Under higher soil fertility regimes the seedlings were larger and the
permanent lateral roots were also larger but we were still able to separate
them from other emphermal lateral roots. Based on examination of seedlings
sent to the Forestry Sciences Laboratory, Athens, Georgia, from different nur-
series, soil texture appears to have limited affect on lateral root numbers
and distribution. Both soil fertility and soil texture, of course, can
affect the feeder root development but this has not affected the number of
permanent lateral roots produced in our tests.
We now need to determine how extremes in seedbed density affect the
development of permanent lateral roots and our ability to distinguish root
grades. This is the objective of the 1985 nursery testing program. We do
know, however, that diameters of lateral roots of seedlings in the interior of
the beds are significantly smaller than those seedlings lifted from the exterior
border rows. The data do not indicate, however, that the number of lateral
roots from the border rows represent a different distribution than occurs on
the interior seedlings. Within a given half-sib seedlot, I do not believe it is
very important how big the permanent lateral roots are initially, within
reasonable limits, but rather how many are present just as long as we can iden-
tify them as permanent or emphermal.
There is little doubt that there exists a distinct distribution of lateral
root grades on sweetgum seedlings lifted from the nursery. This distribution
appears to be stable from selected half-sib progeny and a comparable distribu-
tion has been found with seedling populations obtained from mixed seedlots of
unknown parentage. Extensive nursery trials show that 40 to 50 percent of all
seed germinated produce seedlings with fewer than four permanent lateral roots,
which may be a minimum for satisfactory early plantation performance. This
assumption must be tested in the field over time but two sweetgum plantations
established in 1982 and 1983 appear to support it.
295
The ease with which aboveground characteristics can be changed without
altering lateral root characteristics may in part explain why improved nursery
practices, even with seed from selected trees, have not resulted in improved
performance of nursery stock in the field. If lateral root distribution is
positively correlated with plantation performance, a biological grading scheme
may be fairly easy to develop. Standards, of course, will probably vary by
species. Our early data indicate that lateral root characteristics are quite
different for the seven different commercially important forest species we
have examined. Six lateral roots may be sufficient for sweetgum seedlings to
be competitive, but oaks (Quercus spp.) and black walnut (Juglans nigra) may
need twice this number.
Recent surveys indicate that productivity of plantations is not as good as
predictions indicated. The cause may be production of seedlings with large
tops but without accompanying large root systems. If we are correct in our
assumption that number of lateral roots occurring on young seedlings is a good
estimator of potential productivity of that seedling and that this trait may
be predictable within a given species, the economic gains in yield per hec-
tare would warrant extensive trials of biological grading. Unfortunately, at
this time, we may not be able to alter root morphological characteristics as
readily as we can change aboveground characteristics. Thus, we must determine
how important lateral root development of seedlings really is and be prepared
to cull many seedlings if marginal plantation performance is shown to be
correlated with lateral root development.
What is a quality seedling? Could it be one with at least a specific
number of lateral roots?
‘ LITERATURE CITED
Burdon, R. D.; Sweet, G. B.. The problem of interpreting inherent differences
in tree growth shortly after planting. In: Cannell, M. G. R.3 Last, F. T.,
eds. Tree physiology and yield improvement. London: Academic Press; 19/76:
483-502.
Duryea, M. L. Nursery cultural practices: impacts on seedling quality. In:
Duryea, Mary L.; Landis, Thomas D., eds. Forest nursery manual, production
of bareroot seedlings. Boston: Marinus Nijhoff/Dr W. Junk Publ.; Forest
Research Lab., Oregon State Univ., Corvallis. 1984: 143-164.
Forest Nursery Manual: Production of bareroot seedlings. Duryea, Mary L.;
Landis, Thomas D., eds. Boston: Marinus Nijhoff/Dr W. Junk Publ.; 1984:
885) pe
Kormanik, P. P.; Bryan, W. C.; Schultz, R. C. Influence of endomycorrhizae
on growth of sweetgum seedlings from eight mother trees. For. Sci. 23:
500-506; 1977.
Kormanik, P. P.3 Schultz, R. C.; Bryan, W. C. The influence of vesicular-
arbuscular mycorrhizae on the growth and development of eight hardwood
CREM SDeCIIES | EOI SGils (Con wool DSO mg oee
296
Ritchie, G. A. Assessing seedling quality. In: Duryea, Mary L.; Landis,
Thomas D., eds. Forest nursery manual, production of bareroot seedlings
Boston: Marinus Nijhoff/Dr W. Junk Publ.; Forest Research Lab., Oregon
State Univ., Corvallis. 1984: 243-259.
ScnUlezerie (Gas NOGMdNa Ks hs Pes Bayan, We. C.. Effects of fertilization and
vesicular-arbuscular mycorrhizal inoculation on growth of hardwood
Seedlings. Soll Sei. Soc. Am. UJ. 45: 961-965; 1981.
Southern Industrial Forestry Council (SIFRC). Rep. No. 3. Am. Pulpwood
Assoc. Inc., Jackson, MS, and Southern Forest Products Assoc., New Orleans,
LA. 1984.
Wakeley, Philip C. Planting the southern pines. Agric. Monogr. No. 18.
Washington, DC: U.S. Department of Agriculture, Forest Service; 1954.
2333) Do
Webb, Charles D. Uniform seedling density is important in hardwood progeny
test nurseries. In: Proceedings of the tenth southern conference on forest
tree improvement; 1969 June 17-19; Houston, TX: 1969: 208-216.
297
CONIFER GENETICS Ill
MODERATED BY DR. HANS VAN BUIJTENEN
Texas A&M University
298
i
id
- a
'
Ta Be
hal
VAS
aa Fa | ha ;
AASY
MONOTERPENE PHENOTYPES IN LOBLOLLY PINE POPULATIONS:
NATURAL SELECTION TRENDS AND IMPLICATIONS
A.E. Squillace, Harry R. Powers, Jr., and S. V. Kossuth!
Abstract.-- The degree of discrepancy between observed pro-
portions of various monoterpene phenotypes involving two or
more loci and those expected under random association be-
tween loci was studied in 111 populations scattered through-
out the range of loblolly pine. Results suggest that some
phenotypes are being favored by natural selection while
others are being disfavored. Natural selection varies among
regions and appears related to variation in resistance to
fusiform rust.
Additional keywords: Linkage disequilibrium, fusiform rust.
Previous work has shown that contents of four of the major monoterpenes
in cortical oleoresin of loblolly pine (Pinus taeda L.) are largely con-
trolled by single genes, with high content being dominant over low in all
cases (Squillace et al. 1980). The four loci involved have also been shown
to be rather closely linked (Squillace and Swindel, in press). The objec-
tives of the present study are to: (1) examine deviations between observed
proportions of phenotypes involving two or more monoterpene loci and those
expected under random association between alleles at different loci (link-
age disequilibrium) in loblolly pine populations, (2) interpret the results
from the standpoint of natural selection, and (3) seek relationships with
resistance to fusiform rust (Cronartium quercuum f. sp. fusiforme).
MATERIALS AND METHODS
In this study we utilized data previously reported by Squillace and
Wells (1981) and McRae and Thor (1982). In those reports, cortical mono-
terpenes were sampled in trees from a total of 111 loblolly pine popula-
tions scattered throughout the species range (Table 1). Most of the trees
were in seed source study plantations containing trees of known geographic
origin. Details on sampling and analytical techniques are available in the
two publications cited. In some of the plantations sampled by Squillace
and Wells (1981), data on occurrence of fusiform rust infection were also
available and were utilized in interpreting results of the disequilibrium
analyses.
: Squillace is Adjunct Professor, University of Florida, and Volunteer,
USDA Forest Service, Gainesville, FL. Powers and Kossuth are Project
Leaders, Southeastern Forest Experiment Station, USDA Forest Service,
Athens, GA and Gainesville, FL, respectively. The study was also sponsored
by the Georgia Forestry Commission. Journal Series Paper 6469 of the
Florida Agricultural Experiment Station.
299
Table 1.--Numbers of loblolly pine populations and trees sampled for
cortical monoterpene composition by regions of the southern
U.S.
Region?/ Populations Trees
Western 21 565
Central 85 2141
Northeastern 18, 143
Totals 111 2849
a/ Western = populations west of the Mississippi River. 7
Central = populations between western and northeastern regions.
Northeastern = populations in Virginia, Maryland, and Delaware.
In order to study linkage disequilibrium between pairs of loci, we
first determined the frequencies of trees in each phenotypic class. Since
dominance occurs at each locus, there are four possible phenotypes for each
pair of monoterpenes. For example, designating two monoterpenes as A and
B, with capital letters representing high amounts and lower case letters as
low amounts, the four possible phenotypes are:
1. AB, which includes genotypes AABB, AABb, AaBB and AaBb
2. Ab, which includes genotypes AAbb and Aabb
3. aB, which includes genotypes aaBB and aaBb
4, ab, which includes only aabb genotypes
An estimate of linkage disequilibrium, D, is given by the following
(Cavalli-Sforza and Bodmer 1971, Hill 1974)
Xu (x2 + xy) (x3 + Xy)
n n
o>
"
t
in which X»5, X3, and x, are numbers of phenotypes Ab, aB, and ab in the
population, respectively, and n = total number.
If the product of the AB and ab phenotypes (coupling types) exceeds the
product of Ab and aB phenotypes (repulsion types), D will be positive. If
the reverse is true, D will be negative. A positive D indicates that the
proportion of coupling types observed is greater than that expected from
random association. It can mean, for example, that natural selection is
favoring coupling over repulsion types. A negative D would suggest the
reverse.
To test the significance of D, we used the likelihood criterion (K)
given in Hill (1974):
300
4n D2
K = , which is a x2 distribution, with 1 d.f.,
A A a
p (2-p) q (2-q)
where p = estimated frequency of the A allele = l- NEL als
n
and q = estimated frequency of the B allele = 1- a ea
n
As an example, D and K will be computed for 8-pinene vs. myrcene in
population #2 (Marion Co., FL). The numbers of phenotypes BM, Bm, bM,
and bm were 14, 2, 6, and 1, respectively (summed from the second column of
table 2). Substituting these values into the above equations, we obtain: D
=0094 pe= 448: q = 2639, and k = .01.
Since our data involved four loci, we were also interested in determin-
ing evidence of disequilibrium that may occur among four-locus phenotypes.
With four loci showing dominance, there are 16 possible phenotypes as shown
in Table 2. We could find no procedure in the literature for estimating
and testing linkage disequilibrium in such cases and hence used the follow-
ing procedure to get indications of natural selection favoring or disfa-
voring each phenotype. Expected frequencies (proportions) of each phenotype
were computed on the basis of frequencies of single-locus phenotypes and
these were subtracted from observed phenotypic frequencies. We shall des-
ignate the differences by D'.
Table 2.--Computation of observed-expected proportions of four-gene phenotypes (D'),
for Population #2, Marion Co., FL
a/ b/
Phenotype — Observed # Observed prop. Expected prop. — D'
BMLP 2 .087 -101 -.014
BMLp 0 .000 -005 -.005
BM&P ll -478 .478 .000
BM2p 1 .043 .022 -022
BmLP 1 .043 -015 -028
BmLp 0 -000 -001 -.001
BmzP 1 -043 .072 ~.028
Bmip 0 -000 .003 -.003
bMLP 0 .000 044 -.044
bMLp 0 -000 .002 -.002
bM&P 6 .261 -209 -052
DMZ p 0 -000 -010 -.010
bmLP l -043 -007 -037
bmLp 0 -000 -000 -000
bmeP 0 .000 .031 -.031
bmg p 0 .000 -001 -.001
Totals 23 .998 1.001 .000
a/
— 8B, M, L, and P represent high amounts of §8-pinene, myrcene, limonene, and
3-phellandrene, respectively, wnile lower case letters represent low amounts.
y/ Computed from observed proportions of one-gene phenotypes, summed from col-
umn three, above: B = .696, b = .304, M = .870, m= .130, L = .174, Q= .826,
P = .956, and p = .044, Thus, for example, the expected proportion of BMLP
trees is (.696)(.870)(.174)(.956) = .101. See text.
301
Thus, an estimate of D' for phenotype BMLP was computed as:
D if =o Nips all se uienanee
BMLP _'BMLP BML 'p?
in which fRaMLP is the observed frequency of the BMLP phenotype in the popu-
lation, and fps fy> f and fp are the observed frequencies of B, M, L, and
P phenotypes in the same population, respectively. D' values were likewise
obtained for the other 15 phenotypes:
D foe cat eet uaa
BMLp ‘BML ‘p?
pmep ~ ‘pep ~ ‘ptm fy fp»
BMLp —
D
bmgp ~ *
cif Pair
2 bm] p Deal: gies ips
An example is given in Table 2. The above procedure was also used for
getting estimates of disequilibrium for groups of populations (regions).
It should be noted that expected four-locus phenotypic frequencies can also
be computed from observed frequencies of individual alleles--rather than
using observed frequencies of individual phenotypes--with the same result,
but the latter procedure is simpler.
The numbers of trees in phenotypes within individual populations were
too few to make reliable tests of significance because in most populations
many phenotypes were represented by four or less trees, preventing reliable
use of chi-square tests. However, we tested significance of average dif-
ferences for each phenotype across all populations, using a t-test with the
null. hypothesis that the observed-expected values = zero. When pooling
populations within regions, we tested for significance by computing
2
(observed number-expected number)
expected number
for each phenotypic class in which expected numbers were five or more and
summing these to obtain y*. Degrees of freedom here were presumed to be
total number of four-locus phenotypes minus number of single-locus pheno-
types used in computing expected values.
RESULTS |
Two-locus Phenotypes |
Significant pair-wise linkage disequilibrium was found rather frequent-
ly (Table 3). Myrcene vs. limonene was especially notable--84 of the 98
302
populations permitting this test showed negative D values, with 31 of them
being significant. None of the positive values were significant. Curious-
ly, the populations showing positive values were clustered in three areas:
Table 3.--Results of linkage disequilibrium (0) analyses in 111 loblolly pine populations
Loci compared Range of Positive n) values Negative D values
D values Total no. af No. significant by Total no. a/ No.significant Ly
p-Ppinene & myrcene (B,M) 0.08 to -0.27. 17
B-pinene & limonene (B,L) «L3'to,= .27 43 ; i in
6-pinene & g-phellandrene (B,P) .22 to - .23 8 1 43 ;
Myrcene & limonene (M,L) -07 to - .37. 14 0 84 :
Myrcene & g-phellandrene (M,P) .15 to - .27. 31 1 46 -
Limonene & g-phellandrene (ESP) Pe aliator=9 27) 48 1 22 :
a/
— The total number of positive and negative values in each c
omparison are less than 111
populations both the observed and expected proportions of phenotypes were zero, negating eae ee
by Significant at 0.05.
southwestern Alabama, southeast Georgia-northeast Florida, and the Caroli-
nas. The results suggest that M& and mL phenotypes are being favored by
natural selection, while ML and mg types are being disfavored in most por-
tions of the species range. Note also that D values forg -pinene vs. myr-
cene and for myrcene vs. 8-phellandrene also tended to be negative,
judging from both numbers of negative vs. positive values and significance
(Table 3). These findings suggest that BmLP trees are being favored by
natural selection, which will be examined further.
Four-locus Phenotypes
Analyses of four-locus phenotypes in individual populations showed that
observed frequencies of some phenotypes exceeded expected values while for
others the reverse was true (Table 4). Phenotype BmLP had both the great-
est positive average deviation and the greatest proportion of populations
showing positive deviations. Values obtained for this phenotype are plot-
ted in Figure 1. Note that positive values prevail in all areas except
southwest Alabama, southeast Georgia-northeast Florida, and the Carolinas.
Thus, aS suggested earlier, BmLP phenotypes seem to be favored by natural
selection over most portions of the species range. Phenotype BM&P also
showed a significant trend toward positive deviations. Finally, the rela-
tively rare phenotype bMgP also showed a significant but small positive
average deviation. These results may be partly a reflection of the very
strong negative disequilibrium between myrcene and limonene (M& and mL
types being favored), noted earlier.
303
Table 4.--Results of analyses of observed vs. expected proportions of
each monoterpene phenotype, within populations
TT TTT Oo ———————__
Phenotype ay No. of b/ Percent of Average t value G/
populations in test — populations deviation
showing + deviations over all
populations
Bie ee
BMLP 102 18 -.039 9.38**
BMLp 71 19 -.005 2.44*
BMgP 109 68 -030 Do /ehe
BMgp 79 55 -004 1.12
BmLP 100 82 -043 D9 =*
BmLp 70 22 -002 88
BmeP 107 25 -.034 Voices
Bmgp 78 31 -.001 24
bMLP 61 21 -.001 252 r
bMLp 45 6 -001 12
bMeP 67 66 -012 3.64**
bMep 52 21 -.001 -67
bmLP 60 14 -.001 57
bmLp 45 4 -001 -56
bmgP 66 12 -.010 4.92**
bmep 52 2 -.001 1.68
TY {- -"'IXYX ——-rrrO ss
a/ B, M, L, and P represent high amounts of -pinene, myrcene, limonene, and
B - phellandrene, respectively, while lower case letters represent low amounts.
b/ These values are less than the total (111 populations) because in many cases
both the observed and expected proportions were zero, in which case no test
was possible.
c/
=" Test of the hypothesis that the population average is zero, with N-l d.f.
An * = significant at the 0.05 level; ** = significant at the 0.01 level. a
\S
13
Figure 1.--Observed-expected percent of BmLP phenotypes. Clusters of zero
and negative values are outlined.
304
Analyses in which populations were pooled by regions gave similar results
(Table 5). However, the deviations of observed vs. expected values were
generally largest in the central region, smaller in the West, and very
small in the northeast.
Table 5.-- Observed (0) and expected (E) numbers of trees in each of 16 monoterpene
phenotypes within regions?/ and tests of significance
b/
Phenotype — Western populations Central populations Northeastern populations
) E (0-E)? 0 E (0-E)? 0 E (O-E) 2
E E E
BMLP 140 177 UGU 249 348 28.2 3 3 --
BMLp 0 1 -- 21 35 5.6 0 0 --
BM £P 78 44 26.3 1039 953 7.8 89 91 .0
BM gp 0 0 -- 107 95 1.5 8 6 of,
BmLP 307 268 5.7 260 143 95.7 2 1 --
BmLp 1 1 -- 20 14 2.6 0 0 --
Bm 2P 30 66 19.6 290 392 26.5 21 21 0
Bm £p 1 0 -- 34 39 6 1 2 --
bMLP 4 2 -- ll 21 4.8 0 1 --
bMLp 0 0 -- 2 2 -- 0 0 --
DM £P 3 1 -- 80 57 9.3 16 14 <3
bM £p 0 0 -- 8 6 ot 0 1 --
bmLP 1 4 -- 9 9 0 0 0 --
bmLp 0 0 -- 1 1 -- 0 0 --
bm @P 0 1 -- 9 24 9.4 3 3 --
bm sp 0 0 -- 1 2 -- 0 0 --
Totals SI 365 565 59.3** 2141 2141 192.7** 143 143 1.0
wp” Rass les Ty Sega Se ae ee ee eee ee ee
a/
Western = populations west of the Mississippi River.
Central = populations between western and northeastern regions.
Northeastern = populations in Virginia, Maryland, and Delaware.
b/
B, M, L, and P represent high amounts of §-pinene, myrcene, limonene, and
8 -phellandrene, respectively, while lowercase letters represent low amounts.
S/ the totals of (O-E) 2/E are chi-squares, with 12 d.f. Classes having fewer than
five expected values were omitted in computing x2.
** = significant at the 0.01 level.
IMPLICATIONS
The results strongly suggest that natural selection in most portions of
the range of loblolly pine is favoring certain monoterpene phenotypes and
disfavoring others. Here we examine possible reasons.
305
Note first that the phenotype which seems to be most strongly favored
by selection (BmLP) is very prevalent in the western region, °°7/,.,. = 54
percent being of this type (Table 5). It is much less prevalent in the
central region (12 percent) and almost absent in the northeast. The geo-
graphic pattern is shown more clearly in Figure 2. Note further that jin
many respects it conforms to regional patterns of resistance to fusiform
rust (Grigsby 1973, Squillace and Wells 1981). Western trees have large
Figure 2.--Percent of BmLP phenotypes.
proportions of BmLP trees and are relatively resistant, and this rela-
tionship tends to extend into western Mississippi, although to a lesser
extent. Populations in the panhandle of Florida and in southeast Georgia-
northeast Florida have few BmLP trees and are very susceptible to rust.
The pattern fails in the northeast where BmLP trees are scarce and re-
Sistance is high. We shall return to this point later.
With these observations in mind, we hypothesize that natural selection
is favoring BmLP trees because they tend to be more resistant to fusiform
rust than other types. Data from progeny tests reported by Squillace et
al. (1984) also suggest that BmLP trees tend to be more resistant than
other types. Perhaps natural selection has been favoring such trees in the
West over a longer period than in the central region, explaining its great-
er prevalence in the West.
306
A somewhat similar situation seems to exist for phenotype bM&P, which
also tended to be favored by selection. The proportion of such trees was
11.2 percent in the northeast, 3.7 percent in the central region, and 0.5
percent in the West (Table 5). As is well known, northeastern populations
tend to be relatively resistant [see, for example Grigsby (1973)]. Thus,
natural selection for this phenotype may also be a reflection of resistance
to rust.
CONCLUSIONS
Natural selection is definitely favoring BmLP trees in both the west
and central regions of loblolly pine. This phenotype presently comprises a
large proportion of Western populations, which are relatively resistant to
fusiform rust. Although it is now rather infrequent in central populations,
it is presumably increasing with each generation. We hypothesize that BmLP
trees tend to be more resistant to fusiform rust and that this is why they
are favored by natural selection. A similar situation seems to occur for
bMgP trees, although this is less certain. The latter phenotype is rela-
tively most prevalent in the northeast, where resistance to rust also oc-
curs and it is presumably being favored by natural selection, especially in
central populations.
