THE GENETIC AND ENVIRONMENTAL BASIS OF
FRUITFULNESS AND GROWTH IN LOBLOLLY PINES
RONALD CARL SCHMIDTLING
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I am indebted to Dr. Ray Goddard, chairman of my graduate
committee, for his many suggestions and reviews of the manuscript.
I am also appreciative for the help I received from the other
members of my graduate committee, Drs. A. E. Squillace, Charles
Hollis, E. S. Horner, and W. L. Pritchett, for their helpful
I also gratefully acknowledge help I have received from U.S.
Forest Service research technicians Norm Scarbrough, Horace Smith,
Herschel Loper, and Victor Davis. Sincere appreciation is also
extended to past and present tree improvement personnel of the
Southern Region of the U.S. Forest Service, Johnsey King, Will
Schowalter, and Jim McConnell.
TABLE OF CONTENTS
I GENETIC AND ENVIRONMENTAL VARIATION IN
FRUITFULNESS IN A LOBLOLLY PINE SEED ORCHARD..... 1
Literature Review............................ 2
Genotypic Variation..................... 3
Environmental Variation................... 5
Materials and Methods......................... 11
Results and Discussion....................... 16
Genetic Variation.......................... 16
Environmental Variation................... 41
Genetic Variation......................... 61
Environmental Variation.................... 63
II INHERITANCE OF PRECOCITY IN LOBLOLLY PINE
AND ITS RELATION TO GROWTH....................... 65
Literature Review............................ 66
Materials and Methods......................... 70
Results and Discussion....................... 76
SUMMARY AND CONCLUSIONS.......................... 92
Genetic Variability........................... 92
Environmental Variability.................... 93
LITERATURE CITED................................. 95
Abstract of Dissertation Presented to the Graduate Council of the
University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
THE GENETIC AND ENVIRONMENTAL BASIS OF
FRUITFULNESS AND GROWTH IN LOBLOLLY PINES
Ronald Carl Schmidtling
Chairman: Ray E. Goddard
Major Department: Forest Resources and Conservation
Genetic and environmental variation in fruitfulness in loblolly
pines (Pinus taeda L.) were examined in clones and seedlings.
In a young loblolly pine seed orchard in south Mississippi clonal
variation was substantial. Over 50% of the variation in female flower-
ing, cone production, and seed production, and about 40% of the variation
in male flowering was attributable to genetic effects. The broad-sense
heritability for female flowering varied over time, however, increasing
at first to about 0.6 at 8 years of age and declining thereafter. The
decrease in heritability was mainly due to a large increase in environ-
mental variation after the 8th year.
Although the more fruitful clones tended to be fruitful every year,
significant year x clone interactions were found. The top 20% of seed
producing clones was not the same every year. Year x clone interactions
in male flowering also indicated that the genetic makeup of seeds collect-
ed from year to year may vary considerably, even if collected separately
Flowering was inversely related to height and diameter growth, as
flowering was best on soils where growth was poorest. Yearly variation
in flowering also could be related to moisture stress, as a drought
during the strobilus initiation period seemed to favor abundant flower-
ing the following year. This was true of male as well as female
flowering, but the response to drought differed with respect to timing.
Male flowering was favored by an early summer drought (May-June) and
female flowering was favored by a late summer drought (July-August). It
was proposed that moisture stress exerts its effect by causing a cessa-
tion of vegetative growth coinciding with the time of flower induction,
thus allowing floral initials to form.
Genetic variation in average amount of flowering was very strong
among families in a diallel, and most of the variation was additive.
Diallel analysis and parent-progeny regressions yielded narrow-sense
heritability estimates of 0.61 and 0.52, respectively. The inheritance
in flowering at age 2, or precocity, was not as strong as that of
average flowering, being only 0.13 on an individual basis and 0.47 on
a family basis. Flowering traits were negatively correlated with growth
traits, indicating that selection for average flowering or precocity
alone would result in some growth loss. The results of a half-sib
progeny test incorporating precocious individuals, however, indicated
that selection for both precocity and growth could be profitable.
GENETIC AND ENVIRONMENTAL VARIATION IN FRUITFULNESS IN A
LOBLOLLY PINE SEED ORCHARD
The goal of any tree improvement program is the production of
genetically superior trees. At present, southern pine plantations
are established with seedlings (rather than vegetative propagules);
and the tree improvement method commonly used is to establish clonal
seed orchards by grafting scions from phenotypically superior trees
onto seedling rootstocks. Traits such as growth, form, and quality
of the parent trees selected for the production of seed in either
type of orchard, clonal or seedlings, are of prime importance. How-
ever, fruitfulness also is important because the selections or their
progeny need to produce seed in sufficient quantities for the tree
improvement program to be successful.
A related factor influencing seed quality is pollen production
of the orchard trees. Breeders assume that most of the pollen effect-
ing fertilization in orchards will originate within the orchard. If
not, the realized genetic gain will be greatly reduced. Very little
is known about pollen production, for a major part of genetics research
has been done on female flowering and seed production. Although it
has been assumed that pollen from within the orchard will be adequate
for good pollination, this has not been so in many cases, perhaps
because conditions which were optimum for seed production may not have
been optimum for pollen production. Very small amounts of pollen are
required for controlled crosses in progeny tests. Often even these
small quantities are unavailable from some clones, making completion
of progeny tests very difficult.
Thus a major problem in tree improvement programs is the develop-
ment of an understanding of the inherent and environmental conditions
contributing to variation in flowering, pollination, and seed produc-
tion in southern pine orchards.
Flowering and seed yield data collected from a loblolly pine (Pinus
taeda L.) seed orchard over a period of several years were examined in
this report. The objectives of the study were to:
1. determine the extent of genetic control of male and female
flowering and their intercorrelations;
2. determine the year-to-year consistency of clonal differences
in flowering to evaluate the importance of yearly variation
in genetic worth of the seed produced and to examine the pre-
dictive value of early flowering data;
3. examine environmental variation in flowering and its relation-
ship to growth, to provide information on suitable conditions
for optimizing orchard output.
The emphasis of this review is on genetic and environmental effects
on flowering in conifers from an applied point of view. The role of
growth regulators, especially gibberellins (Pharis 1976), is surely
important, because they probably play a central role in floral initia-
tion, but is not directly pertinent to this research report.
IT is generally recognized that large differences exist in fruit-
fulness between southern pine clones. This is considered a serious
problem in seed production since often only a few clones in an orchard
produce most of the seed. Bergman (1968) estimated that 2 of 15 clones
produced over half of the seed in a loblolly pine orchard. Others
estimated that 20% of the clones produced 80% of the seed (N.C. State
University 1976). A less pessimistic estimate was provided by Beers
(1974) who found that the top 20% of clones in seed production produced
56% of the seed in a slash pine (P. elliottii Engelm.) orchard.