The nature and degree of the relationship between monoterpenes and
fusiform rust resistance is still unclear. Although significant rela-
tionships were found on a regional basis (Squillace and Wells 1981), we
have not yet completed studies comparing trees of different phenotypes
within families and populations. We do not believe monoterpene composition
actually affects resistance--more likely it may be an indicator of the
presence of some types of resistance. It is possible, for example, that
the monoterpene phenotypes being favored by natural selection are results
of hybridization or introgression with other species, such as shortleaf
pine (P. echinata Mill.)(Hare and Switzer 1969) and pond pine (P. serotina
Michx.)(Saylor and Kang 1973). Studies of cortical monoterpenes in these
species would be desirable. Also, it would be desirable to artificially
inoculate trees having different monoterpene phenotypes and observe their
reactions to the rust.
LITERATURE CITED
Cavalli-Sforza, L.L. and W.F. Bodmer. 1971. The genetics of human popula-
tions. Freeman & Co., San Francisco, CA. 965 p.
Grigsby, H.C. 1973. South Carolina best of 36 loblolly pine seed sources
for Southern Arkansas. USDA For. Serv. Res. Pap. S0-89. 10 p.
Hare, R.C. and G.L. Switzer. 1969. Introgression with shortleaf pine may
explain rust resistance in western loblolly pine. U.S. Forest Serv.
Res. Note S0-88. 2 p.
307
Hill, W.G. 1974. Estimation of linkage disequilibrium in randomly mating
populations. Heredity 33:229-239.
McRae, J. and E. Thor. 1982. Cortical monoterpene variation in 12 loblolly
pine provenances planted in Tennessee. For. Sci. 28:732-736.
Saylor, L.C., and K.W. Kang. 1973. A study of sympatric populations of
Pinus taeda L. and Pinus serotina Michx. in North Carolina.
Squillace, A.E., H.R. Powers, Jr., and S.V. Kossuth. 1984. Relationships
between cortical monoterpenes and fusiform rust resistance in loblolly
pine. Abstract in Southwide Forest Disease Workshop, Long Beach, MS.
Squillace, A.E. and B.F. Swindel. In press. Linkage among genes controlling
monoterpene constituent levels in loblolly pine. (Accepted by Forest
Science)
Squillace, A.E., and 0.0. Wells. 1981. Geographic variation of monoter-
penes in cortical oleoresin of loblolly pine. Silvae Genetica 30:
127-135.
Squillace, A.E., 0.0. Wells, and D.L. Rockwood. 1980. Inheritance of mono-
terpene composition in cortical oleoresin of loblolly pine. Silvae
Genetica 29:141-151.
308
\
ONE-QUARTER CENTURY OF TREE IMPROVEMENT
ON NATIONAL FORESTS IN THE SOUTHERN REGION
Robert N. Kitchens 7,
Abstract.--In 1983, the Southern Region of USDA Forest Service
Harvested over 15 tons of first generation orchard seed from six seed
orchards. This represented a milestone for a tree improvement program
that began in 1959. That was the year researchers and foresters made
plans for breeding improved trees for restocking National Forest lands.
The program is large and complex, encompassing some 13 species and 52
geographic sources, and serves a land base of about 10 million acres.
Some tough seed orchard management problems were solved along the way -
examples, Net Retrieval system for seed harvest and aerial application of
pesticides. Payoffs are impressive by any measure. Early progeny test
results indicate that large gains in volume and other traits can be
expected through genetic tree improvement. Over 400 acres of progeny
tests have been planted mainly to provide a source of selections for
another generation of breeding. The Second Generation Plan has been
developed and is being implemented.
Additional keywords: Seed production, progeny testing, hardwood tree
improvement, net retrieval system.
The Southern Region's Tree Improvement Program is one of the really great
success stories of the USDA Forest Service. In the last 2 years, 47,000
pounds of first generation orchard seeds were harvested. Through the 1985
planting season, approximately 325,000 acres of improved trees have been
planted on National Forests in the South. This acreage is increasing at the
rate of about 50,000 acres per year.
Payoffs are great by any measure. These trees will produce 10 to 20
percent more volume than average wild trees. They are also bred to be
straighter, more disease-resistant, and to have better wood qualities than
their wild cousins.
It took 25 years of hard work by many people to get to this point, but the
most important phase lies just ahead--that of second generation superior
trees. These trees are expected to grow a whopping 35 to 45 percent faster
than the wild population:
First Generation Program
In 1959, under the guidance of Thomas F. Swofford, the first Regional
Geneticist, planning for the program was started. He consulted many people
and organizations in developing the program. The following is a list of some
who contributed and deserve recognition: John Kraus, Bruce Zobel, Hans Van
1/Regional Silviculturist, Timber Staff Unit, Southern Region, Atlanta, Georgia
309
Buijtenen and Ray Goddard, Southern, Southeastern and Southwestern Forest
Experiment Stations, and Tree Improvement Cooperatives at North Carolina State
University, University of Florida, and Texas A & M University. Programs
already underway in Portugal, Sweden and England were also examined. Swofford
retired in 1975 and was replaced by Jim McConnell, the present Regional
Geneticist.
The pine program was started first. Actual selections begun in 1961 and
were essentially completed by 1967. Thirty-eight species-geographic source
combinations were recognized in the original selection program; consisting of
shortleaf (Pinus echinata Mill.), loblolly (P. taeda L.), longleaf (P.
palustris Mill.), slash (P. elliottii Englem.), eastern white (P. strobus 1G)
Virginia (P. virginiana Mill.), and Ocala sand pines (P. clausa var. clausa
D.B. Ward) [McConnell 1978]. All original orchard sites and species were
successful except for sand pine. It was first included as part of the
Erambert Seed Orchard in south Mississippi. Survival and growth was poor so
the sand pine orchard was moved to central Florida, which is in the native
range of the species.
Approximately 6 million acres of National Forest pinelands were
intensively searched for the very best trees as candidate parent trees for
first generation orchards. After a candidate was found, it had to pass
several screens before it was finally accepted. Faster growth, pruning
ability, straightness, disease resistance, and specific gravity were the
traits sought after in the,superior tree selections. About 50 selections for
each species-geographic source were approved. Then the selections were
grafted into clonal orchards. The Ocala sand pine orchard has both a clonal
orchard and a 120-family seedling seed orchard. The Region now has 2,177 pine
selections in 1,256 acres of pine seed orchards at 6 orchard sites.
The first collectible crop of seed was harvested in 1970; through 1984,
collections have totalled 81,000 pounds. Most pine sources in the program are
now producing enough seeds for total planting requirements.
The hardwood program started in 1968. Six species are in the program-—-
black (Quercus velutina Lam. ), white (Q. alba L.), northern red (Q. rubra
L.), and chestnut oaks (P. prinus L.), cherry (Prunus serotina Ehrh.) and
yellow-poplar (Liriodendron tulipifera L.). To-date, 382 selections have been
made for clonal orchards and 29 acres of clonal orchards have been
established. A 220-family, 16 acre, northern red oak seedling seed orchard
was established, which was originally a Tennessee Valley Authority progeny
test on National Forest land.
Of the hardwood clonal orchards, only yellow-poplar and black cherry are
producing enough seed for operational plantings. Sure-fire techniques for
successful oak seeding or planting are still not developed; however, within a
few years, crops will be harvested from the orchard and attempts at using them
for reforestation will be made.
Managing first generation orchards presented some unique problems that had
to be solved. Foremost was how to harvest all the cones or seeds without
harming the trees. During regular woods cone collection, trees were usually
310
cut down and the cones picked; however, since cutting trees was not possible
in a seed orchard, various methods were used to place people in the trees,
including ladders and bucket trucks. Besides being expensive and slow, these
methods were somewhat hazardous.
With the cooperation of the Georgia Forestry Commission, a new system was
developed, called the Net Retrieval System. Netting was placed on the ground
where the seeds fell, and a combine-type machine was used to roll the net and
separate the seeds. The Net Retrieval System is now in operation on all or
part of 4 Forest Service orchards (Edwards and McConnell 1983, McConnell and
Edwards 1985), and other organizations are considering using this system.
Because a seed orchard has many trees of the same age, it is an attractive
home for insects--especially those that like to eat cones and seeds. Safe and
effective ways had to be found to control these seed-destroying insects.
Entomologists worked closely with orchard managers on pesticide formulation,
application, and timing for effective control. With the help of several
organizations, technology for the aerial application of insecticide was
developed. Now an orchard can be treated in hours instead of weeks that were
required for ground application methods. In addition, aerial applications
place the insecticide in the top portion of the crown, where the cones are.
This means less insecticide is necessary to do an effective job.
Progeny Testing
In 1974, controlled crosses among orchard trees began according to a plan
that employed disconnected half-diallels for all species except sand pine.
Individual matings were made to match desirable characteristics as indicated
by the original scoring sheets, fusiform rust disease resistance screening
tests, and progeny performance (McConnell 1983). Over 9,500 individual
crosses will have been made when the plan is completed.
Progeny testing was done to; (1) measure gains, (2) test worth of parents,
and most importantly, (3) as a source of selections for second generation
orchards. A few open-pollinated tests were installed, mainly for demonstra-
tion purposes. To date, over 250 tests have been planted representing about
6,500 families. About 20 percent of the tests are 5 years old or older.
Early results have been quite surprising. Of course, early results must be
used with a great degree of caution. Nevertheless, they indicate that large
genetic gains can be made.
A white pine open-pollinated test at the Cradle of Foresty, on the Pisgah
National Forest in North Carolina, showed orchard trees to have a 25 percent
superiority in diameter growth (dbh) and a 15 percent superiority in height
growth over general forest area stock at age 5. The 10-year results for the
same test showed an accelerating difference--28 percent in height and 36
percent in diameter. The 10-year mean for orchard stock was 24.75 feet tall
and diamater of 4.97 inches; for general forest area stock the respective
Means were 19.25 feet and 3.65 inches.
311
Another important result, and one expected by geneticists, was that the
range of trait variation for height and diameter was the same for seed orchard
material as general forest area stock. Only the mean of the two populations
was different. This is evidence that orchard populations will continue to
have large amounts of variation for some traits.
One of the largest actual heights and diameters occurred in a loblolly
pine test in Southern Mississippi. At age 5, the average of all orchard
families was 16.6 feet, the tallest family was 18.2 feet, and the tallest
individual tree was 29.1 feet.
Other early results are quite impressive. However, the number of tests
analyzed is small relative to the total number planned. During the next few
years many more tests will be analyzed so that greater confidence can be
placed in the gain percentages.
Plans are to use 8- to 10-year test results to begin making selections for
second generation orchards. That time is almost here.
Second Generation Breeding
The second generation plan for pines has been developed. Actually, it
goes beyond the second generation because selection of new genetic material to
infuse into second and successive generation breeding is also planned.
Y
As stated earlier, since second generation gains are expected to double
first generation gains, full speed ahead is in order.
Orchard site selection has already begun. In general, second generation
Sites will be near first generation sites in order to efficiently utilize
present personnel, facilities, and equipment.
The 38 pine geographic source-species combinations used in the first
generation were streamlined into 20 breeding populations for the second
generation. The breeding populations are based on seed movement and planting
zones as defined by research results in most cases, but a few were designed
using a combination of intuition and/or administrative necessity (Wells and
McConnell 1983). Reducing the breeding population to 20 will increase program
efficiency. It also gives a broader genetic base for second generation
selections since some first generation populations were combined.
The 20 breeding populations for the second generation have been
prioritized based on the species importance in National Forest reforestation i
and on progress of first generation progeny tests. The highest priorities
will be developed first and others will be done as timing and budgets allow.
Other flexibilities have also been built into the plan so that developing
technology can be incorporated along the way.
Guiding Principles
Several guiding principles have been used which contributed greatly to the
success of the program.
312
Knowledge and experience of a large part of the tree improvement community
have been drawn upon in formulating strategic plans and critiquing the
program. The Regional Geneticist and a small staff are responsible for
Strategic planning, but they consult frequently with many others. John Kraus
and Ozzie Wells, Southeastern and Southern Forest Experiment Stations
respectively, are constant advisors. The interchange of tree improvement
information under the umbrella of the Southern Forest Tree Improvement
Committee is used. One almost has to be a part of tree improvement in the
South to fully realize the spirit and degree of cooperation within this
community.
Maximum involvement of Forest and District personnel is fostered. There
is no tree improvement organization as such below the Regional level. Forest
Supervisors are responsible for the program on their respective forests.
Orchard Managers and workers who do actual test plantings and measurements are
on the District Ranger's staff. By having this type of involvement, ownership
of the program is vested in all levels of the organization and each level
takes pride in program accomplishments. Work is accomplished in a timely
fashion and the quality of work is high.
The Tree Improvement Program is continually evaluated with respect to
current silvicultural practices used on National Forest lands. Greater
productivity is the goal of tree improvement. Greater productivity cannot
take place without the proper integration of genetics and silviculture. [In
fact, on the National Forests, tree improvement is viewed as an integral part
of silviculture.
Closing
A quarter-century has brought the Southern Region's program a long way,
but the challenge to breed another generation of trees to reach higher
production goals is great and exciting. The next 25 years will no doubt bring
accomplishments unimaginable today.
LITERATURE CITED
Edwards, J. L. and McConnell, J. L. 1982. Forest tree seed harvesting system
for loblolly pine. ASAE paper No. 82.1589, Winter Meeting, 10 p.
McConnell, J. L. 1978. Region 8 program. In Proc. Service-wide conference on
Genetics, p. 94-97. USDA Forest Service, Washington, D.C.
McConnell, J. L. 1983. Progeny tests - R8 objectives and design. In
Proceedings Servicewide Genetics Workshop on Progeny Testing, p. 258-259.
USDA Forest Service, Washington, D.C.
McConnell, J. Le. and Edwards, J. L. 1985. The net retrieval seed collection
system for Southern Region seed orchards--an economic study. In Proceedings
Third Biennial Southern Silvicultural Research Conference, pp. 252-254. USDA
So. For. Ex. Sta., Gen Tech. Report SO-54, New Orleans, LA.
Wells, O. O. and McConnell, J. L. 1983. Breeding populations in the R8
tree improvement program. In Proceedings Service wide Genetics Workshop on
Progeny Testing, p. 61-67. USDA Forest Service, Washington, D.C.
313
THIRD-YEAR COMPARISONS OF LOBLOLLY AND SLASH PINE SEED SOURCES FOR FUSIFORM
RUST RESISTANCE AND GROWTH POTENTIAL IN NORTH CENTRAL FLORIDA
John A. Pait, III, Lee Draper, Jr., and Robert A. Schmidt 2/
ABSTRACT
Four loblolly and three slash pine seed sources were evaluated for
rust resistance and growth potential in a provenance test established in
1981 in a high rust incidence area in northcentral FL. The site was
moderately well-drained and the soil was a sandy loam overlying clay at 24
inch depth; site index was 75 ft at 25 years. Loblolly pine seed sources
were East Texas, Livingston Parish, Marion County, FL and FL seed orchard.
Slash pine seed sources were FL and GA seed orchards and a rust-rogued seed
production area in North FL. Survival, diameter, height and rust incidence
(% trees with one or more galls) were measured after the third growing
season. Seed sources within species only were compared statistically.
Within loblolly sources, survival, diameter and height were greatest in
Marion County and least in East Texas; Livingston Parish and FL seed
orchard were intermediate. All loblolly sources were significantly differ-
ent for rust incidence. East Texas was least rust-infected (9.6%),
Livingston Parish was intermediate (21.0%), Marion County was highly
infected (50.6%), and the Florida seed orchard source was the most infected
(68.7%). Within slash pine sources survival, diameter and height were
greater in the seed orchard sources than in the rust-rogued seed production
area, but the latter had significantly less rust (27.4%) compared with
59.0% and 61.7% for the FL and GA orchards, respectively. For loblolly
sites in this area recommendations are to plant East Texas sources on the
highest rust incidence sites, Livingston Parish sources on the intermediate
rust incidence sites and Marion County sources in eastern area where rust
incidence is low. For slash pine sites where rust incidence is high, seed
from the rust-rogued seed production area is recommended.
Additional Keywords: Pinus elliottii var elliottii, P. taeda, Cronartium
guercuum f. sp. fusiforme, disease resistance, provenance tests.
y
Research Foresters, Container Corporation of America, Timber Research
and Development, Callahan, Florida, 32011, and Professor of Forest
Pathology, Department of Forestry, University of Florida, Gainesville,
32611, respectively.
INTRODUCTION
In areas where the incidence of fusiform rust, caused by Cronartium
uercuum f. sp. fusiforme, is great forest managers must consider
alternatives to reduce the impact of this disease (Schmidt et al. 1977,
314
Schmidt and Klapproth 1982, Anderson et al. 1984). Currently, the primary
means of mitigating the epidemic in high rust incidence areas is planting
rust-resistant seed sources (Schmidt et al. 1985). Several sources of
resistance are available. These include Livingston Parish and East Texas
provenances of loblolly pine and rust-rogued seed production areas for
slash pine (Goddard and Wells 1977, Schmidt et al. 1981). Recommendations
for a specific area must consider both growth and rust response. Decis-
ions are best made with data obtained from tests established in the
immediate area, since both growth and disease resistance (perhaps patho-
genic variability) vary geographically (Draper 1975, Powers and Matthews
1980, Pait and Draper 1983). The objectives of this study were to compare
the growth and rust resistance of several loblolly and slash pine seed
sources in a high rust incidence area in North Central FL for the purpose
of providing management recommendations.
METHODS AND MATERIALS
Location and Site Characterization. The seed source test is located
in North Central FL in Marion County, approximately ten miles northwest of
Ocala. The soil is a moderately well-drained, fine sandy loam overlying
clay at a 24 inch depth. Site index is 75 ft at 25 years for loblolly and
slash. Fusiform rust is considered a serious problem in this location,
e.g., rust incidence on a seven-year-old loblolly plantation in this area
exceeded 75% of the trees infected.
Seed Sources. Four sources of loblolly and three sources of slash
pine seed were planted. These sources represented the best regeneration
alternatives at the time and all were bulk seed collections. The loblolly
pine sources were from 1) East Texas 2) Livingston Parish, and 3) Marion
County, FL provenances, and 4) Container Corporation of America's (CCA)
loblolly seed orchard selections. The slash pine sources came from CCA's
1) GA slash pine seed orchard, 2) FL slash pine orchard, and 3) a rust-
rogued seed production area in Madison County, FL. Seed orchard sources of
both species were generally unimproved for rust resistance. The East Texas
and Livingston Parish provenances were known to contain appreciable rust
resistance (Wells and Wakely 1966, Wells and Switzer 1975) and the Marion
County provenance has shown good growth and, on occasion, some rust resis-
tance (Draper, 1975). The rust-rogued slash pine seed production area was
located in a high rust incidence area in Madison County, FL and was
expected to have considerable rust resistance (Goddard et al. 1975), but
was not previously tested in an appropriate trial.
Site Preparation and Planting: Sites were prepared by pushing debris
from the area of the plots, followed by single drum chopping and burning.
In January 1981 seedlings were hand-lifted from the nursery at Archer, FL
and dibble-planted at a spacing of 5.5' x 12' (660 trees/acre).
Study Design. The study design was a randomized complete block.
Within species each seed source was randomly planted in each of nine
replications (plots). Each plot consisted of 130 seedlings (10 rows of 13
seedlings each.) Species were analyzed separately since they were not
mixed, and were separated by a fire break.
315
Maintenance and Data Collection. In May of 1982 hardwood sprouts,
which threatened survival and growth, were killed with cut-surface
application of Banvel CST. The strips between the rows were mowed in the
summer of 1982 and 83.
After the third growing season (October 1983) survival, DBH, total
tree height and rust incidence were measured. Rust incidence was recorded
on the total number of living trees and included trees with 1) stem galls
only, 2) branch galls only, and 3) both branch and stem galls. Rust
associated mortality averaged < 1% and was excluded from the analyses.
Within species data were analyzed with the general linear models procedure
for analyses of variance (Statistical Analysis System) and seed source
means were compared with Duncan's Multiple Range test (p < 0.05).
RESULTS
Survival, Height and DBH (Table 1).
Loblolly. The four loblolly pine seed sources averaged 77.3%
survival, 6.6 ft in height and 0.78 inches DBH. There were significant
differences among seed sources for each of these parameters. The Marion
County source performed best and the East Texas source performed the
poorest; the Livingston Parish and FL seed orchard sources were inter-
mediate.
Slash. The three sJash pine seed sources averaged 77.6% survival, 5.9
ft in height and 0.92 inches DBH. There were significant differences among
seed sources for each of these traits. Generally, the FL and GA seed
orchards sources performed the best and the seed production area source
performed the poorest for these growth traits.
Table 1. Comparison of mean survival, DBH and height at age 3 years for loblolly
and slash pine seed sources planted in Marion Co., Florida.
Survival DBH Height
Species Source ho inch ft
Loblolly Marion County, FL 93.4 a¢/ osemie Toe
FL Seed orchard 37). i 0.78 b 6.8
Livingston Parish 19152 2 0.78 6 6.5 d
East Texas 72.9 0.69 ate)
Slash FL Seed orchard 74.6 : 0.97 3 6.2 3
GA Seed orchard / 80.9 b 0.93 rs out b
Seed production area — 74.6 0.85 Bao
au Rust-rogued and located in a high rust incidence area in
2/ Madison County, FL
=’ Statistical comparisons are within a parameter (column)
among seed sources within species; means followed by differ-
ent letters are significantly different (Duncan's Multiple
Range Test,(p < 0.05)
316
°(SO° > d) 3808, eBuey
SLTAGTAIINW s,ueDUNO of ButPuosa5e SH8aA38T YeTtwts Aq petuedwosse
32> AUBISAITP ATRUEDTLTYUBTS 30u sue setseds UTURTM PUB saduNnoS
pess BSBuowe sedAR uoTtRAD.AaJZUT PUe (syuRzBuUET 489) votz.,eyut asnu TeI0Ly
(%X) ASOH UATM seauy
OZ 03 oS OV O€ 02
yp — eT6bu0a9g
PyYeudsUON peas
| Pueusug peas
* Lilla nae
e
* Li ddlddddilidddddddldlddddd vere
)
YY XX ustued
4 WITT uo ASButAtT
q 5
Q e
wa3s pue quty Wy
eS
Y e Yy % sexal
Y ' a3se3
ATuo waas ex Lied 7
ATuO gutq [=a] ' ATTOTQO74
edAL votzoaJ,ur
Sos Seve a7 alisha). Wiese
Ut PeRAueTA sseounos pass Bsutd uoOs Bau sabe 72e
Seppe S saSmcd UStMeaSoass OAn Ee eho “spe wwe oace Ce Sik, sete) eles
. '
4 Pp - 8Ptvoty
317
Fusiform Rust (Figure 1).
Loblolly. The four loblolly pine seed sources averaged 1.6% stem
galls only, 10.8% limb galls only, 25.1% both limb and stem galls, and
37.5% total rust infected live trees. Average rust associated mortality
among these sources was s 1%. The mean total rust incidence on live trees
was 9.6, 21.0, 50.6 and 68.7% in the East Texas, Livingston Parish, Marion
County and FL seed orchard sources, respectively. Rust incidence was
significantly different for each loblolly source.
Slash. The three slash pine seed sources averaged 3.2% stem galls
only, 17.1% limb gall only, 29.1% both limb and stem gatls and 49.4% total
live trees with rust. The mean total rust incidence on living trees was
27.4, 59.0 and 61.7% in the seed production area and FL and GA seed
orchards sources, respectively. The seed production area source exhibited
significantly less rust than the seed orchard sources.
DISCUSSION
Average survival (loblolly, 72.4-83.4%; slash, 74.6-80.4%) and average
rust incidence (loblolly, 9.5-68.7%; slash 27.4-61.7%) were sufficient for
a reliable test of these seed sources. Growth rankings among sources may
change with time and these data must be considered preliminary. Rust
incidence will increase with time, but it is unlikely that relative rank-
ings among seed sources will change.
As suggested in earlier tests (Draper 1975, Pait and Draper 1983) the
Marion County source has superior growth in this area, as well as in some
northern areas. The East Texas and Livingston Parish sources grew more
slowly corroborating age five results published previously (Pait and Draper
1983). The slash pine seed orchard sources grew significantly better than
the seed production area source. Among the seed sources, survival, height
and DBH variation was greater in loblolly than in slash pine.
Rust incidence on the East Texas source was significantly less than
all other loblolly sources. Similar results were reported by Pait and
Draper (1983) for this and other areas in FL and GA. Although a statis-
tical comparison was not appropriate - due to the experimental design - the
East Texas source was less infected than all slash pine sources.
Livingston Parish exhibited good rust resistance in this area, but other
data (Pait and Draper 1983) suggests this source is very susceptible when
planted in Madison County, FL, 100 miles northwest of Marion County, FL.
The reason for the poor performance of Livingston Parish in the earlier
study is not known.
As was reported by Goddard et al. (1975) the seed from heavily
infected rust-rogued stands possess substantial rust resistance. This was
substantiated here as the seed production area source was significantly
less infected than the other slash pine sources. In fact, the rust-rogued
seed production area source performed nearly as well for rust resistance as
the Livingston Parish source. In the absence of rust improved orchard seed
and resistant provenance sources, seed from rust-rogued slash pine seed
318
production areas provides a good alternative for planting in areas of high
or intermediate rust incidence.
Despite a report (Schmidt et al. 1985) that rust incidence is higher
on loblolly than on slash pine in this geographic area, susceptible
loblolly (68.7%) was only slightly more infected than was susceptible slash
pine (60.4%).
CONCLUSIONS
Among the loblolly sources East Texas had significantly less fusiform
rust, but also had the poorest survival and growth. The Livingston Parish
source was intermediate in rust incidence and growth. The Marion County
source grew best, but had significantly more rust than either the East
Texas or the Livingston Parish source. The seed orchard source had signif-
icantly more rust than all other sources and was intermediate in survival
and growth.
Among the slash pine sources the rust-rogued seed production area had
significantly less rust than the seed orchard sources. In fact the resis-
tance of the rust-rogued seed production area source compared favorably
with the Livingston Parish source, although a statistical comparison was
not appropriate because of the experimental design.
Indications from these early observations, combined with information
from previous tests suggest the following seed source allocation.On
loblolly sites East Texas and Livingston Parish sources should be utilized
on the high and intermediate rust incidence areas, respectively. The
Marion County source should be restricted to the eastern portion of this
area where rust incidence has been low. On slash pine sites the seed
production area source should be planted in the high and intermediate rust
incidence areas and the seed orchard sources elsewhere on the low rust
incidence areas only.
LITERATURE CITED
Anderson, R. L., Schmidt, R. A., and Snow, G. A. 1984. Integrated pest
management in regeneration - early growth phase of pine stands:
Diseases. p. 54-71 in Blanham, S. J. and Hertel, G. D. eds. Proc.
Integrat. For. Pest Manage. Symp., Univ. GA, Athens. 281 p.
Draper, L., Jr. 1975. Provenance study of five geographic sources of
loblolly pine. p. 83-88 in: Proc. 13th South. For. Tree Improv.
Cone F262 pi.
Goddard, R. E., Schmidt, R. A., and Vande Linde, F. 1975. Effect of
differential selection pressure on fusiform rust resistance in
phenotypic selections of slash pine. Phytopathology 65:336-338.
Goddard, R. E., and Wells, 0. O. 1977. Susceptibility of southern pines
to fusiform rust. p. 52-58 in R. J. Dinus and R. A. Schmidt, eds.
319
Management of fusiform rust in southern pines. Symp. Proc. Univ.
Fla., Gainesville. 163 p.
Pait, J. A., III and Draper, L., Jr. 1983. Fifth year performance of wide
ranging loblolly pine provenances. p. 245-252 in Proc. 17th South.
For. Tree Improv. Conf. 375 p.
Powers, H. F., uUr., and Matthews, F. R. 1980. Comparison of six
geographic sources of loblolly pine for fusiform rust resistance.
Phytopathology 70:1141-1142.
Schmidt, R. A., Cowling, E. B. and Dinus, R. J. 1977. Recommendations for
a regional rust control strategy. p. 10-16 in R. J. Dinus and R. A.
Schmidt, eds. Management of fusiform rust in southern pines. Symp.
Proc. Univ. Fla., Gainesville. 163 p.
Schmidt, R. A., Holley, R. C. and Klapproth, M. C. 1985. Results from
operational plantings of fusiform rust resistant slash and loblolly
pines in high rust incidence areas in Florida and Georgia. p 33-41 in
J. Barrows-Broaddus and H. R. Powers, eds. Proc. Internl. Union For.
Res. Work Group. Rust of hard pines. Athens, GA. 331 p.
Schmidt, R. A. and Klapproth, M. C. 1982. Delineation of fusiform rust
hazard based on estimated volume loss as a guide to rust management
decisions in slash pine plantations. South. J. Appl. For. 6:59-63.
Schmidt aReAs., Powers, H. R., Jr., and Snow, G. A. 1981. Application of
genetic disease resistance for the control of fusiform rust in
intensively managed southern pine. Phytopathology 71:993-997.
Wells, 0. O. and Switzer, G. L. 1975. Selecting populations of loblolly
pine for rust resistance and fast growth. p. 37-44 in Proc. 13th
South. For. Tree Improv. Conf. 262 p.
Wells, 0., O0., and P. C. Wakely 1966. Geographic variation in survival,
growth, and fusiform rust infection of planted loblolly pine. For.
Sci. Mono. 11, 40 p.
320
COMPARATIVE PHYSIOLOGY OF LOBLOLLY PINE SEEDLINGS FROM SEVEN
GEOGRAPHIC SOURCES AS RELATED TO GROWTH RATE
Bruce C. Bongarten, Robert O. Teskey and Brett A. Boltz*
Abstract.--Growth, photosynthesis and water relations charac-
teristics were examined in loblolly pine seedlings from seven
diverse geographic sources. At the end of the first year, Florida
trees were largest in height and dry weight, while Texas and
Arkansas trees were smallest. Seed source size rankings were
established by the fifteenth week of growth and were correlated
with seed weights and both earliness and completeness of bud-set.
They were also correlated with net photosynthesis at each of
eleven dates during the growing season. This appears to result
primarily from differences in leaf area accretion. When photo-
synthesis was measured on a unit leaf area basis, differences
among the provenances were absent, except late in the year when
Florida trees were most active.
Few differences in water relations characteristics were shown
among the provenances. No differences in osmotic potential at
saturation or turgor loss were detected. The degree of osmotic
adjustment appeared to be equal, as well. Continental seedlings
(Texas, Arkansas and Georgia Piedmont) exhibited greater stomatal
conductance than seedlings from coastal origins when drought
stress was never imposed, however, in trees pretreated with
drought, no provenance differences were observed.
Differences in first year growth rate appear to be due, in
part, to differences in seed weight, leaf production, and late
season growth and photosynthesis. The measured water relations
traits do not appear to be important although other water rela-
tions traits may be.
Additional keywords: Genetic differentiation, provenance testing,
photosynthesis, water relations, pressure-volume curves, Pinus
taeda.
INTRODUCTION
Thirty years of loblolly pine provenance testing have clearly demon-
strated the presence of geographic differentiation for growth rate. Prin-
cipally, trees from south coastal areas, with mild winters and heavy summer
rainfall, are faster growing than trees from interior or north coastal regions
LU Rescelerec ent: Professor, Assistant Professor and Graduate Research
Assistant, respectively, School of Forest Resources, University of Georgia,
Athens, GA 30602.
321
(Wells, 1983; Wells and Wakeley, 1966). Differences are sufficiently great
that trees from non-local provenances are often planted to improve productiv-
ity. This is particularly true in Arkansas where loblolly pines from coastal
North Carolina are now widely planted with projected volume increases of
20-30% over the local stock (Lambeth, et al., 1984).
In this work, we examined some of the possible physiological causes for
provenance differentiation in growth rate in the seedling phase. The growth
superiority of trees from regions with short, mild winters and wet summers
suggests that provenance differences may be due to differences in the duration
of growth activities, including photosynthesis, and/or differences in water
relations traits. We have, therefore, emphasized these in our initial work.
The information obtained may be useful in predicting responses of trees to
different environments, thus providing a means for assessing the risks in-
curred upon planting trees of foreign provenance. Additionally, knowledge of
the physiological basis for growth rate differences may be used to design
inter-provenance breeding plans which maximize growth by combining complemen-
tary components. As our understanding of the physiological basis for growth
increases, it may also be possible to select more effectively for rotation-age
volume in the juvenile stage.
MATERIALS AND METHODS
Seedlots
Seeds were obtained from seven first generation seed orchards represent-—
ing different portions of the loblolly pine range (Figure 1). The North
Carolina, South Carolina, Florida and Louisiana sources are considered
"Coastal", while the Georgia, Arkansas/Oklahoma and Texas sources are con-
sidered "interior" or "continental". The Georgia Piedmont and Coastal
Louisiana seed orchards had been rogued, while the others had not. The Texas
seed orchard was composed of ramets from ortets selected in nursery beds for
drought resistance by the Texas Forest Service; most were originally from
Bastrop and Lee Counties, Texas. Ortets for the other orchards were selected
primarily for phenotypic superiority in volume, and crown and form factors.
Growth and Photosynthesis
In May, 1984, seeds from six seedlots (all except Georgia Piedmont) were
sown in DEEPOT containers (646cm*) in a greenhouse using a randomized complete
block design with four replicates. Each replicate contained one plot of 20
trees per seedlot. Seedlings were kept well-watered and fertilized throughout
the study period.
Beginning 12 weeks after sowing, and recurring every 10 to 14 days
through mid-December, seedling height, root collar diameter and net photo-
synthesis were measured on the interior six trees from each plot. The number
of seedlings with buds was also recorded on each date. In all, measurements
were made on eleven dates. Seed weights, based on 500 seeds, were obtained
before sowing.
322
Wr eeee CUT LAL Hat \ aie
| | \ ty | | | I | hi i i a Mey [Pores {
MH : Srnnieg
he il
a
sO
OY
So.
Te
eR
(x
KOSS ae
rN sgresey
1
=A
tq
sa
ins
&$
S
cae
LY
ws
e, MEY
Sy
04
Sy
4
Ls
ues
rr
RRR SINS
Reta neeann
Nea
%
c)
a,
Re
Re
Egiecad
a
aN ee
SS RS : ~
cs apleee 3 3
+t <
Ba 4
oe 832
ee \
~~ re Eee
| - oy ° 100 200 300 400 MILES wala
|_| y/ (ere eS Sn Se Ln een) tr
We eas (LI eS Me a lat
A\ © 100 200 300 400 $00 600 KILOMETERS
mo
Figure 1.--Locations of the seedlots used in this study. Shaded area repre-
sents the natural range of loblolly pine.
Net photosynthetic rates were obtained under steady-state conditions.
Trees were placed in a cuvette in which the temperature was maintained at
30°C, relative humidity at 50%, light at 1500 ymolm’™ s (middle of the
cuvette; equivalent to full sun light) and CO, concentration at 340 ppm.
Carbon dioxide (1%) was pumped continuously from a compressed gas cylinder
into the chamber to maintain the CO, level at 340 ppm. Steady-state equilib-
rium was achieved when a constant f tow of CO, maintained the prescribed
cuvette CO, concentration. Because the volume of CO, pumped into the chamber
displaced an equal volume of air having the ambient CO, concentration, the
rate of CO, fixation was calculated as the product of the CO, flow rate and
the difference in cO., concentration between the pumped and displaced gases
(Griffiths and Jarvis, 1981). Approximately one hour was required to achieve
steady-state equilibrium for each sample of six trees from a seedlot-replicate
combination. For each measurement period 24 observations (six seedlots x four
replications) were made over two days.
The cuvette, constructed of clear Lexan, measured 47 x 21 x 41 cm in
length, width and height, respectively. Only the shoot portion of the tree
$23
was enclosed within the chamber. Light was provided from above by a sodium
vapor lamp, and from the sides by two incandescent lamps. Temperature was
controlled with two copper, radiator type heat exchange units controlled by
Neslab thermostats (Portsmouth, NH). Relative humidity was regulated by a
Honeywell dehumidifier control (H46C 1000) which directed chamber air through
a desiccator column as required.
Estimates of leaf areas on each of the sample dates were obtained from
regressions of height on leaf area developed at the end of the study. In this
way photosynthesis per unit area could be estimated without destructive
sampling. Such regressions are subject to errors because of leaf and inter-
node growth subsequent to each measurement. However, these errors may be
roughly compensating and of minor importance, particularly near the end of the
study, when growth was terminating.
Analysis of variance was used to test for seedlot differences on each
measurement date. Differences between dates within seedlots were analyzed by
paired t-tests.
Water Relations
Water relations measurements were also taken on trees grown in DEEPOTS in
a greenhouse. Eight blocks} each containing one plot of 20 trees from each of
six seedlots (coastal South Carolina excluded), were used. Trees in four of
the blocks were kept near field moisture capacity by watering on alternate
days (high moisture regime). Trees in the other four blocks were subjected to
recurring drought cycles; rewatering occurred only when flaccid shoot tips
were observed in early morning (low moisture regime). Before the first
measurements were taken, trees in the low regime had experienced eight drying
cycles and were approximately one-third the size of the trees in the high
moisture regime.
Stomatal conductances were determined for each seedlot, in both high and
low moisture regimes, under three different humidity conditions. Steady-state
methodology was employed. A cuvette measuring 15 x 20 x 15 cm was constructed
of glass, Lexan and propafilm-c. Individual seedlings were placed in the
cuvette, and dry air was introduced, displacing the moist chamber air. When a
constant flow of dry air maintained the chamber air at a prescribed relative
humidity, steady-state equilibrium was achieved. Transpiration and stomatal
conductance were calculated from the water lost in the displaced air.
The prevailing envirgnmental conditions included a temperature of 30°C,
and light at 1500 umol m= s . Measurements were made at relative humidities
of 77%, 54% and 31%, which corresponded to absolute humidity deficits of 7, 14
and 21 g/cm?, respectively. For each seedlot seven to nine seedlings were
sampled, individually, in each the high and low moisture regimes. Differences
between seedlots, moisture regimes and humidity treatments were assessed by
analysis of variance.
Pressure-volume curves were also developed to examine differences in
turgor loss points, osmotic potentials at saturation, and osmotic adjustment
324
(Tyree and Hammel, 1972). Seedlings were cut near the root-collar and placed
in water in a dark closet overnight to insure full saturation. Xylem water
potentials, measured with a Scholander pressure bomb, and relative water
contents (the proportion of total shoot water content) were then periodically
recorded as the seedlings were allowed to dry. The curves were plotted as the
inverse of xylem water potential versus relative water deficit (RWD), the
proportion of water lost (Figure 2). Turgor loss points were estimated
visually for each seedling. Osmotic potentials at saturation were estimated
Zot A SEEDLOT 835, WEST TEXAS
HIGH MOISTURE REGIME
— ¥"' (MPa)
UL
UT] PTR LT LE SL TL I
8.8 Q.1 @.2 8.3 8.4 8.5 8.6
RELATIVE WATER DEFICIT
Figure 2.--A sample pressure-volume curve indicating the osmotic potential at
saturation (W~,), and at turgor loss (W_,), and the relative water deficit at
turgor loss (RWD D: This plot was constructed from the combined data of six
sampled trees.
for each seedlot-moisture regime combination as the intercepts of linear
regression lines based on all points beyond turgor loss (Figure 2). Osmotic
adjustment was determined as the difference in osmotic potentials between
trees in the high and low moisture regimes. Differences between seedlots were
tested with analyses of variance.
325
RESULTS
Growth
Of the seedlings monitored for photosynthesis, those from each of the
Coastal Plain origins grew faster than those from Arkansas or the Lost Pines
region of Texas. Seedlings from Florida grew fastest, achieving 50% greater
height and dry weight than seedlings from Arkansas, which, on average, were
smallest (Figure 3).
48
35
5
~ 30
te
ag
©
lJ
= 5
FA 25
es
20a
is) : - :
Vin Uin Cin @, UU UT pred LI PL) Tal Le ere SE SUS Ly PUSS TESA ELS RY PE INI RT FI
@2AUG 22AUG titSEP. @10CT 210CT 1 @NOV 3ONOV
DATE
LEGEND: SOURCE +—+—+ Arkansas 8-6-6 Florida
*#—+— 4 Louisiana ##A-#& North Carolina
2—e—e South Carolina e-©—-¢ Texas
Figure 3.--Height growth in the first year for loblolly pines from six prove-
nances.
Provenance differentiation in height was apparent by the first measure-
ment, 12 weeks from sowing. At that time, seedlings from Texas and Arkansas
were already significantly shorter than seedlings from Coastal sources. This
early size difference may be related to seed weight as seeds from Texas and
Arkansas averaged 2.7 mg, compared to 3.4 mg for seed from Coastal sources.
Significant differences in height at 12 weeks were also detected among the
326
Coastal provenances (Louisiana and Florida seedlings being taller than North
Carolina seedlings), but they were not related to seed weight. The early
difference between Coastal and Continental seedlings in height growth was
accentuated by differences in bud-set later in the study. Compared with
Coastal trees, those from Texas and Arkansas began bud-set earlier, and more
of them had terminal buds at the end of the study (Table 1). Among the
Coastal seedlings, those from Florida showed significantly less bud-set than
the others. Overall, provenance differences in bud-set closely parallelled
final tree heights and dry weights, although provenance differences in seed-
ling height were readily apparent well before the first seedlings set buds.
At the end of the study, provenance differences in height more or less
mirrored differences in root collar diameter, total dry weight and dry weights
of leaves, stems and roots (Table 1). By contrast, Florida and Texas seed-
lings, which represented the extremes in size, had the greatest shoot-root
ratios. However, shoot-root ratios vary with seedling size (Ledig and Perry,
Table 1.--Growth measurements at the end of the first growing season for
lobllolly pines from six provenances.
Source Height Dry weight Shoot-root Bud-set
cm g % of seedlings
SC ee see On525 45.8
LA 31.8. 3.5. 0.49. 50.0
FL 35.94 4.2. 0.64, L2Z5S
AR 24.7). 2.875 0.52, 87.5
NC 30.604 3.6), 0.46, 41.7
TX Diier2 Sil 0.46 58.3
“Means not having a superscript in common differ at the 5% level of
significance.
1965) and time of bud-set (Cannell and Willett, 1976). Therefore, the
shoot-root ratios calculated here may not be indicative of the real differ-
ences in the relative growth of shoots and roots. This essential information
may only be obtained by sampling shoot and root weights throughout the study
period.
Photosynthesis
The seasonal pattern of seedling net photosynthesis was similar for each
of the six examined seedlots. From August 15 photosynthesis rose slowly until
October 1, then rapidly to a maximum in late October. Thenceforth, net
photosynthesis fell until levelling off in November (Figure 4). The rise in
327
1580
h
= 1250
fo
E
a
S 1800
=I
(a)
js}
lJ
2)
a
in 788
uJ
aC
re
ee
>
o
e 580
[o)
at
oa
fi
a 258 Ty fae Mar I TL RS SS
12AUG @1SEP 21SEP . 110CT 310CT 2EGNOV 1@DEC
DATE
: +—+— Ark 8-8-8 Florida
naa a a a eee eee A&A A-* North Carolina
>—e—e South Carolina eee Texas
Figure 4.--Seasonal course of net photosynthesis per seedling. Each point
represents a mean of 24 trees.
net photosynthesis through October resulted primarily from accretion of leaf
area. The increases were modest through early October as respiration from the
production of new needles largely offset the increased photosynthetic capaci-
ty. However, when the rate of new leaf production declined, net photosynthe-
sis rose dramatically. The decline in net photosynthesis after October was
apparently related to internal physiological changes, and is common in temper-
ature zone trees entering the winter season (Ledig, 1976).