Danbury (1971) estimated that seed production could be increased by 50%
and seed cost reduced by one-third if only the most productive half of
the clones available in a radiata pine (P. radiata D. Don) seed
orchard in Australia were retained. He assumed that growth and fruitful-
ness were weakly, if at all, correlated genetically. Inherent fruitfulness
can be especially important, since as a general rule trees which are
inherently unfruitful do not respond as well to treatments as do fruitful
trees (Bergman 1968).
The inherent ability of the individual tree to flower is probably
the most important factor influencing fruitfulness (Schmidtling 1974,
Shoulders 1967). It is important to consider this variation in the
design of experiments. In several fertilizer experiments, from 33%
(Schmidtling 1974) to 56% (Schmidtling 1975) of the total variation in
fruitfulness was attributable to clonal effects, even though the treat-
ment effects were large and significant. Broad-sense heritability
estimates for fruitfulness of 0.50 for slash pine (Varnell et al. 1967)
and 0.4 to 0.7 for loblolly pine (Schmidtling 1974) reinforced this
There is a general consensus that inherent variation in male
flowering is also large, but the effects seldom are quantified. Male
strobilus production is more difficult to measure than female, because
pollen catkins do not persist after pollen is shed and counts must be
timely. They are also relatively numerous, and thus costly to count.
Barnes and Bengtson (1968) and Schultz (1971) found strong clonal effects
on male flowering in a slash pine fertilization and irrigation study.
Webster (1974) also observed strong clonal effects on male flowering
in a loblolly pine orchard.
Most studies have shown that male and female flowering were not
closely related. Stern and Gregarius (1972) found that the correlation
between male and female flowering was very weak. Schultz (1971) found
a slightly negative genetic correlation between male and female flower-
ing in slash pine. This weak relationship between male and female
flowering is an important consideration in seed orchard management. If
we follow Danbury's (1971) suggestion and rogue 50% of the clones on
the basis of seed production, we might very well eliminate some of the
best pollen producers in the orchard, subsequently narrowing the genetic
base and perhaps eliminating some of the better genotypes.
The physiological basis for genetic differences has not been
determined, though Smith and Stanley (1967) showed that high cone
producing trees had higher N concentrations in needles than low cone
producers in response to N fertilizers. Considering all the observed
genotype x treatment interactions, a simple explanation does not seem
warranted. In general, the better flowering trees responded well to
fertilizers and the poorer trees responded poorly or not at all
(Bergman 1968, Schmidtling 1974, 1975). Deviations from this have
been noted. In addition to the kinds of response noted above, it has
been observed that some relatively unfruitful trees responded well to
fertilizers and some very fruitful trees did not (Schmidtling 1974, 1975).
Beers (1974) noted large differences among clones in response to level
of fertilizer, which lead him to suggest that individual clones should
be fertilized on a "prescription" basis. It appears that the limiting
physiological factor for flowering might differ considerably by genotype.
A large volume of literature exists that relates fruitfulness in
conifers to environmental variables (primarily those induced by experi-
mental means). Reviews by Jackson and Sweet (1972) and Puritch (1972)
give comprehensive coverage of the literature up to 1972.
The complex nature of the relationship between environmental
variation, either natural or induced, and flowering is clearly evident
in these reviews and in subsequent work. Many conditions or treatments
which promote flowering in conifers are those which promote the overall
growth and vigor of the tree, such as fertilization, thinning, and
increased insolation. On the other hand, treatments such as girdling,
moisture stress, root pruning or restriction, and poor mineral nutrition
which are detrimental to growth and vigor also promote flowering.
The greatest amount of research deals with fertilization and
mineral nutrition and their relationships to flowering. Management
plans for all southern pine seed orchards now include fertilization,
especially with N, as it is assumed that this will be necessary to
induce and maintain high levels of seed production. This assumption
has a sound basis in past research.
Puritch (1972) tabulated 25 references in which fertilizers were
applied to conifers. In all but three, fertilizers stimulated female
strobilus production. All 22 of the successful experiments included N
in the fertilizer. Nitrogen appeared to be the key fertilizer component.
An impressive number of studies showed a positive response to N alone
(Smith et al. 1968, Hoist 1959, Barnes and Bengston 1968, Cayford and
Jarvis 1967, Stephens 1961, Giertych and Forward 1966, Schultz 1971,
Ebell 1967, 1972, Morris and Beers 1969, Stoate et al. 1961, Barnes
1969, Goddard and Strickland 1966, Kraus 1925, Beers 1974).
Relatively few experiments distinguished the effects of each com-
ponent in fertilizer combinations of N, P, and K; and research on
nutrients other than these three has been lacking. In an NPK factorial
experiment with slash pine, Morris and Beers (1969) found that only N
stimulated flowering. Similarly, Goddard and Strickland (1966) reported
that N was of great importance, but that P and K also had some effect
when combined with N. Phosphorous and K applied without N, however,
depressed flowering. In an N-P factorial test conducted in a loblolly
pine seed orchard, ammonium nitrate tripled female strobilus production,
P had no effect, and the N + P treatment produced the same number of
female strobili as N alone (Schmidtling 1975). On the other hand,
Giertych (1973) observed that only K, in an NPK factorial, stimulated
female strobilus production in Scots pine (P. sylvestris L.). In
1968, van Buijtenen reported that P increased flowering in loblolly
pines and that N alone or in combination with P actually depressed
flowering. Although fertilization research has strongly supported a
positive role for N, under some conditions other nutrients may be
Analyses of tissue N in conifers also has demonstrated a positive
role for N in flowering. Smith and Stanley (1967) and Smith et al.
(1968) found that good cone producers had a higher proportion of N
in their foliage than poor cone producers. Barnes and Bengston (1968)
showed that fertilizing with ammonium nitrate, which doubled female
flowering, increased N content, free arginine, and total amino acids
in twigs. However, N content per se may not be important. Ebell and
McMullan (1970) found that nitrate N and ammonium N increased foliar N
content and shoot growth in Douglas-fir (Pseudotsuga menziesii [Mirb.]
Franco) by similar amounts, but only nitrate-N increased female flower
production. Free arginine content was greater in nitrate-treated trees
than in the others and appeared to be quantitatively associated with
increased flower production.