When leaf area effects were removed by considering net photosynthesis on
a unit leaf area basis, maxima were observed in August and late October, while
minima occurred in early October and after mid-November (Figure 5). The
decline in net photosynthesis per leaf unit area from August to early October
probably resulted from increased respiration due to leaf production. The
increase in photosynthesis thereafter until late October reflected a reduction
in leaf production. Finally, the decline after late October indicated a
transition in physiological activity associated with the onset of winter, as
mentioned previously.
328
h
t 3.8
E
Te)
E
=
255
te
a
Oo
fa
E
€ 2.8
wn
oO
=
(e)
ae
a
bi 1.54 :
me CUMS) PIG IIE USAIN TER SR IR ITY FUT IT RSE STL ER DR SS FL EER SS ISR SRS A on TA Fa) DT JSS FR ISR Foe ETF) Pe UGK Forse Tr Bk Soa Foy
12AUG @1iSEP 21SEP 110CT 310CT 20NOV 1@DEC
DATE
LEGEND: SOURCE “ +—+—++ Arkansas 5 8-8-6 Florida
; +*#—* 4 Louisiana #&-#-*& North Carolina
@—*@—-@ South Carolina e-<-¢ Texas
Figure 5.--Seasonal course of net photosynthesis per unit leaf area. Each
point represents a mean of 24 seedlings.
Significant provenance differences (0.1 level) in net photosynthesis per
seedling were found on only five of the eleven measurement dates. However,
provenance rankings were consistent throughout the study leading to large
differences in total CO, assimilation (the areas under the curves in Figure
4). The rankings of the seedlots in total net assimilation closely corre-
sponded to their rankings in mean seedling size, seedlings from Florida having
the largest values and those from Texas and Arkansas the smallest. This
correspondence is expected because of the dominating influence of leaf area on
total net photosynthesis.
When calculated on a unit leaf area basis, differences among the prove-
nances in net photosynthesis were detected only on one date near the end of
the study, and the seedlot rankings, in general, did not correspond well with
those for mean seedling size. The late-season net photosynthetic rates were
an exception. Florida seedlings, which had the greatest late season growth,
also showed a significantly slower decline in net photosynthetic rate. Thus,
329
it appears that differences in leaf area accretion are primarily responsible
for differences in net assimilation and growth among the seedlots.
Differences in photosynthetic rate appear to be relatively unimportant except
that the longer period of high photosynthetic rate found in Florida seedlings
may account for their greater late-season growth.
Stomatal Conductance
Stomatal conductances at three absolute humidity deficits are shown in
Figure 6 for droughted and non-droughted seedlings. Trees which had never
STOMATAL CONDUCTANCE (cm gec™’)
e
a
8 tS) 18 11 12 Ss 14 1S 16 UY 18 TOE eZo 2
ABSOLUTE HUMIDITY DEFICIT (g m~)
LEGEND: CLASS a—&A—-*& Hi gh—-Coastal 4--4-% High-Interior
*—e—*® Low-—Coastal @-*-- Low-Interior
Figure 6.--Stomatal conductances, at three absolute humidity deficits, for
Coastal and Interior loblolly pines seedlings grown under moist and droughty
regimes.
been subjected to drought had conductances which were more than twice as great
as those which had. In the trees which had never been subjected to drought,
stomatal conductance declined markedly as the absolute humidity deficit
increased from 7 to 14 g/m (equivalent to a change from 77% to 54% relative
humidity at 30°C), while the droughted trees showed little stomatal response
330
to changing absolute humidity deficit. Curiously, among the trees which had
never been droughted, those from the three interior origins (Texas, Arkansas,
and Georgia) had consistently greater stomatal conductance. After drought
pretreatment, though, no differences in stomatal conductance were observed
among the seedlots.
Pressure-volume Curves
Four water relations parameters were estimated from pressure-volume
curves: osmotic potential at saturation (\,), osmotic potential at turgor
loss ()_,), relative water deficit at turgor loss (RWD_,), and the change in
osmotic potential with declining water content (slope). These parameters are
shown in Table 2. No seedlot differences were detected for any of the
Table 2.--Pressure-volume curve parameters for seedlots and moisture regimes.
Water Regime Source Water potential (MPa) RWD Slope
: el!
Saturation Turgor loss
HIGH LA -1.03 -1.47 0.29 -0.099
FL -1.03 -1.41 0.26 -0.102
AR -1.02 -1.48 0.29 -0.098
NC -1.23 -1.53 0.29 -0.081
TX -1.00 -1.42 0.28 -0.107
GA - -1.01 -1.44 0.28 -0.106
MEAN -1.05 -1.46 0.28 -0.099
LOW LA -1.00 -1.58 0.34 -0.108
FL -0.98 -1.57 0.34 -0.114
AR -0.96 -1.53 0.32 -0.123
NC -0.96 -1.61 05.35 -0.114
TX -0.99 -1.55 O63 -0.116
GA -0.88 -1.59 0.36 -0.142
MEAN -0.96 -1.57 0.35 -0.120
parameters. However, trees in the high moisture regime differed significantly
from those in the low moisture regime in each of the four parameters.
Droughted trees had greater initial osmotic potentials, but lower osmotic
potentials at turgor loss than non-droughted trees. Furthermore, the drought-
ed trees reached turgor loss at lower water contents than the non-droughted
trees. Such differences between droughted and non-droughted trees are common
and are termed "osmotic adjustment.'' The seedlots could not be shown to
differ in their degree of osmotic adjustment, either.
331
DISCUSSION
When undamaged by winter cold or ice or biotic agents, loblolly pines of
Coastal origin grow faster than those of Continental origin over a wide range
of sites. Similarly, southern loblolly pines outgrow northern ones (Wells and
Wakely, 1966; Wells, 1983). In our work we have sought to determine the
causes for these well documented trends in order to better understand the
risks incumbent with seed transfer and to assess opportunities for improving
growth rate with inter-provenance hybrids.
In the investigations considered here, the seedlings conformed to the
expected geographic differences in growth rate. The differences among seed-
lots in total growth were well correlated with differences in total net
photosynthesis; however, the correlation resulted primarily from differences
in leaf area and is of little value in selection or breeding. Differences
between Coastal and Interior seedlings in early growth rate may, however, be
due to the much smaller seed weights and earlier bud-set of the latter.
Furthermore, the superiority of the trees from Florida may result, in part,
from their longer period of high photosynthetic activity. Although differ-
ences in photosynthesis per unit leaf area or partitioning of photosynthate
are often suggested as causes for differences in growth rate, we could not
show significant differences among seedlots in either respect.
The growth measures presented here are based on high levels of moisture
and nutrients and cannot be considered to simulate field conditions. Differ-
ences among the seedlots in growth rate may also be due to factors affecting
growth during water or nutrient stress (Cannell et al., 1978). In our inves-
tigations we have also considered some of the factors that might affect growth
during periods of drought stress, including (1) changes in stomatal con-
ductance with increasing evaporative demand, (2) differences in osmotic
potential, and (3) differences in osmotic adjustment (the latter two may
result in differential stomatal closure with increasing water loss). Although
measured with great precision, no differences among seedlots for these charac-
teristics was detected. Other drought resistance characteristics which might
affect growth rate, such as rapid stomatal closure in response to water stress
and longer and deeper roots, have been shown to differ among seedlots from the
Western Gulf region (van Buijtenen et al., 1976), but these were not examined
in this work.
It is obvious that at this point in time our knowledge of the physiologi-
cal basis for growth rate differences is incomplete, and we are not in a
position to make recommendations applicable to field conditions. However,
this is a rather new area of research endeavor, and, if considered holisti-
cally, promises to offer faster, more effective methods of selection and
improved breeding strategies.
332
LITERATURE CITED
Cannell, M. G. R. and S. C. Willett. 1976. Shoot growth phenology, dry
matter distribution and root:shoot ratios of provenances of Populus
trichocarpa, Picea sitchensis and Pinus contorta growing in Scotland.
Silvae Genet. 25:49-59.
Cannell, M. G. R., F. E. Bridgewater, and M. S. Greenwood. 1978. Seedling
growth rates, water stress responses and root-shoot relationships related
to eight-year volumes among families of Pinus taeda L. Silwvae Genet.
27:237-248.
Griffiths, J. H. and P. G. Jarvis. 1981. A null balance carbon dioxide and
water vapour porometer. J. Expt. Bot. 32:1157-1168.
Lambeth, C. C., P. M. Dougherty, W. T. Gladstone, R. B. McCullough and 0. 0.
Wells. 1984. Large-scale planting of North Carolina loblolly pine in
Arkansas and Oklahoma: a case of gain versus risk. J. For. 82:736-741.
Ledig, F. T. 1976. Physiological genetics, photosynthesis and growth models.
pp. 21-54. In Tree Physiology and Yield Improvement (Cannell, M. G. R.
and F. T. Last, ed.). Academic Press, London. 567p.
Ledig, F. T. and T. O. Perry. 1965. Physiological genetics of the root-shoot
ratio. Proce. Soc. Am. For. Meet., Detroit. pp. 39-43.
Tyree, M. T. and H. T. Hammel. 1972. The measurement of the turgor pressure
and the water relations of plants by the pressure-bomb technique. J.
Exp. Bot. 23::267-282.
van Buijtenen, J. P., M. V. Bilan and R. H. Zimmerman. 1976. Morphophysio-
logical characteristics related to drought resistance in Pinus taeda.
pp. 21-54. In Tree Physiology and Yield Improvement (Cannell, M. G. R.
and F, T. Last, ed.). Academic Press, London. 567p.
Wells, 0. 0. 1983. Southwide pine seed source study - loblolly pine at 25
years. South. J. Appl. For. 7:63-71.
Wells, O. O. and P. C. Wakeley. 1966. Geographic variation in survival,
growth, and fusiform-rust infection of planted loblolly pine. For. Sci.
Monogr. 11, 40p.
333
RESISTANCE TO THE DEVELOPMENT OF PITCH CANKER
IN OPEN-POLLINATED SLASH PINE FAMILIES
GeA.-Lowernts., Mole Zoerbre Jane andpeliabaiat aI
Abstract.--Open-pollinated slash pine (Pinus elliottii
Engelm. var. elliottii) families displayed a significant amount
of family variation in resistance to the development of pitch
canker (Fusarium moniliforme Sheld. var. subglutinans Wollenw.
and Reink). Fertilized slash pine families possessed a
significantly greater level of infection than nonfertilized
families. Percent infection ranged from 13 to 69 among
fertilized and from 6 to 39 percent in nonfertilized
families. Slash pine families originating from selections
indigenous to Florida were significantly more resistant than
families originating from Georgia.
INTRODUCTION
Pitch canker infection of slash pine plantations became a serious
forest management problem on Union Camp Corporation land in late 1975 and
early 1976 (Broerman, 1976). A survey of pitch canker incidence on company
land revealed that 40 to 90 percent of all trees were infected within slash
pine plantations in the Florida counties of Clay, Putnam, Flagler, and Volusia.
In these highly affected areas, the entire crown of a tree would be infected
in contrast to infection of the terminal and perhaps a single branch when
pitch canker was present at an endemic level. Losses due to mortality and
decreased growth were estimated to be in excess of 1.5 million dollars
(Broerman, 1976). In response to the high level of pitch canker infection
and the resultant growth loss, the company decided to: (1) document the
distribution of the disease and assess the intensity and rate of disease
development, (2) develop a management strategy to implement salvage cuttings
when necessary, (3) support basic research on the pathogen and means of
transmission and (4) screen for potential resistance among open-nollinated
slash pine families in the company's first (1.0) generation seed orchards.
This paper contains the results of a genetic test to determine the extent of
resistance to pitch canker infection among open-pollinated slash pine families.
MATERIALS AND METHODS
An adequate level of inoculum must be present to screen slash pine
families for pitch canker resistance. Therefore, a test site was located
1/
— Geneticist and Senior Project Leader, resnectively, Union Camp Woodlands
Research, Savannah, Georgia, and Research Forester, Container Corporation of
America, Callahan, Florida.
334
1/
within a high disease incidence area on Union Camp land — in Volusia
County, Florida. The slash pine plantation present prior to test
establishment was harvested prematurely due to extensive pitch canker
induced damage ard loss. The slash pine plantations surrounding the test
site were also heavily infected. Ninety-two families were available for
pitch canker resistance screening from the Union Camp Corporation's first
generation seed orchards.
In January 1977, the 92 families were planted in two blocks. Due to
space restrictions, Block I and Block II did not receive the same number of
families. Block I and Block II were randomly assigned 56 and 36 families,
respectively. Four commercial checks were the only "families" common to
each block. Each block contains 20 replications with each family planted
in five tree row plots. At the time of planting, Block II received 250
lbs./acre of an 18-40-0 fertilizer applied to the planting beds and each
tree in both blocks received 7 grams of Furadan 10G.
In summer 1984, height, diameter and pitch canker incidence were
measured on each tree. Height was measured to the nearest foot and
diameter to the nearest tenth inch. Trees were scored as either being
infected with pitch canker or not infected. The magnitude of infection
in each tree was not assessed.
Since the commercial checks were the only entities common to each
block, a paired t-test was used to compare block means for height, diameter
and percent infected trees. Replication, family and family by replication
effects were analyzed separately for each block using analysis of variance
procedures. The family by replication interaction was not significant in
the nonfertilized block, but was significant in the fertilized block. This
interaction involved a minor family rank order change of no biological
Significance. Within each block, the family by replication effect was then
pooled with experimental error. Based on family means, height, diameter
and percent infected trees were analyzed with families and replications as
the sources of variation.
The families tested in this study originated from selections in the
Atlantic Flatwoods and Upper Coastal Plain provinces of Georgia and from
Florida. Based on the county of origin of the select parent tree, each
family in both the fertilized and nonfertilized blocks was clustered into
one of four groups: (1) Upper Coastal Plain of Georgia; (2) Northern
Georgia Atlantic Flatwoods; (3) Southern Georgia Atlantic Flatwoods and
(4) Florida. Duncan's new multiple range test was used to compare mean
pitch canker infection, height and diameter among the four groups.
RESULTS AND DISCUSSION
Slash pine families in the fertilized block possessed a significantly
higher incidence of pitch canker infection and a significantly larger mean
height and diameter than those in the nonfertilized block (table 1).
ay The test site was subsequently purchased by Container Corporation of
America.
335
Table 1.--Mean tree height and diameter and the proportion of trees
infected with pitch canker in the fertilized and nonfertilized
blocks. 1/
Block Height Diameter Proportion of trees infected with
(EE) Ging) Pitch Canker
Fertilized 2023S 4.8 40.5
Nonfertilized DAES 4.0 20.6
*
tS1Oe 40Re e=2N96 t=8.87
1/
— where ns = not significant
* Significant at five percent level
** = significant at one percent level
Twenty-three percent of the trees in the fertilized block were infected with
pitch canker while only twelve percent of the trees were infected in the
nonfertilized block. A significantly greater incidence of infection also
occurred among the commercial checks in the fertilized block than in the
nonfertilized block (table 2). The commercial checks in the fertilized block
possessed a significantly larger mean height and diameter. Since the
commercial checks were the same in both blocks, the increased rate of
infection in the fertilized block was apparently due, in part, to fertilization
and is probably not a result of the random assignment of families to each
block. A prolonged growing season and an increase in the amount of succulent
tissue as a result of fertilization may have predisposed the slash pine to
Table 2.--Degrees of freedom, mean and F-value of height, diameter and
proportion of trees infected with pitch canker among the commercial
checks between the fertilized and nonfertilized blocks. iy
Source of DF Block Height (ft. ) Diameter(in.) % Pitch Canker
Variation X F X F X F
Block i BEAM 6 24.6 oh 4.5 es 45.6 PB :
Nonfert. DESO “O5Aks AOS Ee Seg S) AAO) et T0829)
GUC: 3 YO 0.68"° 1,390
*
C.C.* block 3 S.5 le ip 0. 4dam
1/
— where ns = not significant
* = significant at five percent level
** = Significant at one percent level
CiGu.= commercial check
336
pitch canker (Dwinell, et al., 1981). In a general way this possibility
is corroborated by other work indicating that an imbalance in plant
nutrition because of an over abundance or shortage of nutrients can lead to
greater levels of infection by a pathogen (Agrios, 1978). The incidence of
pitch canker infection has been shown to be associated with increased levels
Of tertulezation (Wilkinson, et al. 5 1977). Fertilization,’ applications of
pesticides and mechanical wounding may also be associated with the
occurrence of pitch canker in loblolly (Pinus taeda L.) and slash pine seed
orchards (Dwinell, et al., 1981).
In both the fertilized and nonfertilized block, a weak inverse
correlation existed between family mean height and diameter with family
percent pitch canker infection (table 3). Although family mean height was
Table 3.--Correlation coefficient (r) of family mean height and diameter
with the proportion of trees infected with pitch canker in the
fertilized and nonfertilized blocks. 1/
Block N Ht Diameter
*
Fertilized 40 -0.45 SOU Q6ue With proportion of trees
infected with pitch canker
ie ts ns : ;
Nonfertilized 60 -0.47 -0.246 With proportion of trees
infected with pitch canker
Vy if LOU ee
— where ns = not significant
ca Sonar cantwatce the one percent levied!
significantly correlated with percent pitch canker infection, the correlation
accounted for only 20 and 22 percent of the variation between height and
percent pitch canker infection in the fertilized and nonfertilized blocks,
respectively. The weak height and diameter correlation with percent pitch
canker infection suggests that the growth rate of the slash pine families in
this study probably did not directly influence the host-pathogen disease
complex to any great extent. Growth rate was not correlated with pitch
canker resistance in a slash pine screening study conducted by McRae, et al.
(1985). Arvanitis, et al. (1984) also found that diameter was not related
to pitch canker infection in nonfertilized slash pine plantations.
The proportion of trees infected with pitch canker varied significantly
among the slash pine families in both the fertilized and nonfertilized
blocks (table 4). In the fertilized block, pitch canker infection among
slash pine families ranged from 13.0 to 69.0 percent and from 6.0 to 39.0
percent in the nonfertilized block. Family variation in the fertilized and
nonfertilized blocks accounted for 35 and 32 percent, respectively, of the
total variation present in each block. Resistance to pitch canker infection
Table 4.--Degrees of freedom and F-value for height, diameter and
proportion of trees infected with pitch canker within the
fertilized and nonfertilized blocks. iy
SaaS Fertilized Block ---- --- Nonfertilized Block ----
Source of
Variatvone | Diy mee Ditaee yom lnkectr DR Jake Dia. ~% Infect.
k* *** ** **K* *x* **
Replication 19 10.66 65) ASS UW oi) 65S LOS GO,
k** *k* *K* kK* ns k*
Family BOs De yu 1.68 2.94 SO le 9 ho TO 3.49
RA SA nt
not significant
** = significant at one percent level
|
=
S
@
KR
©
=]
12)
Wl
also varies among clones in both slash and loblolly pine seed orchards
(Phelps and Chellman, 1976; Dwinell, et al., 1977; Dwinell and Barrows-Broaddus,
1981; Kuhlman, et al., 1982). The results of this study suggest that the
slash pine families in both the fertilized and nonfertilized blocks contain
varying levels ‘of resistance’ to pitch canker.
In), both the fertilized and) nontentes lazed block. slash pine: tamales
which were indigenous to Florida displayed a significantly lower level of
pitch canker infection than those families which were indigenous to Georgia
(table 5). There was no significant difference in the level of pitch canker
Table 5.--Comparison of mean height, diameter and proportion of trees
infected with pitch canker among slash pine clones originating
from the Upper Coastal Plain, northern and southern Atlantic
Flatwoods region of Georgia and from Florida. 1/
2/ - Nonfertilized Block - -- Fertilized Block ---
Region — Ht. Datay ni cpulemise Gitte Hite Dia. % Intect.
Ges) (in. ) Git) (Gans)
WGP aGar DEAS Blow 2 50a Oo 7a 20 ey Boy ADD eh
A Palka tyewNice Gale Zab As Oetalp 2S Oba ZOpeleae At Sara alls (0) ai
Ne VeUleKe 5 TSA Gate Bar 10) 3.9 b 20).8 a 2BOnO € 457 @ S755) Bio)
Florida 24.9 a Qala USF 110) BO oe! Be G7 ei 30.9 b
a Means within a trait and block not sharing the same superscript are
Significantly different at the five percent level.
2 ‘ :
ay URGee Gar = Upper Coastal Plain of Georgia
A. Flat. N. Ga. = Atlantic Flatwoods Northern Georgia
AS Eilat S).. «Gate Atlantic Flatwoods Southern Georgia
338
infection among Georgia slash pine families. However, slash pine families
which possess good height and diameter growth with pitch canker resistance
could be found in selections from Georgia and Florida. In Florida, natural
stands of slash pine possess significantly lower levels of pitch canker
infection than slash pine plantations; especially those plantations which
originated from selections in southern Georgia (Blakeslee and Rockwood, 1978;
Dwinell, et al., 1981). This study supports the views of several researchers
that slash pine trees indigenous to Florida are, in general, more resistant
to pitch canker.
The results of this study suggest that fertilization, directly or
indirectly, increases the susceptibility of slash pine families to pitch
canker infection. The role of fertilization and perhaps other environmental
factors (eg. drought) in the host-pathogen interaction is still unknown.
Elucidation of the effect of fertilization and environmental factors on the
predisposition of slash pine to pitch canker infection is necessary since
pitch canker remains a potential disease of epidemic proportions. The large
amount of variation displayed among the open pollinated slash pine families
suggests a tree improvement, program to enhance pitch canker resistance may
be possible. Additional testing is required in order to confirm the
repeatability of resistance since the slash pine families employed in this
study were only tested in one location. Even though the resistance of the
slash pine families was determined in one location, the most resistant
families should be preferentially planted in regions of high pitch canker
infection.
LITERATURE CITED
AgrloSwmGe Neen elon a ys Planta pachology; 95; p. N.Y... Academic Press.