In contrast to the voluminous research that showed increased
flowering related to fertilization and nitrogen content, several reports
have related nutrient deficiencies to enhanced flowering. Lyr and Hoffman
(1964) were able to induce flowering in Cryptomeria japonica D. Don
seedlings by inducing nitrogen deficiency. Kuo (1973) and Kamienska
et al. (1973) observed precocious flowering in N deficient Cupressus
arizonica Greene seedlings. Sweet and Will (1965) found that precocious
male flowering in radiata pine was associated with low nutrient status.
Giertych (1975) examined the mineral distribution in crowns of Scots
pine and concluded that female flowers were initiated under conditions
of mineral deficiency. Thus, it seems uncertain that mineral nutrition
is truly a limiting factor.
Thinning has been an established silvicultural procedure to increase
growth, and also has been one of the principal methods to increase cone
production. The effects of thinning on fruitfulness have been well
documented and were assumed to be associated with increased light,
moisture, and minerals. The initial response to thinning may have been
due to some kind of stress induced by an abrupt change in environment.
Change has not always been necessary, however, as spacing trials in
red pine (P. resinosa Ait.) showed that cone production was inversely
related to competition (Stiell 1971). The increased light intensity
and temperature have been the most likely candidates for thinning effects;
but it has been difficult to separate the two, since increasing insola-
tion nearly always increases temperature. The positive role of light
and temperature has been supported by climatic associations with cone
crops in conifers. Warm, sunny weather during flower induction favored
good flower crops (Eis 1973, 1976, Zasada et al. 1976, La Bastide and
Van Vredenburch 1970). Also, the distribution of cones within the
crowns of slash pines was strongly related to insolation (Smith and
Stanley 1969). Increasing temperature without increasing light was
effective in experiments involving transfer of Picea sitchensis (Bong.)
Carr. to greenhouses (Tompsett and Fletcher 1977) or covering field-
grown P. abies L. Karst. with polyethylene (Chalupka and Giertych 1977).
Giertych (1976) surmised that high air temperature favored both male and
female flowering whereas high light intensity favored only female flower-
In addition to the experimental association between vigor-increasing
treatments and flowering, increased flowering has been associated with
various other manifestations of vigor in conifers, such as tree size
(Andersson and Hattemer 1975, Schmidtling 1969), crown size (Grano 1957,
Cappelli 1958), shoot size (Varnell 1970, K. J. Lee 1978), and branch
order (Thorbjornsen 1960, Rim and Shidei 1974). Treatments which had
an obvious deleterious effect on growth and vigor also were associated
with increased flowering. For example, girdling, strangulation, and
binding were used to effect flowering. These treatments were undoubted-
ly stressful, as girdled shoots or trees often died. The immediate
effects were not much different from increased insolation, however, as
carbohydrate accumulated above the girdle (Ebell 1971, Hashizume 1970).
Root pruning or root restrictions and moisture stress have also
increased flowering, but linking these treatments to carbohydrate accumu-
lation was difficult. Root pruning (subsoiling) done in June reduced
the current year's diameter growth, but increased both male and female
flowering in Virginia pines (P. virginiana Mill.) (Greenwood and
Schmidtling 1980). Similarly, transplanting red pines into pots and
continued confinement greatly increased male and female flowering
(Quirk 1973). Confinement in pots has a well-known negative effect
on growth, even if water and nutrients are optimum.
Drought, although obviously deleterious to growth, has been shown
to promote fruitfulness, especially if it occurs approximately at the
time of floral initiation. Matthews (1963) listed a number of refer-
ences to climatic studies which have shown this to be true in many
woody plant species, and experimental evidence in loblolly pines was
provided by Dewers and Moehring (1970).
Shoulders (1973) found that abundant rainfall early in the year
followed by a drought increased flowering in slash pines. He proposed
that the early rains favored luxuriant vegetative growth, thereby
increasing photosynthetic capacity. Mild moisture stress in summer
curtailed further vegetative growth, allowing the accumulation of
carbohydrate necessary for floral initiation. This is an attractive
hypothesis, as it can be used to explain how treatments which increase
vigor and those which decrease vigor can both have promotive effects
Greenwood (1978) believed that the presence of a "quiescent" bud
is necessary for an extended period of time during the growing season
to form strobilus initials, and that one of the reasons seedlings do
not flower is that growth is continuous and a quiescent bud does not
exist. He induced bud quiescence by shortening photoperiod and lowering
temperature in early spring in very young loblolly pines. A large
number of strobili were subsequently produced, when none would other-
wise have been expected. Rudolph (1979) noted female strobili but
little additional shoot elongation on 12-month-old jack pine (P.
banksiana Lamb.) which had been grown under optimum conditions for
10 weeks in the greenhouse, and subsequently transplanted to nursery
beds in July. In this case, "quiescence" was induced by transplant
In the studies relating mineral deficiencies to increased flowering
(Lyr and Hoffman 1964, Kuo 1973, Kamienska et al. 1973, Sweet and Will
1965, Giertych 1975), nutrient deficiencies observed may not have been
the cause. The treatments probably caused an early cessation of growth,
allowing a "resting" bud during the growing season on which initials were
able to form. Thus, many of the conflicting results relating both growth-
promoting and growth-retarding conditions to increased flowering can be
Materials and Methods
The study area was a seed orchard, maintained by Region 8 of the
U.S. Forest Service, located in south Mississippi about 25 miles (40 km)
southeast of Hattiesburg. The orchard consisted of ramets from 50
superior loblolly pine selections located on the National Forests in
south Mississippi. There were about 4,000 ramets of these clones
planted at 15 x 30 foot (4.6 x 9.1 m) spacing.
The ortets ranged from 31 to 67 years of age as determined by
increment cores. More than half were between 43 and 54 years of age.
All were reproductively mature, as one criterion for selection was
"some evidence of cone production," when selected or after release
and fertilization. The orchard site was cleared of second growth long-
leaf pine (P. palustris Mill.) in 1961. The area was hilly, as the
elevation varied a maximum of 65 feet (19.8 meters). Loamy sands of
five different series made up the soils in this planting, ranging from
an luka loamy sand at lower elevations near a spring-fed stream to a
McLaurin sandy loam on the somewhat drought ridges. A preliminary soil
analysis indicated that the soils were all rather low in NPK. The soils
at lower elevations were generally lower in available P, higher in
total N and organic matter, more acid and generally wetter than those
at higher elevations (Table 1-1).
The orchard was established from 1963 to 1972 using potted grafts
on nursery-run rootstock. Only a few of the clones were represented
among the ramets established the first 3 years, but in 1966 and 1967 more
than 1,000 ramets, well distributed over the 50 clones, were planted each
year. Grafts were planted in 1968 through 1972 to fill in vacant spots
and were also well distributed over the 50 clones, but with very few
ramets per clone.
Male and female flowering, height, and D.B.H. of all ramets were
measured at irregular intervals starting in 1969. Table 1-2 summarizes
the measurements that were taken. Elevation above the stream was also
estimated for all ramets.
The clonal composition varied greatly depending on the year of
grafting. The total set of data was highly unbalanced, so each age
group was analyzed separately. The data thus broken down by year were
simplified into a group of nested designs:
Source of variation Expected mean squares
Clone 02 + Ko2
Table 1-1. Chemical analysis of soil samples taken from two portions
of the Erambert orchard in 1972.
Organic matter (%)1
Total nitrogen (%)2
Extractable phosphorous (ppm)
Exchangeable potassium (ppm)1
Lower elevation Ridge
3Bray and Kurtz (1945)
Table 1-2. Dates on which various traits were measured on all orchard
ramets (indicated by "X").
Male strobili clusters
"K" is the harmonic mean number of ramets in clones (Becker 1967).
The genetic model is also very simple. The clonal component of variance,
o2, represents the genetic effect, and heritability is calculated:
2 + j
The model for computing genetic and environmental correlations
between measured traits is similarly constructed. Since the ramets
are vegetatively propagated, the clonal variance component contains
all genetic variation, nonadditive as well as additive. The herita-
bilities derived are therefore broad-sense.
The number of ramets per clone in each age group varied widely,
so 10 ramets grafted in 1966 were selected from each of 18 clones.
The selected ramets were evenly spread over the orchard to represent
the full range of site variation.
The sample ramets were distributed in the orchard as follows:
1. Lower elevation--three ramets from each clone located on
the luka soils.
2. Middle elevation--four ramets from each clone on the Benndale
3. Upper elevation--three ramets from each clone on the McLaurin
In addition to the measurements indicated in Table 1-2, the follow-
ing parameters were measured on the 180 sample ramets:
1. Total seed and sound seed (separated by ethanol flotation) from
a five-cone sample taken in the fall of 1976, 1977, and 1978.
2. Weight per 100 seed in 1978.
3. Yearly diameter growth measured from an increment core (cores
taken at 3 feet above ground) in fall 1976.
4. Counts of male strobili clusters and female strobili in 1977,
1978, and 1979.
5. Height and D.B.H. in 1977 and 1978.
In addition, depth of the A horizon was measured at three equally
spaced points around each ramet, 5 feet from the trunk, with a soil
auger. Soil samples from a depth of 10 centimeters (Al horizon) and 75
centimeters (B horizon) were bulked from the three locations, and sand,
silt, and clay percentages were determined by the Bouyoucos hydrometer
method for each depth.
A competition index (C.I.) was computed for each of the 180 sample
ramets, based on spring, 1976, D.B.H., measurements of the study tree
and the two trees 15 feet (4.6 m) in either direction in the row (between
row spacing was 30 feet (9.1 m), and competition at that distance was
considered negligible). The index was calculated as:
C.I. = ED.B.H.2 (study tree).
This is similar to an index that Daniels and Burkhart (1975) found
effective in describing competition effects on stand development in
Environmental effects were analyzed in two ways: (1) by classify-
ing the ramets by soil type (or elevational class) and analyzing
flowering, height, and D.B.H.; and (2) by using regression and correla-
tion to determine relationships of flowering and growth (as dependent
variables) with soils, size of ramet, previous cone crop, etc. (as
All count data were transformed to /count + 0.5. Tests of
statistical significance were at the 0.05 level of probability.
Results and Discussion
Yields for all reproductive structures showed a large general
increase with time, as would be expected in a young orchard (Fig. 1-1).
The 180 sample trees seemed to be fairly representative of orchard
production, as they followed the production trend for the whole orchard.
Large year-to-year variation aside from the general increase in produc-
tion was evident, especially in female strobili numbers (Fig. 1-la).
There was an exceptionally good crop of female strobili in 1976, not
only in this orchard but elsewhere in the area. Production the follow-
ing year was lower than 1976, but it probably was near average.
Cone production (Fig. l-lc) followed female flower production,
though survival varied considerably, from 80% in 1970 and 1971 to 33%
in 1973 and 1977. Male strobili production was very low during the
early years, and then expanded tremendously after 1975 (Fig. l-lb).
It increased each year starting in 1976, but decreased in 1979. Female
strobili, on the other hand, showed decreases in 1977 and 1978, and
an increase in 1979. In comparing the yearly patterns of male and
female flowering (Figs. 1-la and l-lb), it was evident that the two
did not seem to be closely related; that is, the increase or decrease
in female flowering did not correspond to an increase or decrease in
Female flowering. Broad-sense heritabilities for female strobili
production varied considerably from year-to-year (Table 1-3). Herita-
bility seemed to increase sharply with age in the younger material,
Yearly variation in production of reproductive
structures of loblolly pines at the Erambert
Seed Orchard. A. Female. B. Male. C.
Cones. Female strobili in 1975 were estimated
from 1976 cone counts and 1976-1977 stobilus
survival data. Grafts made in 1964 were
included to amplify early flowering trends.
The Y axis was square-root scale.
0 0: Entire orchard
S S: 180 sample ramets
4 4: 1964 grafts
75 ( A. FEMALE
1969 1970 1971 1972 1973 1974 1975 1976 1977 197 1979
1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
) 40 1
> 30 -
o, 0 --- 5 ----"
-- I I I I I
1970 1971 1972 1973 1974 1975 1976
1977 1978 1979
1969 1970 1971 1972 1973 1974 1975 1976 1977 1978
\ s s ss
Table 1-3. Broad-sense heritabilities for female flowering by year
for different aged grafts.
Year -------------Year examined--------------------
grafted Clones Ramets 1969 1970 1971 1972 1973 a age1 1976
No. No. ----------------------h2
63 8 41 0.871 0.687 0.704 0.623 0.420 0.622 0.235
64 13 198 0.600 0.594 0.780 0.673 0.568 0.722 0.345
65 14 56 0.056 0.396 0.527 0.555 0.426 0.490 0.401
66 43 1106 0.326 0.584 0.486 0.595 0.481 0.606 0.441
662 18 180 0.274 0.467 0.270 0.541 0.534 0.593 0.501
67 45 1351 0.268 0.298 0.475 0.370 0.439 0.558
68 34 388 0.477 0.430 0.577 0.304 0.423 0.565
69 34 206 0.204 0.422 0.276 0.356 0.391
70 47 186 0.431 0.299 0.395 0.617
1This is the heritability
the four heritabilities.
of 1969-73 average flowering, not the average of
but then leveled off and perhaps even decreased with age. Curves
were fitted to the change in variance with age and are shown in Figure
1-2. Variance and heritability were essentially zero before age 3,
as only sporadic flowering occurred before that age. Genetic variance
increased very rapidly from age 3 to age 5, and continued to increase
but at a slower rate to age 13. Apparently the flowering trait was
better expressed among larger trees. The environmental variance increased
sharply at first, likely because of the large variation among ramets in
graft union formation and in subsequent growth. It leveled off some-
what between ages 5 and 9 as more uniform growth occurred. After age
8 or 9, it increased sharply again, probably due to the large variation
in competition among clones. At this age and spacing, 15 x 30 feet
(4.6 m x 9.1 m), within-row competition was beginning to have an effect
but was not uniform. Some of the ramets had essentially no competition
because the ramets on either side were missing or were very small.