ApRVanintedis ln GarpiJigimGOUDCe Ly Jima .uand 1. Porta, 1984. Pitch canker impact
on volume growth: A case study in slash pine plantations. South.
Jour. Appl. For. 8:43-47. :
Blakeslee, G.M., and D.L. Rockwood. 1978. Variation in resistance of slash
pine to pitch canker caused by Fusarium moniliforme var. subglutinans.
Phytopathology News 12:207-208.
Broerman, F.S. 1976. Pitch canker status report. Union Camp Corporation.
USSG
Dwinell, L.D., and J. Barrows-Broaddus. 1983. Pitch canker in seed orchards.
In Proc. 16th South. For. Tree Improv. Conf. p. 234-240.
PRR vanrandyE Gan KUniman | 197i. «Patch icanker Jot) Loblolly,
pine in seed orchards. In Proc. 14th South. For. Tree Improv. Conf.
Deel SO Sic
, E.G. Kuhlman, and G.M. Blakeslee. 1981. Pitch canker of
southern pines. In Fusarium: Diseases, Biology and Taxonomy. Eds.
P.E. Nelson, T.A. Toussoun and R.J. Cook. p. 188-194.
339
Kuhlman, E.G., S.D. Dianis, and T.K. Smith. 1982. Epidemiology of pitch
canker disease in a loblolly pine seed orchard in North Carolina.
Phytopathology 22:1212-1216.
McRae, C.H., D.L. Rockwood, and G.M. Blakeslee. 1985. Evaluation of slash
pine for resistance to pitch canker. Paper presented at the 18th
Southern Forest Tree Improvement Conference, May 21-23, 1985, Gulfport,
Mississippi.
Phelphs, W.R., and C.W. Chellman. 1976. Evaluation of "pitch canker" in
Florida slash pine plantations and seed orchards. U.S.D.A. For. Serv.,
State -G Priv. For.) Southeast Area. @22"p.
Wilkinson, R.C., E.M. Underhill, J.R. McGraw, W.L. Pritchett, and R.A.
Schmidt. 1977. Pitch canker incidence and fertilizer-insecticide
treatment) Inst. Food Agric, Sei) Unive Fila. Prope eRepe aii— sana
GROWTH MODEL EVALUATION: SOUTH-WIDE LOBLOLLY PINE
SEED SOURCE STUDY
Fan H. Kung!
ABSTRACT.--A growth model useful to the forest geneticists
should have the following five P's: (1) perfection of fit
to the data presented, (2) predictability of growth poten-
tials, (3) possibility of inference based on parameter
estimates, (4) power of discrimination among seed sources,
and (5) persistency of regression coefficients over time.
To illustrate, the function of 1n(HT) = BO + B1/AGE +
B2 In(1 + 1/AGE) was fitted to the height growth of 15
seedlots of loblolly pine in a south-wide study. The
degree of determination was 0.999 at least. The five-year
and the ten-year projections were low by 3% and 5% respec-
tively. The regression coefficients BO, Bl and B2 were
highly significant among seedlots. When the three growth
periods were compared, the coefficient of variation for the
regression coefficients was less than 52%.
Additional Key Words: Growth projection, Model verifica-
tion, Growth curve discrimination
INTRODUCTION
The mathematical characterization of growth is among the
oldest scientific pursuits. Indispensable long-term planning in
forestry requires reliable information about the growth of forest
stands. A wide array of growth and yield models ranging from
whole stand models to individual tree models has been developed
for southern species.
Genetic field tests are subjected to many uncontrolled
disturbances. However, height of dominant-codominant trees is
much less dependent on density and therefore is a better measure
of inherent growth differences. Growth and yield models can be
used to translate differences in dominant-codominant height into
stand differences expected in the absence of uncontrolled disturb-
ances (Nance and Wells 1981).
Foresters have long used the mean height of dominant-codomi-
nant trees for site index, as a universal measure of the potential
a Professor, Department of Forestry, Southern [Illinois
University, Carbondale, IL 62901. The author is grateful to Dr.
Osborn 0. Wells and the Southern Forest Experiment Station for
providing the data for this study, and to Dr. Marilyn A. Buford
for fruitful discussions.
341
productivity of forest land. Thus, a further improvement in
precision for measuring genetic differences in growth rate would
be removing the site effect and provenance by site interaction by
using the regional mean from many plantations.
Suppose that we have determined to use the range-wide mean
height to assess genetic differences, the question still remains
with us as to which growth model should we use as a base for
comparison.
Before growth models can be evaluated, one must outline the
criteria for model selection. We believe a useful growth model
should have the following characteristics:
06 perfection of fit to the data presented,
Die predictability on growth potentials,
BiG possibility of inference based on parameter estimates,
4. power of discrimination among seed sources, and
Die persistency of regression coefficients over times.
In this paper, we use data from the loblolly pine south-wide study
to illustrate such desirable properties of a growth function.
SOUTH-WIDE LOBLOLLY PINE SEED SOURCE STUDY
Complete details of the Loblolly pine experiment are given by
Wells and Wakeley (1966). Fifteen seed sources are represented
and 16 plantings survived after 25 years in the field.
Measurements of total height were made at various ages from 1
to 27 years. However instead of using only the dominant and
codominant trees, we used all available trees measured in 3, 5,
10, 15, 20, and 25 years to calculate the mean height growth for
each seed source. The reason for this selection is based on
statistical and not on silvicultural ground. The statistical
property of dominant and codominant trees is close to the extreme
number distribution while the average height is close to the
normal distribution. The latter is much easier for data analysis.
The south-wide regional means of the 15 seed sources are
listed in Table 1. The mean standard deviation and the coeffi-
cient of variation are also presented.
GROWTH FUNCTION
Nonlinear growth functions have been proposed for total
height. The monomolecular function is useful for site index
curves for age 20 and older which show no point of inflection
(Lundgren and Dolid 1970). Richards' curve gives a good fit for
both height and volume’ growth of Douglas-fir provenances
(Namkoong, Usanis, and Silen 1972). The Weibull function can
describe the growth of trees and stands (Yang, Kozak and Smith
1978). However, we are in favor of linear models over nonlinear
models because linear models give rise to unbiased, normally
distributed, minimum variance estimators, whereas nonlinear
342
Table 1.--South-wide regional means of height growth among
loblolly pine seed sources.
Seed Regional Mean at Age
Source 3 5 10 5 20 745)
----------- 0.01 ft. -----------
301 451 1089 2872 4174 5270 6092
303 455 1092 2907 4283 5388 6339
305 479 eS 3021 4440 5582 6410
307 391 948 2566 3907 5078 5907
309 481 1120 2866 4166 5331: 6305
SHISE 388 940 2599 3955 5146 5964
315 432 1043 2716 4085 byl Ls}s) 6007
319 457 1057 2790 4113 5243 6069
SZ) BY hi/ 903 2545 3898 4976 6028
323 433 1079 2897 4244 D1, 6162
325 438 1048 2793 4020 5006 5849
B27 412 992 2654 3782 4843 5681
329 393 934 2608. 3779 4831 5643
B85 346 905 2585 3915 4912 5949
Mean 421 1014 2729 4039 DEs2 6018
St. Dev. 40 82 159 197 220 222
GeViews Sod role § Syote) 4.9 4.3 37
regression models tend generally to do so only as the sample size
becomes very large (Ratkowsky 1983). With only a total of six
data points we are not very comfortable with the results of a
nonlinear model. Therefore, to illustrate that the five desirable
properties are obtainable, we use the following curve with an
intrinsically linear combination of parameters:
In (HT) = BO + B1(1/AGE) + B2 1n (1 + 1/AGE),
where ln is the natural logrithm transformation and BO, Bl, and B2
are coefficients to be determined by regression analysis.
The function provides precise description of observed data
points and provides trustworthy predictions. The function is
differentiable and is applicable to many growth and yield charac-
ters in life sciences (Kung 1984).
PERFECTION OF FIT
Growth is a continual process but it may be subdivided into
Stages. Tree physiologists indicated that growth consists of
division elongation, differentiation and maturation of cells
(Kozlowski 1971). Forest geneticists divided stand development
into juvenile-genotypic, mature-genotypic and codominance-suppres-
sion phase (Franklin 1979). Forest biometricians believe that a
growth curve begins at the value of zero, climbs slowly at first
and then more steeply. After a turning point, the increment
diminishes and then asymptotically moves towards some final value
343
(Prodan 1968). Thus a growth function should act like an adjust-
able ship curve used in drafting that fits all data points through
various stages equally well and not just for a single stage. For
example, a polynomial, as well as the simple exponential function
may fit a part of a growth series better than the more complex
nonlinear models, but may have a poor fit elsewhere.
In the paper, three periods were used for comparison: (1)
age 3) to did,. (2) 3 \ton20.) andy@) 3) toon sihe) havaluenthemroot
mean square of error and the coefficient of determination are used
to judge the fit of the model.
The F values ranged from a minimum of 3040 to a maximum of
999999 (Table 2). All models were significant at the 0.0001
probability of error.
Table 2. Fit between data and model among seed sources in three
time periods at age 3 to 15, 3 to 20 and 3 to 25 years.
Model Statistics
Seed F value Root mean square
Source Byres ANS) 3—5 20 3u—e25 3 - 15 3 - 20 S25
301 18,909 US AhS)5) 12/333 -009 Oule2 SORES)
303 58,791 32), 07/01: 9,387 .005 .008 -016
305 B25 Ole W265 PaaS) 38,561 007 008 008
307 ZO Ee eae Oo> 6,166 .007 O22 O22
309 WSS OLey/ Sad 7/0) 3,040 .003 019 .029
Syl 999,999 UV auus} 10,099 -O01 .018 5(0)i07/
Syks) 10,901 Tal eas) 7,760 O11 014 One2
Syki7/ 70,568 Woe 4,596 005 017 O25
319 45,775 11,704 AS MAY 006 O13 -014
BVA 483,648 VARS 5, 834 -002 .006 .022
323 29 , 266 SS) L/P 20,694 007 .008 eAO LIE
375 4,926 NUS TR SAS) 10,270 eOile7: One2 .016
327 2,844 4,473 4,068 s023 O22 .029
329 2,104 Hasisial Vj 232 027 -020 .020
33m MES) 5 Abe SMU AOS 6,114 004 .003 O22
Mean 009 013 019
St. Dev 008 006 006
GéWoa 25 85 45 32
The root mean square of fitting errors ranged from a minimum
of .001 to a maximum of .029. On the average the fitting error
was .009 for the 3 to 15 year period and increased to .019 for the
3 to 25 year period. The average for the three periods was .014.
For small values of e, we have approximation of Exp(e) = 1+ e.
Therefore, the small size of the error term represents the rela-
tive error. In other words, the percent of error for the 3 — 15
year period would by only .9 percent, corresponding to the average
RMSE of .009. The average of the relative error for the study is
1.4 percent.
344
The coefficient of determination ranged from 0.9997 to 1.000
among 45 regression models. It may lead one to wonder whether a
perfect fit has been achieved.
PREDICTABILITY OF GROWIH POTENTIALS
One of the most rigorous tests of a fitted equation is cross
verification with a second sample taken at another time (Daniel
and Wood 1980). Because this is impossible, we have used a
longitudinal verification for the growth curve. First, we devel-
oped regressions based on 3 to 15 and 3 to 20 year periods for
each seed source. The second step was to project the height at
age 25 from each regression. The final step was to compare the
projected and the observed height. The results are presented in
Table 3.
Table 3.--Comparison between projected and observed height at 25
years of age.
Seed Observation Projected from ages Error rate
Source at age 25 3-15 3-20 3-15 3-20
------- 0.01 ft. ------ --- £---
301 6,092 5,858 5,944 -3.9 -2.4
303 6,339 6,051 (oy UWA -4.5 -3.5
305 6,410 6,278 6573377. -2.1 -1.1
307 5,907 Soy at 5,741 -5.7 -2.8
309 6,305 5,800 5,963 -8.1 -5.4
sal 5,964 5,699 5,847 -4.4 -2.0
315 6,007 By Ysls) 5,805 -4.8 -3.4
SUZ 5,865 5,456 5,588 -7.0 -4.7
319 6,069 5825 5,932 -4.0 —-2.2
321 6,028 5,681 5/20 -5.8 -5.0
323 6,162 W977 6,027 -3.0 -2.2
325 5,849 DOV 5,686 -3.0 -2.8
BV7/ 5,681 Di, 339 5,444 -6.1 -4.2
329 5643 5,461 55510 -3.2 2.4
331 5,949 5,646 5,641 -5.1 -5.1
Mean 6,018 Di V/ Ge) SYA te7Alk -4.7 -3.3
St. Dev. 222 228 239 6 3
All projections were lower than the observed heights. The 10
year projection was low by 2.8 feet or 4.7 percent; while the 5
year projection was low by 2.0 feet or 3.3 percent. However the
Standard deviation of the three groups were the same. The projec-
tions were as precise as the observations even though the accuracy
may be off a little. The standard deviation for the relative
error rate were 1.6 and 1.3 percent respectively for the 10-year
and 5-year projections.
The bias in projection is not the fault of using transforma-
tion. The transform of the expected value does not equal the
345
expected value of the transform, although they are crude approxi-
mations of each other (Kruskal 1978). The percent of bias calcu-
lated according to the formula given by Wiant and Harner (1979)
was less than 0.04 percent. Because of the small error variance
in the regressions, the adjustment of the prediction (Baskerville,
1972) offered little reduction in the bias. This large and
consistent bias needs to be rectified by adding 3 percent to the
5-year forecasting and 5 percent to the 10-year forecasting.
POSSIBILITY OF INFERENCE
Growth potential and growth rate are of interest to the
forest manager. The function 1n(HT) = BO + B1(1/AGE) + B2 In(1 +
1/AGE) indicates that the asymptotic height should be near the
value of Exp(BO). For example, the maximum value (9.334) for BO
among 15 seed sources was found in provenance 305, therefore the
asymptotic height would be Exp(9.334) x .01 ft. = 113 ft. On the
other hand, a minimum of BO in seed source 327 indicated that the
average asymptotic height could be 92 ft. Notice that the average
height at age 25 for the complete study is 60 feet, the average
site index for the complete experiment could be estimated as 90
ft. at age 50. Which is the average site for loblloly pine.
Trees at that site grow slowly after 50 years of age, hence our
estimate of asymptotic height seems to be reasonable.
The function can be differentiated. The instantaneous growth
rate of log (Ht) at age x is: 9 D
d(1nHT)/dx = -Bl/x” - B2/(x” + x) 3 9
= (-x(Bl + B2) - Bl)/(x +x)
Using the coefficients developed from the 3 to 25 year as an
example we found in Table 4 that seed source 321 would have the
gregtest growth rate of log (Ht) at age x as ((17.15x - 33.90)/(x
+ x), while seed source 325,,would have the smallest growth rate
as ((14.81x - 24.58)/(x + x’) at the age of x years. Because of
the extremely high correlation between Bl and B2 (r = 0.997),
selection of growth rate can be simplified as selection for Bl.
POWER OF DISCRIMINATION
If simple functional differences among genotypes were exis-
tent or nonexistent, a growth function should have the power of
discrimination to prove or to disprove the differences in growth
form. By fitting the mean height from age 5 to 55 years old of 13
populations of Douglas-fir in Wind River, Oregon, Namkoong, Usanis
and Silen (1972) found that the parameters A, C, and m of the
Richards' function were nonsignificant among populations. Using
the Weibull function to quantify sweetgum germination, it was
found that the coefficients b and c were significant among fami-
lies within the stand, but all of the three coefficients (a, b, c)
were not significant among stands (Rink et al. 1979). The signi-
ficant difference in a given parameter indicated that selection in
that characteristics of the growth curve may be possible. On the
other hand, no growth rate can be selected if all growth curves
are the same.
346
Table 4.--Summary of regression coefficients in growth models for
15 loblolly pine seed sources during three time period.
—————— eee
Coefficient
Soi eB ORS Sl eee) earn B2
Seedlot 3-15 3-20 3-25 c.V. 3-15 3-20 3-25 C.V. 3-15 3-20 3-25 c.V.
301 9.222 9.248 9.267 24 22.56 23.89 25.02 5.17 -36.98 -38.58 -39.96 3.87
303 2712: 9.289 9.318 25 24.66 25.56 27.22 5.03 -39.53 -40.62 -42.65 3.87
305 9.309 9.324 9.384 13 25.04 25.85 26.38 2.62 -39.93 -40.91 -41.56 2.01
307 9.217 9.267 9.290 40 27.36 29.96 31.28 6.75 -42.98 -46.16 -47.77 5.34
309 9.206 92/253 9.298 50 23.41 25.82 28.40 9.64 -37.65 -40.51 -43.74 7.50
311 9.252 9.296 9.311 33 28.58 30.80 S72 5.32 -44.55 -47.27 -48.38 4.22
315 9.212 9.239 9.266 29 24.46 25.82 27.40 5.68 -39.26 -40.93 -42.85 4.38
317 9.196 9.236 9.275 43 27.13 29.20 31.44 7.37 -42.75 -45.28 -47.99 5.78
319 9.240 9.270 9.289 27 26.65 28.22 29.28 4.72 -41.71 -43.63 -44.92 3.72
321 9.268 9.281 9.322 30 30.84 31.53 33.90 5.00 -47.33 -48.17 -51.05 3.99
323 9.249 9.264 9.281 17 22.09 22.83 23.85 3.86 -36.65 -37.55 -38.79 2.85
325 9.193 9.197 9.220 16 23.04 23.26 24.58 3.53 -37.51 -37.79 —-39.39 2.65
327 Or125 9.160 9.194 38 22.15 23.83 25.80 7.64 -36.35 -38.52 -40.92 5.92
329 9.189 9.204 9.223 19 26.65 27.42 28.53 3.43 -42.06 -43.01 -44.35 2.67
331 9.235 9.236 9.277 26 24.70 24.72 2713 5.48 -40.41 -40.43 -43.35 4.10
Mean 9.226 9.250 9.278 29 25.29 26.58 28.13 5.41 40.37 41.95 43.84 4.19
St. Dev. 043 042 040 -10 2.54 2.80 2.93 1.83 3.18 3.46 3.62 1.45
Gavin, x -47 45 -43 10.06 10.55 10.41 7.89 8.25 8.26
The functions presented in this paper have great power of
discrimination among seed sources of loblolly pine. All three
coefficients BO, Bl and B2 are significant beyond the 0.001
probability of error (Table 5). The repeatability or provenance
heritability calculated from the F value (Kung and Bey 1978) was
0.98 for each coefficient.
Table 5.--Analysis of variance of the regression coefficients.
Coefficient
BO Bl B2
Source alent MS F MS F MS F
Seed source 14 .0050 40° 2220. 63 BLO" 655
Age period 2 ROLOM 62s 30531 86. G52), 860
Error 28 .0001 35 52
Total 44
“Significant at the 0.0001 level.
PERSISTENCY OF COEFFICIENT
A good growth curve should be relatively independent of the
range of data base. If the growth curve developed from the
juvenile growth period were the same as that developed from the
complete life span, we would be more successful in making early
selection.
The persistency of coefficient in the three periods (3 to 15,
347
3 to 20, and 3 to 25 years of age) is evident in Table 4. The
average coefficient of variation among three periods was only .29
in£Or BORD Die474 fOr Bil ande4 2=4 store B2.
As the range of ages becomes wider, the absolute values of
the BO, Bl and B2 also increases. The differences are significant
beyond the 0.001 probability of error (Table 5). Although one
would like to have a constant regression coefficient throughout
the years if possible, the second best would be that for any given
parameter, it may differ from one period to another, but it should
vary in a predictable manner.
The persistency of coefficient can be shown also by the
correlation coefficient for BO, Bl and B2 among three periods
(Table 6). All correlations are significant at the 0.001 level.
From the period, of 3 to 15 years to the period of 3 to 20 years,
the Bl increases by 6 % and the B2 increases by 4 %; while from
the period of 3 to 20 years to the period of 3 to 25 years both Bl
and B2 increases only 2 Z.
Table 6.--Correlations among three periods for parameters BO, Bl
and B2.
Correlation between Correlation coefficients for parameters
periods BO Bl B2
342520 Bie 2015 wg 97 98 98
3.— 15 3 - 20 94 96 96
3 - 15 3 - 25 .89 94 94
DISCUSSION
Many growth and yield models are available to tree improve-
ment workers. From a practical point of view we recommend the 5-P
criteria for model selection. However, we have not assigned any
weight to each criterion which may differ from one program to
another.
We use the range means of the seed source and not the indivi-
dual tree in hope that if all the environmental errors could be
averaged out we would have a more accurate evaluation of the
performance of the growth model as well as the genetic difference
among seed sources.
The function used for illustration is almost ideal in the 5-P
criteria. However, we are still searching for a perfect curve
which is totally independent of the age range. Would it be
possible to have a model which regression coefficients are change-
less between any period range from 1 to 100 years? or is it an
impossible dream?
348
LITERATURE CITED
Baskerville, G. L. 1972. Use of logarithmic regression in the
estimation of plant biomass. Can. J. For. Res. 2:49-53.
Daniel C., and F. S. Wood. 1980, Fitting Equations to Data. 2nd
ed., 458 p. Wiley, New York.
Franklin, E. C. 1979. Model relating levels of genetic variance
to stand development of four North American conifers. Silvae
Genetica 28:207-212.
Kozlowski, T. T. 1971. Growth and Development of Trees, Vol. I,
Seed Germination, Ontogeny and Shoot Growth. 443 p. Academic
Press, New York and London.
Kruskal, J. B. 1978. Transformations of data. In International
Encyclopedia of Statistics, Vol. II. 1,350 p. The Free Press,
New York.
Kung. shaeie, and Coy r. Bey., 1978." Heritability construction for
provenance and family selection. Lake States For. Tree Impro.
Conf. 13:136-146.
Kunpiin He a He 1984. A law-like empirical growth function.
Abstract Am. Stat. Ass. Meeting 1984:234.
Lundgren, A. L., and W. A. Dolid. 1970. Biological Growth
Functions Describe Published Site Index Curves for Lake State
Timber Species. U.S.D.A. For. Serv. Res. Paper NC-36. 9 p.
Namkoong, G., R. A. Usanis, and R. R. Silen. 1972. Age-related
variation in genetic control of height growth in Douglas-fir.
Theo. Appl. Genet. 42:151-159.
Nance, W. L., and O. O. Wells. 1981. Estimating volume potential
in genetic tests using growth and yield models. South For. Tree
Impro. Conf. 16:39-46.
Prodan, M. 1968. Forest Biometrics. 447 p. Pergamon Press, New
York.
Ratkowsky, D. A. 1983. Nonlinear Regression Modeling. 276 p.
Marcel Dekker, Inc., New York and Basel.
RingwaGagsals Rs, Dell, iG. Switzer, ‘and Hi. el Bonner: » 19/9 psUsevot
the Weibull function to quantify Sweetgum germination data.
Silvae Genetica 28:9-12.
Wells, O. O., and P. C. Wakeley. 1966. Geographic variation in
survival, growth and fusiform-rust infection of planted loblolly
pine. For. Sci. Monogr. 11:40.
349
Wiant, H. V., and E. J. Harner. 1979. Percent bias and standard
error in logarithmic regression. For. Sci. 25:167-168.
Yang, R. C., A. Kozak, and J. H. G. Smith. 1978. The potential
of Weibull-type functions as flexible growth curves. Can. J.
For. Res. 8:424-431.
350
Evaluation of Slash Pine for Resistance to Pitch Canker
C. H. McRae, D. L. Rockwood and G. M. Blakeslee!
Abstract.--Two- to three-year-old orchard open-pollinated
seedlings from 46 slash pine (Pinus elliottii Engelm. var.
elliottii) clones were evaluated for resistance to Fusarium
moniliforme var. subglutinans (FMS). These families, repre-
senting fast growing and/or fusiform rust resistant genotypes,
were planted near Gainesville, Florida. Terminal and lateral
shoots of 21 to 24 seedlings per family were wounded and
inoculated with a polymix of four isolates of FMS. Family
mean symptom expression ranged from 16.6 to 91.7%; shoot
mortality ranged from 4.2 to 91.7%. Strong individual and
family heritabilities suggested that genetic resistance may be
useful in management of pitch canker. Estimated gains from
four improvement options are presented. There were no signi-
ficant correlations between pitch canker resistance and either
fast-growth or fusiform rust resistance.
Additional Keywords: Pinus elliottii var. elliottii, Fusarium
moniliforme var. subglutinans, heritability, genetic varia-
tion, genetic gain, fusiform rust, Cronartium quercuum f. sp.
fusiforme.
INTRODUCTION
FMS infects many southern pines. This fungus has been especially
damaging in slash pine, inciting resin-soaked cankers on the branches and
main stem resulting in shoot dieback, stem deformity, reduced growth,
and mortality (Blakeslee et al. 1980, Blakeslee and Oak 1979, Dwinell et
al. 1985, Phelps and Chellman 1976). Prospects for long-term control of
pitch canker are strengthened by studies showing genetic variation
within pine species for resistance to pitch canker (Barnett and Thor
1978, Blakeslee and Rockwood 1978, Dwinell and Barrows-Broaddus 1979,
ne lwvand= "1983, ~ Dwinell” et all; §1977). This paper reports further
evidence of genetic variation in slash pine and estimates genetic gains
that could be obtained from using resistant genotypes.
sl Geshe Research Assistant, and Associate Professors, respectively,
Department of Forestry, University of Florida, Gainesville, FL. The
assistance of the Cooperative Forest Genetics Research Program at the
University of Florida is gratefully acknowledged.
351
MATERIALS AND METHODS
Open-pollinated progeny of 46 fast-growing and/or rust-resistant
clones were planted in a randomized complete block design on a fertile,
well-drained site in north central Florida in 1981 and 1982. The trees
were planted at a 10' x 2' spacing in 10-tree row plots that were
replicated three times.
From September 24 through October 18, 1983, the terminal and one
lateral shoot of each tree were inoculated with an aqueous suspension of
conidia of FMS. At the time of inoculation, the trees averaged 7.25
feet in height and were vigorous and healthy. The eight most vigorous
of the ten trees planted per row plot were selected for inoculation.
Precipitation and ambient témperature were monitored on the site during
the inoculation period.
The inoculum consisted of four proven-pathogenic isolates of FMS
obtained from slash pines in Volusia, Franklin and Gilchrist counties in
Florida. The isolates were single spored and grown on carnation
leaf-water agar for about 10 days priox to use. An aqueous suspension
of conidia from each source (1-1.5 x 10° conidia/ml) was prepared daily,
and equal aliquots of each source were combined just prior to use.
Post-inoculation germination checks were made daily, and viability
consistently exceeded 96%.
Prior to inoculation, shoots were surface sterilized by spraying
with 95% ethanol and allowed to dry. Each shoot received two wounds at
the same level, located on opposite sides of the third flush of 1983
growth. The wounds were made with an 18-gauge needle, and two drops of
inoculum were placed in each wound. Control branches, selected at
random within the family, were treated in the same manner except that
sterile water was used in place of the inoculum.
At about 18-day intervals for the next eight months, each shoot was
examined for symptom development. At each observation, the condition of
the shoot was rated as 1 (no symptoms), 2 (foliage discolored) or 3
(foliage brown and shoot dead). So that maximum disease expression
could be obtained for all trees, the shoots were harvested according to
disease severity, with dead shoots being harvested and processed in the
laboratory before living shoots were harvested. In July 1984 when new
symptom expression had essentially ceased, all remaining shoots were
collected and brought to the laboratory for detailed examination and
isolation.
Two responses on terminal shoots were analyzed; percent of trees
with pitch canker symptoms (conditions 2 and 3) and percent of trees
with pitch canker induced mortality (condition 3 only). To determine
significance of blocks and families, analysis of variance was conducted
suing plot means. For genetic analyses, these two binomial responses
were handled in the same manner as percent rust-infected data (Rockwood
and Goddard 1973). Individual tree and family mean heritabilities were
calculated on the assumption of half-sibs. Genetic gains employed
techniques presented by Shelbourne (1969). Selection intensities
assumed were: 10% in seed production areas and for tested clonal
352
orchards, 2% for untested clonal orchards and 40% in existing clonal
orchards.
Correlations between traits were based on family means. Clonal
evaluations for growth and rust resistance included in the correlations
were weighted, standardized comparisons from numerous progeny tests
involving open-pollinated progenies and an unimproved check lot.
RESULTS AND DISCUSSION
Symptom development followed a typical disease progress curve
(Figure 1). Observed symptoms included those regularly associated with
pitch canker on slash pine, via. discoloration and death of needles
around the inoculation sites, exudation of pitch from infected tissues
surrounding the inoculations, discoloration and death of needle and
shoot tissues distal the the cankers, and, for those trees where
advanced symptom development did not occur until the following spring, a
second flush of symptoms coinciding with the spring flush of growth.
The pitch canker pathogen was reisolated from about 90% of the
inoculated shoots and 2% of the control shoots, thus indicating that the
observed symptoms were due to the pathogenic action of the introduced
100 —
% Disease Expression
Block 1 Mean
Block 2 Mean
A Block 3 Mean
ie} 43 nS 11 161 197 IL 264 300
(e}
J
0
et (5)
Days From Inocuiction
Figure 1. Pitch canker symptom expression disease progress curves based
on block means.
353
fungus. Failure to obtain higher yields from the inoculated shoot
reisolation may be due to incomplete isolation from the canker, or loss
of viability of the fungus due to active defense of host, thus limiting
tissue available for colonization. Reisolation of the pathogen from a
small percentage of the control shoots can be readily attributed to the
presence of idigenous FMS inoculum that can infect susceptible tissue
when wounded.
Family mean responses to the pathogen were normally distributed,
ranging from highly susceptible to highly resistant (Table 1). A
similarily wide range of incidence has been observed in loblolly pine
clones (Dwinell et al. 1977). However, there were certain trees that
became infected but progressive tissue colonization did not occur as the
canker was stabilized and the infection delimited by the production of
hypertrophic callus around the canker. Those trees, based on mean
symptom expression, would appear to be poor performers, but when
evaluated with respect to shoot mortality, appear to have functional
resistance. Overall, there was a strong correlation between symptom
expression and mortality response.
Significant differences existed among families and among blocks
(Table 2). The differences between blocks may be related to differences
in edaphic conditions, topography, variations in rainfall during the
inoculation period or the slightly higher percentage of younger
(two-year-old) trees included in the third block.
Table 1. Slash pine family means for percent pitch canker symptom
expression (Symp) and mortality (Mort) of terminal shoots.
Pitch Canker Pitch Canker
Family Symp Mort Family Symp Mort
1-60 16.6 Sis 56-56 62.5 S}3}58)
23-59 20.8 Gren: 327-56 64.9 26.2
M-835 29.2 G3) 102-57 66.7 2 Diced
57-56 3373 26S) 261-56 66.7 29 2:
239-56 3125 2550 285-55 66.7 ANE?
M-204 40.3 13.9 69-56 66.7 ZOD
205-55 45.8 125 163-57 68.3 So 05
24-60 45.8 16;.7, 173-57 70.8 25.60
13-59 47.0 8056 33-58 70.8 fliee2
89-57 48.2 176 48-59 70.8 SES)
M-308 50.0 S83 60-56 70.8 29.2
16-59 50.0 25.0 B-106 7560 20.8
163-58 JOO W255) 65-56 a0) 45.8
31-60 50.0 Dd) 5 2 91-58 7560 50.0
66-73 50.0 20.8 M-109 76.4 45.8
27-58 50.8 Na 3} 13-56 TQ2 66.7
M-817 A572 4.2 293-55 UO 2 BAG
64-56 DAG 3755) 106-56 333} 8) BKC)
357-56 Sah 26.8 265-55 S33 66.7
57-61 58.3 25550) 76-58 83.3 45.8
100-56 62.5 29.2 130-60 87 E 30.9
330-56 6255 EN 7 347-56 Biers FIO
342-56 62.5 50.0 70-56 Olea Ok ul,
354
Table 2. Analyses of variance, heritabilities and genetic gains for
percent pitch canker symptom expression (Symp) and mortality
(Mort) in slash pine.
Pitch Canker
Ss) ) ee Soe Morte aaed
Analysis of Variance
Source DE: MS F MS F
Block yp. 3857.6 10.40% 1574.8 5.58%
Families 45 927.8 2.50% 1070.2 3.79%
Error 90 Sy/raO DOLeD
Heritabilities
Individual 5203 - 383
Family Mean - 600 750
Genetic Gains
Short-term Options -
Seed Production Area 39.4 LOW <7
Tested Clones in
Existing Seed Orchard 16.8 41.4
Long-term Options -
Orchard of
Untested Select Trees S575) 148.9
Orchard of
Tested Clones 60.7 149.3
*
Significant at the 1% level.
Family differences were somewhat greater for percent mortality than
for percent symptoms, and this relationship was further evidenced in
heritabilities and genetic gains (Table Di Individual tree
heritabilities were strong and family heritabilities also suggested
potential for genetic gain.
A variety of short- and long-term options are available for
realizing the genetic potential for reducing pitch canker incidence.
Short-term alternatives include seed production areas in
heavily-infected plantations or natural stands in epidemic areas, a very
successful strategy for developing fusiform rust resistance (Goddard et
al. 1975), and collection of seed from tested clones in established seed
orchards. The more productive alternative appears to be seed production
areas (Table 2). Assuming that 20 clones of a typical 50-clone orchard
will be resistant and contribute seed, existing orchards provide a
relatively small, but still meaningful, gain.
The long-term options involving new clonal orchards offer greater
improvement. An orchard of proven pitch canker resistant clones would
provide slightly more gain than an orchard of untested clones derived
from pitch canker epidemic areas.
355
Simultaneous implementation of short- and long-term options, if
possible, is desirable for developing progressively resistant planting
stock. Conversion of stands to seed production areas is a relatively
low cost alternative applicable to some forestry organizations. Initia-
tion of new orchards is immediately possible due to ongoing screening
efforts which have identified more than 25 resistant trees.
Correlations between pitch canker resistance, rust resistance, fast
growth and tree height were insignificant (Table 3). Lateral shoot
response showed a strong relationship to the response of terminal shoots
(Table 3) notwithstanding the differences in shoot size and phenology
These results suggest that fast growth and rust resistance need not be
compromised in selecting for pitch canker resistance. Several of these
46 clones were superior in all three characteristics.
Additional testing involving different environments and additional
clones will be conducted in 1985-86. Data obtained will permit expanded
examination of the results reported in this paper.
CONCLUSIONS
These results from inoculations of vigorous, healthy, field-grown
trees suggest that genetic resistance to pitch canker is present in
slash pine and that significant genetic gains may be realized.
Pitch canker resistance appears to be unrelated to either rust
resistance or fast growth, indicating that fast-growing, rust-resistant
and pitch canker resistant trees may be selected.
Table 3. Correlation coefficients among slash pine family means for
terminal (Term) and lateral (Lat) shoots pitch canker symptom
expression (Symp) and mortality (Mort), tree height and
clonal rust and growth evaluations.
Shoots Clonal
Term. ats Evaluation
Tree
rad t Mort Symp Mort Height Rust Growth
Term —- Symp ./8% ./9% 5 [aes iO -.01 505
Mort ./3% ~//% -09 -.O0l -.05
Lat = Symp 87% -.14 03 -.04
Mort -.1l1 -.01 -.0l
Tree Height 02 Oks)
Clonal Rust -.25
FS
Significant at the 1% level.
356
LITERATURE CITED
Barnett, P. E., and E. Thor. 1978. Effects of site and inheritance on
Fusarium incidence in Virginia pine. USDA For. Serv. State
and Priv. For. Tech. SA-TP2:159-161.
BilakesteesuG. | M.,4 bebe. Dwinelll and») Ri. LiovAnderson: 1980. Pitch
canker of southern pines: identification and management
considerations. USDA For. Serv. Southeast. Area State and
Private For. Rep. SA-FRI1. 15 p.
Blakeslee, G. M., and S. W. Oak. EOS Significant mortality
associated with pitch canker infection of slash pine in
Florida. Plant Dis. Rep. 63:1023-1025.
Blakeslee, G. M., and D. L. Rockwood. 1978. Variation in resistance of
slash pine to pitch canker caused by Fusarium moniliforme var.
subglutinans. Phytopath. News 12:207-208.
Dwinell, L. D., and J. Barrows-Broaddus. TENS Susceptibility of
half-sib families of slash and loblolly pine to the pitch
canker’ fungus, Fusarium moniliforme var. subglutinans.
Phytopath. 69(5):527.
Dwinell, L. D., and J. Barrows-Broaddus. 1981. Pitch canker in seed
orchards. Proc. 16th S. For. Tree Imp. Conf.:234-240.
Dwinell, L. D., and J. Barrows-Broaddus. 1983. Pine wilt and pitch
canker of Virginia pine in seed orchards. Proc. 1/th S. For.
Tree Imp. Conf.:55-62.
Dwinell, L. D., J. Barrows-Broaddus, and F. G. Kuhlman. NOS 5." Pitch
canker: A disease complex of southern pines. Plant Dis. Rep.
69:270-276.
Dwinelsl as Law. «7b. Ryan and, E.G. Kuhlman. LOT. \ Pitch canker” of
loblolly pine in seed orchards. Proc. 14th S. For. Tree Imp.
Conf .:130-137.
Goddard.coR?. Ey Re Aks Schmidt; and FE. Vande) Linde. ILS) FAY Immediate
gains in fusiform rust resistance in slash pine from rogued
seed production areas in severely diseased plantations. Proc.
14th S. For. Tree Imp. Conf.:197-202.
Phelps, W. R., and C. W. Chellman. 1976. Evaluation of "pitch canker"
in Florida slash pine plantations and seed orchards-1976.
USDA For. Serv. Southeast. Area State and Private For. 22 p.
Rockwood, D. L., and R. E. Goddard. 1973. Predicted gains for fusiform
rust resistance in slash pine. Proc. 12th S. For. Tree Imp.
Conf 31-3) «
Shelbourne, C. J. A. 1969. Tree breeding methods. New Zealand For.
Serv. Tech. Paper No. 55. 43 p.
357
COLD TOLERANCE VARIATION IN LOBLOLLY PINE NEEDLES FROM DIFFERENT
BRANCH TYPES, FAMILIES, AND ENVIRONMENTS
Ly oh cKkolbmands kK. aC. steiner!!
Abstract.--The cold tolerances of loblolly pine needles
from different open-pollinated families, branch types, field
blocks, and test locations were measured by the electrical
conductivity method. Significant differences in tolerance
were found between families, "upper" and "lower" growth inter-
nodes, blocks, and locations. Family differences in tolerance
were more pronounced among l-year-old seedlings in a nursery
environment than among 12- and 13-year-old trees in plantation
environments. Results indicate that non-genetic sources of
variation and genotype x environment interaction may bias as-
sessments of cold tolerance genetic variation in loblolly pine.
Keywords: Pinus taeda, cold tolerance, genetic variation.
INTRODUCTION
Loblolly pine (Pinus taeda L.) is the preferred pulpwood species for
many areas located immediately north of its natural range (Allen 1953,
Aughanbaugh 1957, Lambeth et al. 1984). However, winter injury to loblolly
pine in these areas may reduce growth rates or cause mortality, thus reducing
productivity below the potential for the site (Boggess and McMillan 1954,
Minckler 1952, Thor 1967, Wells and Rink 1984). Consequently, loblolly pine
genetic improvement programs should stress the development of varieties pos-
sessing both cold tolerance and rapid growth rates.
Reliable techniques of screening loblolly pines for genetic differences
in cold tolerance are needed to expedite the production of hardy varieties.
Assessments of cold tolerance differences among progenies in field tests are
desirable since the trees acclimate under natural environmental conditions.
However, field assessments depend on the fortuitous occurrence of adequate
test winters and may require observations over many years to detect anything
but large differences in tolerance. A modification of the electrical conduc-
tivity method (Dexter et al. 1930, 1932) was used by Kolb et al. (1985) to
accurately measure differences in cold tolerance among open-pollinated fami-
lies of loblolly pine growing in a western Kentucky field test. This paper
reports the results of two studies designed to identify non-genetic sources
of variability which may bias assessments of cold tolerance genetic variation
in loblolly pine. Study One emphasized within-tree and field block sources
of variability, while Study Two emphasized ontogenetic and plantation sources
of variability.
one authors are, respectively, Graduate Research Assistant, and Associate
Professor of Forest Genetics, Forest Resources Laboratory, The Pennsylvania
State University, University Park, Pennsylvania 16802. Journal Article No.
of the Pennsylvania Agricultural Experiment Station. The authors wish to
acknowledge the financial assistance provided by Westvaco Corp. in performing
the research and the valuable contributions of Henry F. Barbour in executing
the study.
358
STUDY ONE
Methods and Materials
On December 15, 1982, needle samples from three open-pollinated families
(1-31, 6-20, 18-94) were collected from each of four field blocks in the
Westvaco Corporation's 1976 progeny test in Calloway Coynty, Kentucky. These
families represent a wide range of hardiness based on assessments of winter
injuries that occurred in the progeny test in 1977 (Kolb et al. 1985). Family
1-31 was the most hardy of these families, 6-20 was intermediate, and family
18-94 was the least hardy.
In each field block, collections consisted of two lower and two upper
growth internodes from branches which flushed two or three times in 1982.
The "upper" growth internodes on three-flush branches were defined as the
middle internode, and "lower" internodes were consistently those in the
lowermost position on both two-and three-flush branches. Two- and three-
flush branches were collected from one tree each in a ten-tree row plot for
each family in a block. The collection of samples was limited to branches
formed from the main stem in 1981 on the north side of the tree. This sam-
pling scheme produced 48 treatment combinations in factorial arrangement: 3
families, 4 field blocks, 2 branch types (two- and three-flush), and 2 inter-
nodal positions, each combination represented by samples from two branches
on the same tree. Samples were promptly packed into coolers and mailed to
University Park, Pennsylvania, where they arrived the morning following
collection.
Samples were prepared for laboratory exposure to low temperatures by
bulking an approximately equal number of needles from the two branches
representing each treatment combination, cutting these into 5 cm segments
from the fasicle end, and randomly choosing approximately 25 segments for
each desired temperature exposure. Needle samples of each treatment combi-
nation were commonly exposed in a freezing chamber to the following tempera-
tures: 5°C (unfrozen control), -l10°C, -15°C, -20°C, -25°C, -30°C, -35°C,
-40°C, -45°C. The temperature in the chamber was lowered at a rate of 4°C
per hour, and each desired temperature exposure was maintained for 30 minutes.
Needle samples were removed from the chamber following each desired exposure,
and slowly thawed to an ambient temperature of 5°C. Electrolytes from each
sample diffused into 10 ml of deionized water for 24 hours after thawing. The
electrical conductivity of each diffusate solution was measured both before
and after an autoclaving treatment at 245°C for 30 minutes.
As initially described by Dexter et al. (1930, 1932), the electrical
conductivity of diffusate from plant tissues injured by low temperatures is
higher than that of diffusate from uninjured tissues. To obtain a measure of
injury due to low temperature exposure, a "relative electrical conductivity"
was calculated for the diffusate solution from each sample by dividing the
electrical conductivity before autoclaving by the electrical conductivity
after autoclaving. This index eliminates spurious effects caused by the ten-
dency of some samples to have a higher electrical conductivity due to differ-
ences in needle sample size or nutrient status (Wilner 1959).
359
Analysis of variance on observations of relative electrical conductivity
were used to determine if cold tolerance varied between needles from differ-
ent sources. In these analyses, the temperature x needle source interaction
was of primary interest since the significance of this term indicated whether
needles from different growth internodes, branch types, families, or blocks
varied in their injury response to temperature and consequently cold toler-
ance,
Results
The first hypothesis of interest was that the cold tolerance of loblolly
pine needles does not differ between lower and upper growth internodes. This
was tested by an analysis of variance on observations of relative electrical
conductivity for upper and lower growth internodes averaged over two- and
three-flush branches and families. Needles from lower growth internodes were
significantly (p < 0.01) less cold tolerant then those from upper growth
internodes. This difference is illustrated in Figure 1-- needles from lower
internodes were injured more rapidly in response to decreasing temperature
than needles from upper growth internodes.