Others had trees larger than themselves growing on either side. As a
consequence, variance increased sharply as ramets were affected by
competition in widely varying degrees.
Male flowering. Heritabilities for male strobilus production were
lower than for female strobilus production, and more variable (Table 1-4).
Significant heritable variation occurred only in the older material,
probably due to the low frequency of male flowers before 1976 (Fig. l-lb).
It appeared, however, that eventually the heritability for male flower-
ing would be nearly as high as that for female flowering in older material,
probably above 0.4. The heritabilities for male flowering for grafts
to a r-
40 u J
4-4 0 -H
l 0 a0
H 0 0
C *l o1
o o44 a)
cb 3 0
nrt 60 <
a) *rl *rH
I I I mI *i
N 4 I
0 Ln u0 o
( N r-4 rH
ZD ONIRaMOia aqVKIyw SONVINVA
Table 1-4. Broad-sense heritabilities for male flowering by year for
different aged grafts.
grafted Clones Ramets 1970 1971 1972 1973 1970-731 1976
No. No. --------------------h2 --
63 8 41 0.069 0.461 0.364 0.243 0.458 0.433
64 13 198 0.245 0.529 0.611 0.016 0.506 0.407
65 14 56 03 0 0 0 0 0
66 43 1106 0.103 0.129 0.165 0.010 0.111 0.272
662 18 180 0.002 0.102 0.168 0 0.055 0.208
67 45 1351 0.243 0.332 0.232 0.037 0.270 0.364
68 36 388 0 0.115 0.065 0.174 0.434
1This is the heritability of 1969-73 average flowering, not the average of
the four heritabilities.
3Negative heritabilities were assumed to be zero.
made in 1965 inexplicably were always slightly negative or zero (negative
heritabilities were assumed to be zero). Heritabilities for female
flowering for these same trees seemed reasonable (Table 1-3). And many,
but not all, of the same clones were included in the 1964 grafts where
heritability for 1970-1973 average male flowering was computed as 0.506.
The range of clonal means for male flowering was much smaller in the
1965 grafts than in the 1964 grafts, however, and probably accounted for
the low heritabilities. The 1970-1973 average clonal means for male
flowering ranged from 0 to 140 clusters per ramet on the 1964 grafts.
But male flowering ranged from 0 to only 20 clusters per ramet on the
1965 grafts. With respect to clonal variation, the 18 clones in the
sample population seemed to be fairly representative of the 1966 grafts
for both male and female flowering (Table 1-4).
Age and rootstock could bias clonal variation. The age of the ortet
influenced growth (Franklin 1969) and flowering (Schnidtling and Griggs
1976) in slash pines. Ortets in this study did not vary greatly in age
and were all reproductively mature, so age effects were probably minimal.
Correlations based on the 1976 flower crop in the 1966 and 1967 grafts
revealed no important age effect, as the correlation between ortet age
and female flowering was r = -0.053, and between age and male flowering
r = -0.212. Neither of these was significant at the 0.05 level. There
was clearly no age bias in female flowering and very little, if any, in
All ramets in this orchard were grafted on nursery-run seedling
rootstocks, which surely affected variation in growth and flowering.
Schmidtling (1978) found that rootstock variation significantly affected
flowering of loblolly pine grafts. This was true for different species
of rootstock as well as different half-sib loblolly families used as
rootstocks. Specific clonal reactions to grafting, such as incompati-
bility, could have increased clonal variation. But the rootstock
variation undoubtedly contributed substantially to the within-clone
(error or environmental variation), lowering heritabilities. Because
of this, true broad-sense heritabilities were likely higher than those
Cone and seed yields. Cone and seed production are even more
important aspects of orchard performance than flowering because im-
proved seed are the desired end product of any orchard. In addition
to being affected by the same variables involved in production of the
female strobili from which they are derived, cone and seed production
are further affected by variations in conelet and seed survival. Much
of this added variation is caused by insect predation, but other environ-
mental and genetic factors may contribute to it (McLemore 1977).
Genetic variation in cone production was important, as heritabili-
ties of cone counts were in the same range as heritabilities for female
flowering. (Cone production figures for any given year should be
compared to the previous year's flower production; e.g., the 1970 cone
crop originated from the 1969 flower crop, etc.) The heritability
estimates approached and sometimes exceeded 0.5 but they were highly
variable (Table 1-5). Heritabilities for the 180 sample trees averaged
slightly lower than the 1966 grafts as a whole. In addition to those
listed in Table 1-5, the heritability estimate for cone production was
0.501 in 1977 and 0.648 in 1978 for the 180 sample trees. The genetic
Table 1-5. Broad-sense heritabilities of cone prc-Xcction by year for
different aged grafts.
Year ---------- Year exa-inej-------------
grafted Clones 7.Laets 1969 1970 1971 :172 1973 1976
No. o. --------------------h-------------------
63 8 41 0.593 0.808 0.647 0.375 0.154 0.714
64 13 198 0.550 0.523 0.565 0.763 0.244 0.649
65 14 56 0.571 0.396 0.368 0.329 0.154 0.679
66 43 1106 0.083 0.328 0.507 0.405 0.542 0.535
661 18 180 0.788 0.252 0.429 0.120 0.454 0.454
67 45 1351 0.030 0.238 0.102 0.424 0.534
68 36 388 0.264 0 0.518 0.518 0.300
69 34 206 0.342 0.367 0.368
correlation bet-..-eE flowering and subsequent cone crops was high: r =
0.905 for 1976 fe->le flowers with 1977 cones, an. r = 0.984 for 1977
female flowers .i:- 1978 cones. Those clones with the heaviest flower
crops produced the =.st cones.
Conelet sur-.i--al, i.e., the proportion of female flowers that pro-
duced cones, varied widely among clones, and averaged 32.7% for 1976-
1977 and 63.8% for 1i77-1978 (Tables 1-6 and 1-7). Conelet survival
was moderately heritable: h= 0.356 for the 1976-1977 crop and h
0.453 for the 1977-1978 crop.