The second hypothesis of interest was that tolerance does not differ
between needles from two- and three-flush branches. This was tested by
analysis of variance which compared the tolerances of needles from the two
branch types averaged over families for lower and upper internodes, respec-
tively. Needles from comparable internodes (lower or upper) did not differ
significantly in tolerance when obtained from either two- or three-flush
branches. As shown in Figure 2, needles from lower internodes on two- and
three-flush branches were injured approximately the same. However, Figure 3
shows a possible difference in response between needles from upper internodes
of two- and three-flush branches. Thus, needles from upper internodes may
differ slightly in cold tolerance between branch types (although not signifi-
cant in this experiment), while needles from lower internodes appear to be
relatively stable in tolerance with respect to branch type.
The final hypotheses of interest were that the cold tolerance of needles
does not differ among the three families and four field blocks sampled in
this experiment. To test these hypotheses, data for the three families were
averaged over two- and three-flush branches using only lower internodes, as
suggested by the results of the previous analysis. Families did not differ
significantly in tolerance. However, the pattern of response of families to
temperature shown in Figure 4 is identical to that suggested by winter injury
to families under field conditions in 1977: family 18-94 was injured the
most rapidly, 6-20 was intermediate, and 1-31 was injured the least rapidly.
This comparison suggests that cold tolerance varied among families, but that
these differences were not statistically detectable at the level of precision
present in this experiment.
It was also apparent in this analysis that needles from the four field
blocks differed significantly (p < 0.005) in tolerance. As shown in Figure
5, needles from block one were injured much more rapidly than needles from
other blocks. These block influences on cold tolerance are presumably
related to variations in microsite within the plantation and perhaps related
to the fact that block one is located in a somewhat moister area than the
others.
360
STUDY TWO
Methods and Materials
On February 8, 1984, needle samples were collected from each of four
open-pollinated families of loblolly pine (3-4, 3-41, 6-8, 6-22) at
Westvaco's 1972 test in Livingston County, Kentucky (13-year-old trees),
Westvaco's 1973 test in Hickman County, Kentucky (12-year-old trees), and
in Westvaco's progeny test seedlings at the J. P. Rhody (Kentucky State)
Nursery, Marshall County, Kentucky (l-year-old trees). For each family,
progenies at all three test locations originated from the same seed orchard,
but not necessarily the same seedlot. Progenies 3-4 and 3-41 are from the
Champion International seed orchard in Newberry, South Carolina, and proge-
nies 6-8 and 6-22 are from the Champion International seed orchard in
Tillary, North Carolina.
For the 1972 and 1973 tests, a lower internodal segment of 1983 twig
growth from a lateral branch formed on the main stem in 1982 was collected
from seven trees per family in each of three field blocks. For each family
in the nursery, twenty whole seedlings were collected from each of two
blocks. All collections were packed into coolers and mailed to University
Park, Pennsylvania, where they arrived the next morning.
The preparation of samples, the freezing process, and the measurement of
injury was identical to that described for Study One with the following excep-
tions: 1) needles collected from blocks in the field were divided into four
replications for laboratory analysis, and 2) temperature treatments of -10°C,
-40°C, and -45°C were excluded. Relative conductivity data from each of the
test locations were subjected to analysis of variance to determine if genetic
differences in tolerance among families were detectable in each environment.
A combined analysis of variance was used to determine whether test locations
influenced overall levels of tolerance, as well as relative differences among
families.
To make specific comparisons of cold tolerance, injury response curves
were formulated by regressing mean relative conductivity on temperature treat-
ment using the following model:
= + Geol aa h
OMe ange) 244 dq, ceo
a = mean relative conductivity for family "i" at test location "j" for
each temperature treatment
X = temperature treatment
Wet
Ey. = residual for family "i" at test location "j
Similar regressions were performed on overall means at each test location to
compare injury responses among environments. The model fit the data ade-
quately, as indicated by R* values which ranged from 0.91 to 0.99.
361
Results
Overall levels of tolerance differed significantly (p < 0.005) among the
three test locations as illustrated by injury response curves shown in Figure
6. One-year-old seedlings from the nursery were generally the least tolerant,
trees from the 1973 test were intermediate, and trees from the 1972 test were
the most tolerant. Test location also influenced differences in tolerance
among families (family x temperature x location interaction significant, p <
0.005). Injury response curves shown in Figures 7 and 8 indicate no signifi-
cant differences in tolerance among families in the 1973 and 1972 test. In
contrast, family 3-41 was significantly (p < 0.005) less tolerant than the
other families in the nursery (Figure 9).
DISCUSSION
These studies indicate that differences in the cold tolerance of lob-
lolly pine needles may arise from both genetic and non-genetic sources of
variation. Consequently, confounding the sampling of tissues from families
or clones with branch positions, field blocks, or test locations may seri-
ously bias genetic assessments of cold tolerance. Consistent sampling is
especially important when measuring cold tolerances by the electrical con-
ductivity method because of its sensitivity in detecting differences in
injury. Exploratory studies to identify potential sources of variation are
a necessary precaution in using indirect measures such as the electrical
conductivity method for genetic assessments of cold tolerance in any species.
The differential injury response of families from the three test loca-
tions suggests that genetic variation in tolerance may be more pronounced
among seedlings in the nursery than among older trees in the field. The
reasons causing this interaction cannot be determined in this study since
environmental and ontogenetic effects were confounded. It is conceivable
that expression of genetic variation was greatest in the nursery environment
because of a correlation between tolerance and some aspect of seedling phys-
iology such as response to fertilizers or other cultural treatments used in
the nursery. Useful assessments of cold tolerance in loblolly pine breeding
programs will be complicated if such genotype x environment or genotype x age
interactions are common.
LITERATURE CITED
Allen, J. C. 1953. A half century of reforestation in the Tennessee Valley.
Jewbor.~ >) FlO6—113%
Aughanbaugh, J. E. 1957. Loblolly pine winning Ohio pine race. Ohio Farm
and Home Res. 42:95-96.
Boggess, W. R., and F. W. McMillan. 1954. Cold weather and glaze damage to
forest plantations in southern Illinois. Bull. Ill. Agr. Exp. Sta. No.
574,
Dexter, S. T., W. E. Tottingham, and L. F. Garber. 1930. Preliminary
results in measuring the hardiness of plants. Plant Physiol. 5:215-223.
362
Dexter, S. T., W. E. Tottingham, and L. F. Garber. 1932. Investigations of
the hardiness of plants by measurements of electrical conductivity. Plant
Physiol. 7:63-78.
Kolb, T. E., K. C. Steiner, and H. F. Barbour. 1985. Seasonal and genetic
variations in loblolly pine cold tolerance. For. Sci. in press.
Lambeth, C. C., P. M. Dougherty, W. T. Gladstone, R. B. McCullough, and
O. O. Wells. 1984. Large-scale plantings of North Carolina loblolly pine
in Arkansas and Oklahoma: a case of gain versus risk. J. For. 82:736-741.
Minckler, L. S. 1952. Loblolly pine seed source and hybrid tests in south-
ern Illinois. U.S. For. Ser. Central States For. Exp. Sta. Tech. Pap.
We
Thor, E. 1967. A ten-year-old loblolly pine seed source test in Tennessee.
J EOGs O5°3526—32)/ «
Wells, 0. O., and G. Rink. 1984. Planting loblolly pine north and west of
its natural range. Proc. Southern Silvicultural Research Conf. p. 261-
265.
Wilner, J. 1959. Note on an electrolytic procedure for differentiating
between frost injury to roots and shoots in woody plants. Can. J. Plant
Scines geo z= 3%
uw
(oe)
Yi
Sa
Lower Internode /
L
N
o
Upper
Internode
&
Relative Electrical Conductivity
3 a
o
Control -5 -10 -I5 -20 -25 -30 -35 -40 = -45
Temperature Treatment (°C)
Figure 1.--December 1982 injury (Relative Electrical Conductivity) to needle
tissues versus temperature treatment for "lower" and "upper" growth inter-
nodes. Graphs adjusted to a common y-intercept.
363
w
fe)
1
nN
on
ml
ot)
(e)
Relative Electrical Conductivity
a
peal oS
Sia lL
=i ny 4 — ee 1 uu —|
Control -5 -10 -15 -20 -25 -30 -35 -40 -45
Temperature Treatment (°C)
Figure 2.--December 1982 injury (Relative Electrical Conductivity) to needle
tissues versus temperature treatment for lower internodes of "two-flush" and
"three-flush" branches. Graphs adjusted to a common y-intercept.
30,
257 3-Flush / /
“i
4
2-Flush
20F- A
Relative Electrical Conductivity
_t — 1
Control -5 -10 -15 -20 =20) -30 =35 -40 -45
Temperature Treatment (°C)
Figure 3.--December 1982 injury (Relative Electrical Conductivity) to needle
tissues versus temperature treatment for upper internodes of "two-flush" and
"three-flush" branches. Graphs adjusted to a common y-intercept.
364
w
(2)
=
&
wo
aS
8-20
7 \-3i
= 25+ 7
2 AY /
3 4
DooL La,
Eze UA
rs) i.
2 15
Oo
2
W 19
@
>
i=)
@5
ic il _t Jie = ~ ait It 1 —j
Control -5 SOM EISeet 20825. 9-30) 359 240 45
Temperature Treatment (°C)
Figure 4,--December 1982 injury (Relative Electrical Conductivity) to needle
tissues versus temperature treatment for lower internodes of three open-
pollinated families. Graphs adjusted to a common y-intercept.
ol
a
ul
°o
nN
(e)
Relative Electrical Conductivity
Control -5 -10 -15 -20 =29) (-s0 -35 -40 -45
Temperature Treatment (°C)
Figure 5.--December 1982 injury (Relative Electrical Conductivity) to needle
tissues versus temperature treatment for four field blocks. Graphs adjusted
to a common y-intercept.
RELATIVE CONDUCTIVITY (%)
2
>
TEMPERATURE (°C)
Figure 6.--February 1984 injury (Relative Conductivity) to needle tissues
versus temperature treatment for four families in each of three environments.
Graphs adjusted to a common y-intercept.
36 @ oe @ 3-4
@ ——-—-@ 3-41
32 a.—.—. a 6-8
eo 6-22
28
24
RELATIVE CONDUCTIVITY (%)
TEMPERATURE (°C)
Figure /.-- February 1984 injury (Relative Conductivity) to needle tissues
versus temperature treatment for each of four families growing in the 1973
plantation environment. Graphs adjusted to a common y-intercept.
366
36 Cpshscenern=oe @ 3-4
e-——-—-@ 3-41
32 a.—.—.«a 6-8
e———8 6-22
RELATIVE CONDUCTIVITY (%)
TEMPERATURE (°C)
Figure 8.--February 1984 injury (Relative Conductivity) to needle tissues
versus temperature treatment for each of four families growing in the 1972
plantation environment. Graphs adjusted to a common y-intercept.
36 ( peoseonoeenen @ 3-4
@—-—— —-@ 3-41 ee
32 46-8 oy
@————_8 6-22 a
RELATIVE CONDUCTIVITY (%)
0 -5 -10 “15 -20 -25 -30 -35
TEMPERATURE (°C)
Figure 9.--February 1984 injury (Relative Conductivity) to needle tissues
versus temperature treatment for each of four families growing in the nursery
environment. Graphs adjusted to a common y-intercept.
367
e
x
t
are sha
Mas
mn o.
: rT = au t hun
rue t ee By Pagar y ev lvos &
, % a ee
A tea 5, oy goene? ;
Ad A ” SUE APE ole } b tet
; <i -_ »'S t 4
x
N\ .
\,
—_
* ~~
au
‘ if
si &
io Fat
en ‘ aN i oo 7 Per svat ; oa
ee & P ‘ ‘
a A WN A OE HP) is
in vibe tpt
et aaa
Mio he ae ee
ey bo tots
‘ near iD
POSTERS
ADVENTIVE EMBRYOGENESIS IN YELLOW-POPLAR TISSUE CULTURES:
A PRELIMINARY REPORT
S. A. Merkle, H. E. Sommer and C. L. BEowne
Among hardwood forest tree species, there have been only a few
reports of adventive embryogeny. We initiated cultures of yellow-poplar
(Liriodendron tulipifera) from embryos from seeds collected from a
single tree at two week intervals between mid-August and late
October, 1984. Proembryogenic nodules developed from the explants 5-6
weeks after cultures were transferred to a medium containing 2,4-D and
6BA. Within a month following transfer of proembryogenic nodules to a
hormone-free medium supplemented with casein hydrolysate, embryoids
differentiated. Prembryogenic cultures have also been grown in
suspension culture and produced embryoids. All four cultures that
produced embryoids originated from immature embryos from seed collected
during the last week of August and the first week of September.
Although most embryoids appeared abnormal, those with well-formed
cotyledons and radicles were capable of developing into normal
plantlets.
poerdocreral Associate, Associate Professor and Professor
Emeritus, School of Forest Resources, University of Georgia, Athens, GA
30602.
369
DIFFERENCES IN SEED PROPERTIES AMONG RECIPROCAL CROSSES OF
LOBLOLLY PINE
(A preliminary report)
Thomas O. Perry and Theodore H. Shear
ABSTRACT
In order to determine the relative effects of seed and pollen parents on
seed properties and initial germination and growth of loblolly pine seedlings,
selected reciprocal crosses among 14 loblolly pine clones were made.
Measurements of the total seed weights, seed coat and gametophyte weights, and
the percent of filled seeds are reported in this display.
Both the seed parent and the pollen parent affected the weight of the
seed. Statistical analyses indicated that the seed parent accounted for 66%
of the variation in total seed weight and the pollen parent accounted for an
additional 15%. Some clones were particularly poor seed parents while they
functioned well as pollen parents.
Planned germination studies and progeny tests will determine the
differences in progeny performance among reciprocal crosses. In the
meanwhile, tree improvement workers should be aware that there are large and |
significant differences in the quality of seeds between a cross of A x B and
its reciprocal B x A.
370
VARIATION IN FUSIFORM RUST STEM GALLS ON FIVE- AND SIX-YEAR-OLD SLASH PINES
Charles H. Walkinshaw!
Infection of slash pines (Pinus elliottii Engelm. var. elliottii) by Cronartium
quercuum (Berk.) Miyabe ex Shirai f. sp. fusiforme produces many identifiable
external symptoms as well as internal damage. This study classifies symptoms
on galled 5- and 6-year-old slash pine stems. Observations were made in
progeny tests near Gulfport, Mississippi; Greenwood, Florida; and Savannah,
Georgia. Data and photographs of approximately 20 families and 5 replications
per location serve as the basis for this poster.
Galls on apical dominant stems are thin, one-sided, small, typical, or fat.
The first three of these five types appear inocuous unless they develop into
constricted galls. Some galls cause a loss or reduction of apical dominance
and stimulate the production of auxiliary stems. These are stem to branch,
branch to stem, double stems, or multiple stems (often termed witches
brooms). Trees with auxiliary stems may split in windstorms and generally
die 2-5 years following infection. By the fifth growing season 23% of
rust-affected trees have formed witches brooms and another 22% have grown
auxiliary stems while retaining a dominant main stem. Galls on dominant or
abnormal stems can kill affected slash pines by girdling the cambium.
Constriction and girdling of pine stems have long been associated with wounding
by mechanical means, incompatibility of scion and stock, and infections such as
fusiform rust. Stem tissues above the constriction increase to form an
elongated upward taper. Constrictions may form in thin, typical or fat galls;
the cambium is destroyed within the gall. Incidence of constricted stem galls
in the progeny tests was 0 to 27% for control-pollinated families.
These preliminary results indicate that infected progeny from different
parents exhibit large variations in external symptoms. Families M-601 and
C-115 in the Florida test had similiar percentages of infected trees.
However, their percentages with branch galls were 80 and 46, respectively.
Incidence of witches brooms was 25 and 38%, and mortality was 19 and 547,
respectively. Thus, disease severity was significantly different in these two
families. Disease severity is governed by type of stem gall and the age of
the tree at time of infection. Thin, one-sided, and small stem galls appear
to cause minimal tree distortion unless they become constricted galls. Galls
which cause loss of apical dominance ruin the affected trees for forest
products.
Knowledge of gall types may have application in selecting for rust resistance
in slash pine. A family which forms a high percentage of thin, one-sided or
small galls on dominant stems has the greatest chance of survival. Typical
and fat stem galls, and especially those showing constriction reduce height
growth. Once this occurs, mortality usually follows within one or two growing
seasons.
lprincipal Plant Pathologist, USDA, Forest Service, Southern Forest
Experiment Station, P.O. Rox 2008, GMF, Gulfport, MS 39505.
371
sehoe
tale: oh, ae +07 Hoke iy
, rgbloch ee bmi if ene e e
“4 f esse ‘y cy
APPENDIX
SOUTHERN FOREST TREE IMPROVEMENT COMMITTEE
Membership List
May 5, 1985
GROUP A
Representatives Appointed for 6-year term
Term Ends
(at end of
Committee Membership Annual meeting) Re presenting
E. C. Burkhardt 1989 Hardwood Research Council
Route 12, Box 92
Vicksburg, MS 39180
(601)638-2305
Marvin Zoerb 1989 Pulp and Paper Industry
Union Camp Corp.
Pe O) BOX 2:16
Rincon, GA 31326
(912)826-5556
Siroos T. Jahromi 1986 Pulp and Paper Industry
International Paper Co.
Southlands Experiment Forest
Riel BOxXe5)7 |
Bainbridge, GA 31717
(912)246-3642
Dave Canavera 1986 Pulp and Paper Industry
Westvaco
Summerville, SC 29483
(803)871-5000
Mr. Garner Barnum 1988 State Forestry Agencies
Management Forester
Arkansas Forestry Commission
P. O. Box 4523, Asher Station
3821 W. Roosevelt Rd.
Little Rock, AR 72214
(501) 664-2531
Mr. C. R. Nichols 1986 State Forestry Agencies
Asst. State Forester
South Carolina Forestry Comm.
Pen Os Box 21707
Columbus, SC 29221
(803) 758-6900
373
Term Ends
(at end of
Committee Membership Annual meeting) Representing
James Bright 1988 State Forestry Agencies
Tree Improvement Forester
Mississippi Forestry Comm.
908 Robert E. Lee Bldg.
Jackson, MS 39201
(601)359-1386
Charles G. Tauer 1989 Forestry Schools
08 AG Hall
Oklahoma State University
Stillwater, OK 74074
(405) 624-5462
Bruce Bongarten 1986 Forestry Schools
School of Forest Resources
Univ. of Georgia
Athens, GA 30602
(404)542-7247
Sam B. Land, Jr. 1988 Forestry Schools
Dept. of Forestry
Mississippi State Univ. ,
P. O. Drawer FR
Mississippi State, MS 39762
(601)325-2946
Sam Foster 1986 Southern Forest Products
Crown-Zellerbach Corp. Assoc.
Southern Timber Division
Box 400
Bogalusa, LA 70427
(504) 735-4602
GROUP B
Representatives Appointed for Indefinite Terms
John Kraus U. S. Forest Service
USDA Forest Service
Southeastern For. Exp. Sta.
Georgia Forestry Center
Route 1, Box 182A
Dry Branch, GA 31020
(912)744-0266
374
Term Ends
(at end of
Committee Membership Annual meeting) Representing
O. O. Wells
Institute of Forest Genetics U. S. Forest Service
P. O. Box 2008, GMF Southern For. Exp. Sta.
Gulfport, MS 39503
(601)864-3972
Clark Lantz U. S. Forest Service
Nursery/Tree Improvement Specialist State & Private Forestry
U. S. Forest Service
Suite 300
1720 Peachtree Road, N.W.
Atlanta, GA 30356
(404) 881-3551
GROUP C
Specialists Appointed for 6-year Term
O. O. Wells 1984 Racial Variation
Institute of Forest Genetics
P. O. Box 2008, GMF
Gulfport, MS 39503
(601) 864-3972
Tom Miller, IPM 1987 Pathology
School of Forest Resources
University of Florida
Gainesville, FL 32611
(904) 392-4826
Harry O. Yates 1984 Entomology
Southeastern For. Expt. Sta.
Carlton Street
Athens, GA 30601
(404) 546-2467
Floyd Bridgwater 1987 Pollen Management
N. C. State University
Box 8002
Raleigh, NC 27650-8002
(919) 737-3168
375
Term Ends
(at end of
Committee Membership
Warren Nance 1990
U.S.D.A. - Forest Service
Southern For. Expt. Sta.
Forestry Sciences Lab.
Box 2008
Gulfport, Mississippi 39503
Ralph Mott 1990
Botany Dept.
Box 7612
North Carolina State Univ.
Raleigh, N.C. 27650-7612
GROUP D
James L. McConnell
U. S. Forest Service
Suite 828
1720 Peachtree Road, N.W.
Atlanta, GA 30309
(404) 881-3846
J. P. van Buijtenen
Forest Genetics Lab
College Station, TX 77843
G73)" 845-1325"
Tim White
School of Forest Resources
and Conservation
University of Florida
Gainesville, FL 32611
(904) 392-1850
Bob Weir
Forestry Department
Box 8002
Raleigh, NC 27650-8002
(919) 737-3168
376
annual meeting)
Representing
Stand Dynamics
Forest Biotechnology
U.S. Forest Service
Regional Office Region 8
Western Gulf Regional Tree
Improvement Cooperative
University of Florida Tree
Improvement Cooperative
North Carolina State Tree
Improvement Cooperative
SOUTHERN FOREST TREE IMPROVEMENT CONFERENCE
Abbott, Jerry E.
Kirby Forest Industries, Inc.
25 5 Iker S/7/
Silsbee, TX 77/06
(409) 385-5201
Adams, John C.