Heavy-flowering clones had a slight tendency to produce a smaller
proportion of surviving cones in the 1976-1977 crop, but not in the
1977-1978 crop. Genetic correlation for the 1976 female flower crop,
with subsequent survival of that crop, was -0.121. For the 1977-1978
crop, which was characterized by a smaller flower crop but much better
survival, genetic correlation was 0.504. Possibly the heavier drain
on nutrients of the larger crops of developing cones adversely affected
survival in a year where the overall crop was heavy. Insect predation
could also have explained the difference. This factor was the major
cause of conelet abortion in loblolly pine (McLemore 1977).
If insect control was poor in 1976-1977, the trees with larger
flower crops could have presented a more favorable situation for insect
population development. This would explain the trend for heavier flower-
ing trees to have a smaller percentage of survivors. On the other hand,
if insect control -as better in 1977-1978, those populations could not
have developed as fully in the heavier flowering trees. Consequently,
Table 1-6. Clonal -3ans for flower, cone and see- yield for the 130
sample riets in 1977.
1976 female& Cones/ Conelet Sound Total
Clone flowers/rsane ramet survival seed/- :n seed/cone
No. No. % No.
A143.0 41.0 38.5 29 36.5
B 102.0 23.8 26.0 26.6 39.8
C 144.0 78.8 62.8 45.$ 63.4
D 238.0 87.3 30.8 59.1 79.3
E 41.0 10.8 46.6 49.8 59.0
F 13.0 7.0 59.3 39.2 52.2
G 191.0 120.7 68.5 37.1 47.8
H 103.0 37.2 52.5 42.3 51.1
I 209.6 60.0 34.7 60.0 77.5
J 16.5 7.5 56.7 37.2 43.9
K 65.0 5.0 10.8 50.7 60.4
L 55.0 2.2 26.8 55.3 58.8
M 177.5 26.9 16.4 40.2 48.9
N 21.0 7.3 28.4 24.1 40.5
0 10.0 4.8 57.0 48.1 53.5
P 118.0 22.1 28.3 17.3 25.7
Q 97.0 9.0 14.2 25.6 30.6
R 239.5 97.0 45.9 56. 81.1
Mean 110.2 36.0 32.7 41.? 52.8
Heritabili- 0.501 0.587 0.199 0.293 0.409
iThis column is based on individual tree data. Trees -ith no flowers were
not included, so it is not the same as the percent cha To'.ld be derived
by dividing average 1978 cones per ramet by average 19-6 flowers per ramet.
able 1-7. Clonal means for flower, cone and seed yield f:r -he 180
sample ramets i. 1978.
1977 female Cones/ Conelet Sound T::a- ,eight of
CLone flower/ramet ra-e: survival seed/cone see: c -- 130 seed
No. o. % No. v. Grams
A 9.2 3.2 30.5 21.7 :?.9 3.2
B 15.1 5.3 29.9 11.5 --.5 3.1
C 51.9 17.0 36.8 7.8 ;.3 3.8
D 117.9 93.3 82.8 17.9 73.3 3.1
E 14.9 8.6 78.0 39.2 65.7 2.6
F 1.5 0.7 33.3 3.2 Pn.5 3.0
G 28.0 26.0 92.3 36.5 63,2 2.8
H 59.8 59.1 86.7 26.6 53.5 3.1
I 19.9 14.6 61.0 23.5 53.0 3.2
J 15.1 14.2 64.8 50.8 6.7 3.3
K 17.6 10.0 44.2 39.6 81.5 3.4
L 11.5 4.2 16.6 22.1 -.1 3.5
M 112.0 88.3 69.0 16.2 56.1 4.0
N 16.2 9.8 56.3 33.4 65.4 2.4
0 7.8 4.9 56.4 40.4 61.3 3.1
P 125.5 93.5 69.6 7.5 15.6 2.6
Q 82.1 49.2 52.3 10.8 -3.5 3.5
R 275.1 202.9 80.6 11.6 -1.3 4.4
Average 54.5 39.2 63.8 23.3 55.4 3.2
Heritabili- 0.629 0.618 0.274 0.307 3.265 0.366
there --as a positive relationship bet.;en flowering and survival.
Genetic : rrelation for conelet surviv-al between the two crops was
positive Dut weak (r = 0.228).
Average yields of total seed per zone varied from 42.3 in 1976
(Table 1-3) to 56.4 in 1978. The yield in 1977 (Table 1-6) was similar
to tha: in 1978 (Table 1-7), 52.8 seed per cone. Clones varied widely
in total seed per cone all 3 years (Tablesl-6, 1-7, and 1-8), with a
three zo fourfold difference between those having the largest and those
having the smallest yields per cone. Heritabilities were moderately
high for total seed per cone (0.241, 0.409, and 0.366 for 1976, 1977,
and 1978, respectively).
Yearly averages of sound seed per cone varied much more than the
total seed per cone. They ranged from 16 per cone in 1976 to nearly
42 per ccne in 1977. Clonal averages of sound seed yields per cone
also varied more widely than total seed, especially in 1976 and 1978.
During those years there was a tenfold or elevenfold difference among
clones. Proportion of total seed which were sound was about 38% in
1976 and 41% in 1978. In 1977, however, nearly 80% of the total seed
was sounc, and there was only a fourfold difference among clones.
The yearly differences in proportion of sound seed was probably
due to variation in effectiveness of insect control. Apparently insect
control was effective in 1977, but not so effective in 1976 and 1978.
Losses :c seed bugs (Leptoglossus corculus [Say]) occurred the year
before c=ne harvest and were manifested as empty and aborted seed. Cone-
let/flcwer losses, on the other hand, occurred mainly in early spring
Table i--. Clonal means for cone and seed yields for the 180
sample ramets in 1976.
Clone Cones per Sound seed Total seed
ramet per cone per cone
curing and shortly after flowering (McLemore 1977). The high survival
rae of conelets for the 1977-1978 crop (71.8%), compared with the 1976-
1977 crop (32.7%), also indicated better insect control. McLemore (1977)
showed that most conelet losses in loblolly pines were caused by insects.
Data were not available to attest to the relative degree of insect
losses for the years 1976 through 1978. Some survey data were availalbe
to provide an estimate of the overall seriousness of insect problems in
this orchard. These data were collected in various years to assess the
impact of insect loss and the effectiveness of various treatments.1 In
1975, 85% of female strobili survived on trees sprayed periodically with
the insecticide GuthionR versus 64% on untreated checks. Cones from
trees treated with the systemic insecticide FuradanR averaged 47.6 sound
seed per cone versus 14.3 sound seed per cone in untreated checks in 1976.
For the 1977-1978 flower and cone crop, there was a 21.9% cone loss from
trees treated with FuradanR versus 39.6% from untreated checks. Cones,
caged with a wire mesh to exclude insects, averaged 55 sound seed per
cone in 1978. Uncaged checks averaged only 19 sound seed per cone.
Apparently insects caused heavy losses of female strobili, cones, and
seed in this orchard.
Clonal variation as well as yearly variation in conelet survival
and seed yield could be related to variation in insect predation. Merkel
et al. (1966) found evidence for inherent resistance to coneworm (Dioryc-
tria amatella [Hulst.]) attacks in slash pine.
LData provided by Neil Overgaard, Entomologist, U.S.D.A., Forest
Service, Pineville, Louisiana.
Year x clone interactions. In general, clonal variation in flower-
ir follo~-.ed the general yearly trend; i.e., all clones flower; well
in good years such as 1975, and poorly in bad years such as 197-. The
relative ranking of clones, however, changed considerably from year to
year, as exemplified by 6 of the 18 sample clones shown in Figure 1-3.
Those 6 clones were representative of the range of flowering performances
of the 18 sample clones. The clones varied considerably in female flower-
ing from year to year (Fig. 1-3a), and there seemed to be a lun--term
trend for some clones to move up or down in relative rank. Clone P was
highly variable, and in the early years was the best flowering clone;
but it has been surpassed by several other clones. Clone R, initially
among the poorest, now has become the best.
Strong year x clone interactions were apparent in male flowering,
also (Fig. l-3b). For instance, Clone D was the best pollen producer
during the first 3 years in which measurements were made, but it was
average or only slightly above average the last 3 years. Clone R was
at the bottom during the first 4 years and had no pollen, but it assumed
the number one position the last 3 years. Clone P oscillated year to
year from good to average. An important consequence of such year x
clone interaction in pollen production is that the genetic makeup of
the seed produced by the orchard will vary from year to year, even
for seeds collected separately by female parent. The amount of selfing
could also be expected to vary.
In an analysis of female and male flowering for the years 1976
through 1978, year x clone effects were significant and constituted an
important part of the total variation for both male and female flowering
(Table 1-9). But clonal effects were much greater, and heritabilities
Year-to-year clonal variation in male and female
flowering of 6 representative clones from the 18
sample clones. Flowering of each clone is ex-
pressed as a percent above or below the mean for
each year. A. Female flowering. B. Male
flowering. Letters refer to clone designation.
A. FEMALE FLOWER::;,
A- / A/
1970 1971 1972 -1973 1974 1975 1976
C. 50 -
1977 1978 1979
3. MALE FLOWERING
- -^ N
N. N -
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
able 1-9. Analysis of variance and covariance of male and female
flowering for 1976 through 1978 (year effect considered
fixed, all others random).
Source of Degrees of Expected Mean
variation freedom Squares
Clone (C) 17 Ge + YaR2 + YRac
?amet/Clone (R) 162 e2 + Yo2R
Tear (Y) 2 e2 + Rayc2 + CROy2
Tear x Clone 34 Ge2 + Rayc2
Error 324 Ce2
Variance Variance Correlation
Female Covariance Male Coefficient
Clone 10.88 10.95 17.02 0.805
Ramet 4.82 7.62 18.80 0.801
Year 4.17 -7.72 20.50 -0.835
Year x Clone 3.26 -0.69 3.22 -0.214
Error 5.55 1.45 12.70 0.173
Total 28.68 72.24
for 1976-1978 average male and female flowering were 0.425 and 0.621,
respectively. Although yearly variation and year x clone interactions
were important, the average performance of a clone was highly heritable.
Correlation coefficients computed from the variance-covariance
components showed that male and female flowering were well correlated
by clone (r = 0.805) and by individual ramet (r = 0.801) over the 3
years (Table 1-9). The correlation coefficient for year (r = -0.835)
was strongly negative, however. This indicates that yearly variation
in female flowering has had an opposite basis than for male flowering,
which was evident in comparing Figures 1-la and 1-lb.
In spite of these interactions, female and male flowering for the
whole orchard were correlated positively on a clonal basis, though the
correlations were generally not as strong as in the sample trees.
Those clones with the most female flowers generally were good pollen
producers also. The genetic correlation between male and female flower-
ing in 1976 was almost identical for the 1966 and 1967 grafts (r = 0.589
and r = 0.582, respectively). These were the most reliable figures, as
grafts made in those 2 years had 39 and 45 clones represented, respec-
tively. The 1976 flowering data for the 1966 grafts are shown in Figure
One interesting aspect of Figure 1-4 was the number of clones in
the upper left quadrant. These were below average for female strobili
but above average for male strobili. If only female flower or cone
production is used as a basis for preliminary rogueing of an orchard,
some valuable pollen-producing clones might be eliminated. One clone
I I1 4 1 I I -
10 25 50 75 100 150 200
FEHRLE STRlOILI NUMBER / RRHET
Scatter diagram of clonal means for female versus
male flowering. Data were from 10-year-old grafts
in 1976. Dotted lines indicate means for female
flowering (vertical) and male flowering (horizontal).
Both axes are square-root scale.
in particular averaged only 1.0 female strobili per ramet, but had
over 50 male strobili clusters per ramet, which was -ore than twice
the average (Fig. 1-4). It would be more logical to limit rogueing
to the clones in the lower left quadrant.
Contribution of each clone to the seed produced. Many tree
improvement personnel are concerned that only a few of their clones
produce the major part of their seed. One tree improvement coopera-
tive estimates that 20% of the clones in their orchards produces 80%
of the seed. This is the so-called "20/80 rule" (N.C. State University
1976). A strong clonal effect was evident in these data, although not
as strong as "20/80."
In this study, sound seed per cone and total cone counts per ramet
indicate that the top 4 of 18 clones (22%) produced 76%, 65%, and 62%
of the total seed produced in 1976, 1977, and 1978, respectively. How-
ever, the top four clones were not the same each year. Figure 1-5 shows
the strong year x clone interaction for seed produced for the top 7 of
the 18 clones. Only one clone ("R" in Fig. 1-5) was included in the
top four all 3 years. On the basis of total seed produced for all 3
years together, the top four clones produced 58.3% of the total seed.
The genetic correlations between number of female flowers and the
number of sound seed produced from that particular flower crop were
very high (r = 0.857 for the 1977 crop, and r = 0.884 for the 1978 crop).
Clones which produced female flowers were the ones which produced the
seeds, despite all the hazards which occurred between formation of the
female strobilus and harvest of full seed.