School of Forestry
Louisiana Tech University
Ruston, LA 71270
Allen, Robert M.
Dept. of Forestry
Clemson University
Clemson, SC 29631
(803) 656 3302
Arnold, Billy G.
Georgia Kraft Co.
Genetics Dev. Sec.
P. O. Box 296
Greensboro, GA 30642
(404) 485-8363
Askew, Dr. George R.
Clemson University
P. 0. Box 596
Georgetown, SC 29442
(803) 546-4402
Bailian, Li
Dept. of Forestry
NC State University
709 Latta St.
Raleigh, NC 27607
Barbour, Jill
Owens Illinois, Inc.
P. O. Drawer A
White Springs, FL 32096
(904) 397-2171
Barnum, Garner
Arkansas Forestry Commission
P. O. Box 4523, Asher Station
Little Rock, AR 72214
(501) 664-2531
Barras, Dr. Stanley
USDA-Forest Service
RM T-10210 701 Loyola Ave.
New Orleans, LA 70113
(504) 589-3003
May 21-23, 1985
Registrants
Beineke, Walter
Dept. of Forestry Purdue
Purdue University
West Lafayette, IN 47907
(317) 494-3611
Benson, James D.
Dept. of Forestry
Clemson University
Lehotsky Hall
Clemson, SC 29631
(803) 656-3302
Bartlum, Guy N.
St. Joe Paper Co.
Rt 1 Box 70
Lamont, FL 32336
(904) 997-0526
Blanton, Steve A.
Alabama Forestry Commission
Rt 1 P.O. Box 88-B
Thorsby, AL 35171
Blush, Ton
Westvaco
P. O. Box 1950
Summerville, SC 29484
(803) 871-5000 Ext 547
Boltz, Brett
School of Forest Resources
University of Georgia
Athens, GA 30602
(404) 542-7247
Bongarten, Bruce
School of Forest Resources
University of Georgia
Athens, GA 30602
(404) 542-7247
Bramlett, David L.
U.S. Forest Seravice
Rt 1 Box 182A
Dry Branch, GA 31020
(919)744-0261
Bright, Jim
Mississippi Forestry Commission
908 Robert E. Lee Bldg
Jackson, MS 39201
(601) 359-1386
377
Bridgewater, Floyd
School of Forest Resources
NC State University
P. O. Box 8002
Raleigh, NC 27695
Brooks, Terrell
Georgia Forestry Commission
P. O. Box 819
Macon, GA 31298
(912) 744-3369
Brouard, John
NC State University
201 B Dixie Trail
Raleigh, NC 27607
(919) 737-3420
Brown, Raymond
Champion International
P~ 0.) ‘Box: 1937:
Tillery, NC 27887
(919)826-4182
Bryan, Harry
Hiwassee Land Company
Chatsworth, LA 30705
(404) 334-2422
Bryant, Richard
International Paper Company
Re 1 Box ‘S71
Bainbridge, GA 31717
(912) 246-3642
Buckles, Oliver W.
U.S. Forest Service
Pe) (0%, ‘Box 7/57.
Moncks Corner, SC 29461
(803) 336-3246
Buford, Marilyn
Virginia Polytechnic Inst.
Department of Forestry
Blacksburg, VA 24061
(703) 961-6603
Byram, Tom
Texas Forest Service
Forest Science Laboratory
College Station, TX 77843
(409) 845-2556
Canavera, David
Westvaco
P. O. Box 1950
Summerville, SC 29484
(803) 871-5000
Chase, Charles D.
St. Joe Paper Company
Rt 1 Box 70
Lamont, FL 32336
(904) 997-0526
Coleman, Stephen
Boise Cascade
P. O. Box 45
Singer, LA 70660
(318) 463-9681
Comer, C. W.
University of Florida
118 N-Z Hall
Gainesville, FL 32611
(904) 392-4851
Cox, Russeil A.
Tennessee Div of Forestry
P. O. Box 120
Pinson, TN 38366
(901) 988-5221
Crane, Barbara S.
Georgia Kraft Company
Genetics Dev. Sec.
Greensboro, GA 30642
(404) 485-8363
Cunningham, Michael
NC State University
School of Forest Resources
Box 8002
Raleigh, NC 27695
(919) 737-3420
Davison, R. Marc
Champion International
P. O. Box 97
Tillery, NC 27887
(919) 826-4182
Debarr, Gary L.
USDA-Forest Service
Cariton Street
Athens, GA 30602
(404) 546-2467
Dewald, Laura E.
Virginia Polytechnic Institute
and State University
228 Cheatham Hall
Blacksburg, VA 24061
(703) 961-5300
Dinus, Dr. Ronald J.
International Paper Company
P. 0. Box 797
Tuxedo Park, NY 10987
(914) 351-2101
Duba, Dr. Stuart
Auburn University
School of Forestry
Auburn University, AL 36849
(205) 826-4050
Dvorak, William S.
Camcore Cooperative
NC State University
P. O. Box 8007
Raleigh, NC 27195
(919) 737-2738
Einspahr, Dean W.
Institute of Paper Chemistry
P. O. Box 1039
Appleton, WI 54912
(414) 734-9241
Ext 282
Emery, Bruce
NC State University
2829 Mayview Rd
Raleigh, NC 27607
(919) 737-3168
Feret, Dr. Peter P.
Dept of Forestry
Virginia Polytechnic Institute
and State University
Blacksburg, VA 24060
(703) 961-5943
Fins, Dr. Lauren
University of Idaho
Inland Empire Tree
Improvement Coop
Dept of Forest Resources
Moscow, ID 83843
Foster, G. Sam
Crown Zelierbach Corp.
Southern Timber Division
Box 400
Bogalusa, LA 70427
(504) 735-4602
Frampton, L. John
School of Forest REsources
NC State University
P. O. Box 8002
Raleigh, NC 27695-8002
(919) 737-3168
378
Gates, James
USFS Kisatchie Natl Forest
2500 Shreveport Hwy
Pineville, LA 71360
(318) 473-7123
Gill, John J.
Georgia Kraft Company
P. O. Box 108
Rome, GA 30161
Goddard, Ray
University of Florida
School of Forest Res. & Conserv.
Gainesville, FL 32611
(904) 3921850
Greenwood, Michael S.
College of Forestry
Nutting Hall
University of Maine
Orono, ME 04469
(207) 581-2838
Griggs, Margene M.
U.S. Forest Service
P. O. Box 2008, GMF
Gulfport, MS 39505
(601) 864-3972
Guiness, Bill
Catawba Timber Company
Box 128
Catawba, SC 29704
(803) 329-6653
Hamlin, Jim
U.S. Forest Service
P. O. Box 1008
Roseburg, OR 97470
(503) 672-6601
Handley, Lee
Westvaco
Box 1950
Summerville, SC 29484
(803) 871-5000
Harbin, Michael C.
Chesapeake Corp
Box 311, 19th & Main St
West Point, VA 23181
(804) 843-5652
Hart, Joel David
Mississippi State Univ
P. O. Drawer FR
Mississippi State, MS 39762
Hartwell, Dianna Lynn
NC State University
2834 Mayview Rd
Raleigh, NC 27607
(919) 737-3420
Hatcher, Alice
School of Forest Resources
NC State University
P. O. Box 8002
Raleigh, NC 27695-8002
(919) 737-3168
Hendrickson, John A.
Scott Paper Company
P. O. Box 899
Saraland, AL 36571
(205) 675-2932
Hodge, Gary
NC State University
School of Forest Research
Box 8002
Raleigh, NC 27695
(919) 737-3420
Hodges, James F.
Brunswick Pulp Land Company
Box 860
Brunswick, GA 31521
(912) 265-5780
Hollowell, Alfred
Weyerhaeuser Comapny
P. O. Box 1060
Hot Springs, AR 71902
Hood, Jim
Ontario Ministry of Natural Res
Forest Res. Branch, Queens Park
Toronto, Ontario L4C 1C6 Canada
(416) 963-1334
Hopkins, Sam
Gulf States Paper Corporation
P. O. Box 3199
Tuscaloosa, AL 35404
(205) 553-6200
Huang, F. H.
University of Arkansas
316 Plant Science Bldg
Fayetteville, AR 72701
(501) 575-2603
Hutto, Erika M.
Alabama Forestry Commission
Rta2e P.O. Box 151
Kinston, AL 36453
Isik, Kani
University of California
Forestry Department
Mulford Hall 145
Berkeley, CA 94720
Jahromi, Siroos T.
International Paper Coampny
Southlands Experimental Forest
Rt 1 Box 571
Bainbridge, GA 31717
(912) 246-3642
Jernigan, Richard
Georgia Forestry Commission
P. O. Box 819
Macon, GA 31298-4599
(912) 744-3369
Jett, J. B.
School of Forest Resources
NC State University
P. O. Box 8002
Raleigh, NC 27695-8002
(919) 737-3168
John, Sally
NC State University
201B Dixie Trail
Raleigh NC 27607
(919) 737-3420
Dr. Jon Johnson
Department of Forestry
University of Florida
118 N-Z Hall
Gainesville, FL 32611
(904) 392-4851
Jurado-Blanco, Juan B.
NC State University
P. O. Box 5716
Raleigh, NC 27650
(919) 737-2739
Kanashiro, Milton
NC State University
School of Forest Resources
Box 8002
Raleigh, NC 27695
(919) 737-3420
Karrfalt, Robert
USDA Forest Service
Rt 1 Box 182 B
Dry Branch, GA 31020
(912) 744-3314
379
Kitchens, Frank
School of Forestry
Branch, GA 31020
(912) 744-0266
Kitchens, Robert
USFS Timber RO
1720 Peachtree Rd NW
Atlanta, GA 30367
(404) 881-4039
Kok, H. R.
University of Florida
118 Newins-Ziegler Hall
Gainesville, FL 32611
(904) 392-1850
Kolb, Tom
Penn State University
302-D Forest Resources Lab
University Park, PA 16802
(814) 865-2130
Kormanik, Paul P.
SE Forest Exp Station
Carlton Street
Athens, GA 30602
(404) 546-2435
Kossuth, Dr. Susan V.
USDA Forest Service
1143 Fifield Hall
University of Florida
Gainesville, FL 32611
(904) 371-4360
Kraus, John F.
Southeastern Forest Exp. Station
Rt 1 Box 182-A
Dry Branch, GA 31020
(912) 744-0266
Kung, Dr. Fan H.
Department of Forestry
Southern Illinois University
Carbondale, IL 62901
(618) 453-3341
Lafarge, Timothy
USDA Forest Service
1720 Peachtree Rd NW
Atlanta, GA 30367
(404) 881-4046
Lambeth, Clements C.
Weyerhaeuser Company
900 Whittington Ave.
Hot Springs, AR 71902
(501) 624-8516
Land, Samuel B., Jr.
Mississippi State University
P. O. Drawer FR
Mississippi State, MS 39762
(601) 325-2946
Lantz, Clark W.
USDA Forest Service
1720 Peachtree Rd NW
Atlanta, GA 30367
(404) 881-3554
Layton, Patricia
Building 1503
Oak Ridge National Laabs
P. O. Box X
Oak Ridge, TN 37831
(615) 574-7364
Ledig, Thomas
Institute of Forest Genetics
U.S. Forest Service, PSW
Box 245
Berkeley, CA 94701
(415) 486-3458
Leach, Greg
Chamption International Corp.
Pe Os Box 87
Cantonment, FL 32533
(904) 968-6614
Lesney, Mark
University of Florida
Dept. of Forestry
118 N-Z Halil
Gainesville, FL 32611
(94) 392-4851
Lewis, Ralph A.
USDA Forest Service
380 Rock Meadow Dr.
Stone Mountain, GA 30367
(404) 881-4664
Libby, Dr. Bill
University of California
Dept. of Forestry & Conserv.
Mulford Hall
Berkeley, CA 94720
Lowe, William J.
Texas Forest Service
Forest SCience Lab
College Station, TX 77843
(409) 845-2523
Lowerts, George A.
Union Camp Corporation
Bo 0. Box) 216
Rincon, GA 31326
(912) 826-5556
Mabry, Will
Catawba Timber Co.
Box 128
Catawba, SC 29704
(803) 329-6653
Manchester, Edwin H.
USDA Forest Service
201 Woodland Dr.
Murphy, NC 28906
(704) 837-5152
Manley, David
International Paper Co.
Rt 3), Box 312—B
Natchez, MS 39120
(601) 442-7421 Ext 583
Massie, William E., Jr.
Mississippi Forestry Commission
P. O. Bos 468
Lumberton, MS 39544
(601) 796-8892
_ McConnell, James L.
U.S. Forest Service
1720 Peachtreee Rd
Atlanta, GA 30367
(404) 881-4045
McCutchan, Barbara G.
Westvaco
P.O. Box 1950
Summerville, SC 29484
(893) 871-5000
McKeand, Steve
School of Forest Resources
NC State University
P. O. Box 8002
Raleigh, NC 27695-8002
(919) 737-3168
McKinley, Craig R.
Texas Forest Service
Texas A&M University
Caollege Station, TX 77840
(409) 845-5078
Merkle, Scott A.
University of Georgia
School of Forest Resources
Athens, GA 30602
(404) 542-6112
Miller, Dot
Texas Forest Service
Route 1 P. O. Box 160
Hillister, TX 77624
380
Miller, Larry
Temple-Eastex, Inc.
229 N. Bowie
Jasper, TX 75951
(409) 384-3434
Morrow, Dan
Dept of Natural Resources
Buckingham Forest
Tree Nursery
Harmans, MD 210/77
(301) 768-7367
Mott, Ralph L.
NC State University
Dept of Botany
Raleigh, NC
737-3525
Moye, Jim
Alabama Forestry Commission
Route 2
Kinston, AL 36453
Murray, Gordon
Canadian Forestry Service
Petawawa Natl Forestry Inst.
Chalk River, Ontario
Canada KOJ 1JO
(613) 589-2880
Neal, Ronald B.
American Can Company
P. O. Box 315
Butler, AL 36904
(205) 459-3002
Nelson, Neil D.
Biotechnology Program Leader
NC Forest Experiment Station
Forestry Sciences Laboratory
Box 898
Rhinelander, WI 54501
(715) 362-7474
Newton, R. J.
Dept of Forest Science
Texas A&M University
College Station, TX 77843
(409) 845-8279
O'Gwynn, Claude H.
Weyerhaeuser Company
208 S. Moore Road
Hot Springs, AR 71913
(501) 767-7255
Padgett, Bill
Alabama Forestry Commission
513 Madison Avenue
Montgomery, AL 36130
€205)) 261-2532
Pait, John
Container Corp. of America
Box 626
Callahan, FL 32011
(904) 879-3051
Park, Yill Sung
Canadian Forestry Service
P. O. Box 4000
Fredericton, NB E3B 1P4 Canada
(506) 452-3585
Platt, Adlai
Champion International Corp
P. 0. Box 834
Newberry, SC 29108
(803) 276-5529
Post, Boyd W.
Coop State Research Service
J. S. Morrill Bldg
15th & Independence
Washington, D.C. 20251
(202) 447-2016
Powell, Gregory L.
University of Florida
118 Newins-Ziegler Hall
Gainesville, FL 32611
(904) 392-1850
Ranney, J. W.
Oak Ridge National Lab
Box X
Oak Ridge, TN 37831
(615) 574-7364
Reddy, K. V.
University of Florida
118 N-Z Hall
Department of Forestry
Gainesville, FL 32611
(904) 392-1850
Reighard, Greg
University of Florida
118 Newins-Ziegler Hall
Gainesville, FL 32611
(904) 392-1850
Rhinehardt, George
International Paper Company
Bellamy Seed Orchard
Rt 3 Box 405
Marianna, FL 32446
(904) 594-6001
Rockwood, D. L.
University of Florida
118 Newins-Ziegler Hall
Gainesville, FL 32611
(904) 392-1852
Rounsaville, Marc
USFS
P. O. Box 248
Wiggins, MS 39577
(601) 928-4792
Rousseau, Randall
Westvaco
221 Lansing Dr.
Paducah, KY 42001
(502) 335-3156
Ruzin, Dr. Steven E.
Plant Genetics, Inc.
1930 Fifth St.
Davis, CA 95616
(916) 753-1400
San Fratello, Guy
SC Forestry Commission
P. O. Box 157
Tillman, SC 29943
726-3845
Schlarbaum, Scott E.
Dept of Forestry, Wildlife
and Fisheries
University of Tennessee
Knoxville, TN 37901-1071
(615) 974-7126
Schmidtling, Ron
U. S. Forest Service
P. O. Box 2008, GMF
Gulfport, MS 39505
(601) 864-3972
Schoenihe, Roland E.
Clemson University
Department of Forestry
Clemson, SC 29631
(803) 656-3302
Schultz, Emily B.
Mississippi State University
School of Forest Resources
P. O. Drawer FR
Mississippi State, MS 39762
(601) 325-2946
Sederoff, Ron
Forest Genetics, PSW
P. O. Box 245
Berkeley, CA 94701
(415) 486-3134
Shaw, Douglas U.
International Forest Seed Co.
P. O. Box 290
Odenville, AL 35120
(205) 323-0150
381
Shear, Ted
5016 Newcastle Road
Raleigh, NC 27606
(919) 851-0374
Shimzu, Jarbas Yukio
Embrapa-Centro Nacional De Presq.
De Florestas
Caixa Postal 3319
8000 Curitiba-Parana, Brazil
Sluder, Earl R.
SE Forest Experiment Station
Rt 1 Box 182A
Dry Branch, GA 31020
(912) 744-0266
Smeltzer, Richard H.
International Paper Company
Route 3, Box 312-B
Natchez, MS 39120
(601) 442-7421 Ext 587
Snow, Glenn A.
U. S. Forest Service
P. O. Box 2008, GMF
Gulfport, MS 39505
(601) 864-8256
Sommer, Harry E.
School of Forest Resources
University of Georgia
Athens, GA 30602
(404) 542-2535
Sprague, Jerry
School of Forest Resources
NC State University
P. O. Box 8002
Raleigh, NC 27695-8002
(919) 737-3168
Squillace, A. E.
School of For. Res. & Conserv.
University of Florida
N-Z Hall, Rm 118
Gainesville, FL 32611
(904) 392-1850
Stauder, Albert
Texas Forest Service
P. O. Box 617
Alto, TX 75925
(409) 858-4202
Steele, James A.
Temple-Eastex Forests
229 N. Bowie
Jasper, TX 75951
(409) 382-3434
Steigerwalt, Mark
KMI Land Resources
P. O. Box 7038
Garden City, GA 31418
(912) 964-1871
Stelzer, Henry
Dept. of Forestry
Perdue University
West Lafayette, IN 47907
(317) 494-3613
Stonecypher, Roy W.
Weyerhaeuser Company
534 N. Tower Street
Centralia, WA 98531
(206) 736-8241
Stubbs, Olin L.
Louisiana Office of Forestry
P. O. Box 1628
Baton Rouge, LA 70821
(504) 925-4507
Studyvin, Charles
-U.S. Forest Service
Pi. 0. Box 255
Mt. Ida, AR 7197
(501) 867-2101
Stuhlinger, Chris
LSU
School of Forestry, Wildlife
& Fisheries
Baton Rouge, LA 70803
Talbert, Cheryl B.
Weyerhaeuser Company
WIC-2H2 Technology Center
Tacoma, WA 98477
(206) 924-6991
Tauer, Charles G.
Oklahoma State University
013 S Agriculture Hall
Stillwater, OK 74078
(405) 624-5462
Thielges, Bart A.
University of Kentucky
Department of Forestry
Lexington, KY 40546-0073
(606) 257-7596
Todhunter, Michael N.
International Paper Company
Southlands Experiment Forest
Bainbridge, GA 31717
(912) 246-3642
Tolliver, Dr. John
LSU School of Forestry
Baton Rouge, LA 70803
(504) 388-4131
Turner, Gene F.
NC Div of Forest Resources
2411 Garner Rd
Clayton, NC 27529
(919) 553-6178
Tuskan, Gerald A.
North Dakota State Univ.
Dept. Horticulture & Forestry
Fargo, ND 58105
(701) 237-8476
Van Buijtenen, J. P.
Forest Genetics Lab.
Texas Forest Service
College Station, TX 77843
(409) 845-1325
Walkinshaw, Charles H.
USDA Forest Service
Southern Forest Expt Stn
P. O. Box 2008, GMF
Gulfport, MS 39505
(601) 864-8256
Waxler, Michael S.
Weyerhaeuser Company
P. O. Box 1060
Hot Springs, AR 71902
(50h) 624-8494
Weir, Robert J.
School of Forest REsources
NC State University
P. O. Box 8002
Raleigh, NC 27695-8002
(919) 737-3168
Wells, Ozzie
Southern Forest Expt Stn
P. O. Box 2008, GMF
Gulfport, MS 39505
(601) 864-3972
White, Tim
University of Florida
Gainesville, FL 32611
Wilder, Jean
Louisiana State University
P. 0. Box 22809
Baton Rouge, LA 70893
(504) 388-4131
382
Williams, Claire G.
North Carolina State University
P. O. Box 8002
Raleigh, NC 27695
(919) 737-3420
Williford, Mike
American Can Company
P. O. Box 315
Butler, AL 36904
(205) 459-3007
Wiselogel, Arthur
Texas A&M University
Forest Science Dept
College Station, TX 77843
(409) 845-5004
Wright, Lynn
Oak Ridge National Lab
Bldg 1503
Oak Ridge, IN 37831
(615) 574-7378
Yates, Harry O. III
Forestry Sciences Laboratory
USDA Forest Service
Carlton Street
Athens, GA 30602
(404) 546-2467
Young, Mike
Georgia Forestry Commission
P. O. Box 819
Macon, GA 31298-4599
(912) 744-3369
Zeringue, Mr. Furcy
LSU School of Forestry
Baton Rouge, LA 70803
(504) 388-4131
Zoerb, Marvin H., Jr.
Union Camp Corp
Box 216
Rincon, GA 31326
(912) 826-5556