Yearly seed production of the seven best seed
producers of the 18 sample clones, expressed
as a percent of the total production of all
18 clones each year.
The theoretical genetic contribution of each of the 18 clones
was computed by using the proportion of sound seed and the proportion
of pollen catkins contributed by each clone at time of pollination.
This was done for the 1977 and 1978 seed crops (1975 pollen data were
not available for the 1976 crop) and results are shown in Table 1-10.
There was a distinct year x clone interaction, as there was with male
and female flowering and seed produced. The top four clones contributed
49% of the genes for the 1977 and 1978 crops, respectively. These
values are somewhat lower than those obtained from seed alone (65%
and 62%, respectively). The present approach assumed that the male
contribution of a given clone was proportional to the number of cat-
kins it produced, which was surely not entirely true.
Clearly, the genetic quality of seed from an orchard can vary
considerably from year to year. This will be true even if seed are
collected from the same mother tree, since pollen composition will
vary from year to year. This undoubtedly accounts for the seed crop
year x genotype interaction observed by C. H. Lee (1978).
Soils. Although genetic variation accounted for a large propor-
tion of the variation of many of the characters measured, environmental
variation was also evident in this orchard. It was noted in 1969 that
the trees in the lower elevations were growing better but flowering less
abundantly than the trees located at higher elevations. Those differ-
ences seem to have held up over the years, especially for female
flowering. This seemingly inverse relationship between growth and
flowering suggested the present study.
Theoretical genetic contribution of each of the 18
sample clones to the total output in 1977 and 1978.
Based on sound seed and pollen produced in the year
of pollination (selfs excluded). Ranked by 1977
Clone Year Rank in 1978
% of total ---------
According to a U.S. Forest Service soils survey, completed before
the orchard was established, the orchard was planted on five soil series:
luka, Benndale, McLaurin, Van Cluse, and Lucy loamy sands to sandy loams.
The first three types predominated. luka soils are Entisols, and McLaurin
and Benndale are Ultisols of similar taxonomy (Table 1-11). The soils
did not vary greatly in their physical characteristics. luka, an alluvial
soil located in the lower elevations near the stream, had the highest
proportion of sand (85% and 90% at 10 cm and 75 cm depths, respectively)
(Table 1-11). At the other extreme, McLaurin, which was located on the
ridges, had the least sand at both depths (78% and 84% sand in the upper
and lower samples, respectively). Benndale, which was intermediate in
elevation, was intermediate in texture. Clay content of the surface
horizons of these samples varied from 5% to 7%. Differences in textures
were mainly a function of varying proportions of sand and silt.
The relationship of site variables, such as soil texture and eleva-
tion, to flowering were analyzed in two ways: (1) the soil series were
used as a discrete classification variable; each sample tree was classi-
fied according to a soils map, and (2) regressions were used to relate
soil and site variables to flowering, without regard to soil series.
The latter approach was probably the most valid, as soil variations
were continuous and not discrete; and the soils around individual trees
have been rather arbitrarily classified. However, the data can be
presented in a much simpler way with the former approach, so both methods
will be utilized.
Table 1-11. Description of the soils where the IS0 sample trees were
Series Family Subgroup Order
luka Coarse-loamy, siliceous, thermic Aquic Udifluvents Entisols
Benndale Coarse-loamy, siliceous, thermic Typic Paleudults Ultisols
McLaurin Coarse-loamy, siliceous, thermic Typic Paleudults Ultisols
Series Eleva- Graph Depth 10 cm depth 75 cm depth
tion symbol A horiz. sand silt clay sand silt clay
luka Lower 1 31 85 10 5 90 5 5
Benndale Middle 2 35 80 13 7 87 6 7
McLaurin Upper 3 32 78 15 7 84 9 7
Female flowering. Measurements of flowering and growth generally
confirmed earlier observations that flowering was best on soils where
growth was poorest. In general, growth was best on the l-ka soils at
lower elevations (Fig. 1-6). By fall of 1977 the sample ramets on luka
soils averaged 11.1 m in height while those on Benndale and McLaurin
soils averaged only about 10.3 m (Fig. l-6a). Radial growth was clearly
superior for ramets on the luka soil until 1973, when it -as equaled or
surpassed by those on the Benndale soil (Fig. l-6b). As :he sample trees
grew larger, they seemed to be less influenced by soils.
By contrast, flowering was poorer on the soils on which growth was
favored (Fig. 1-7). Flowering was always poorer on the luka soil for
all years measured except the last, though the differences were statisti-
cally significant only in 1972 and 1973.
Although there were differences in flowering and growth of trees
on the different soils, soil texture measurements were infrequently
correlated with either flowering or growth. Female flowering was posi-
tively correlated with elevation in the early years. It was strongest
in 1973 (r = 0.296) and weakest thereafter. This was opposite the early
relationship between growth and elevation, but the trend toward decreasing
the importance of elevation with time was similar.
The variable most strongly and consistently correlated with female
flowering was D.B.H. Correlation increased with age; in 1973 the
correlation between D.B.H. and female flowering was r = 0.295, and in 1976
it was r = 0.524. This seemed to contradict the negative relationship
between growth and flowering by soil series. But the magnitude of these
correlation coefficients allowed for considerable variation due to factors
other than D.B.H. Correlations between D.B.H. and flowering were even
Growth of the 180 sample ramets by soil series.
A. Height. B. Radial growth, measured from
luka soils (lower elevation)
Benndale soils (middle elevation)
McLaurin soils (higher elevation)
I 7 -
1969 1970 1971 1972 1973 1974 1975 1976 1977
B. RRDIRL ERDWTH
1969 1970 1971 1972 1973 197- 1975 1976
30 Grafts 2
1969 1970 1971 1972 1973
1974 1975 1976 1977 1978 1979
Female flowering of the 180 sample ramets and the
1964 grafts by soil series. Vertical axis is
square-root scale. Asterisk indicates differences
were statistically significant.
luka soils (lower elevation)
Benndale soils (middle elevation)
McLaurin soils (higher elevation)
Fitted regression line for flowering/age
stronger when variation due to soil series was removed. This was done
by computing correlation coefficients between D.B.R. and flowering
separately within each of the three soil series, and then pooling the
three values. In 1973, the within-soil correlation coefficient between
D.B.H. and flowering was r = 0.385 compared to r = 0.295 overall; in
1976, the pooled correlation coefficient was r = 0.553 compared to
r = 0.524 overall.
Flowering in a given year was always correlated positively with the
size of the cone crop in the previous year. This was true even for the
1978 female crop, which was preceded by a cone crop varying from 0 to
390 cones per ramet. Correlations ranged from r = 0.286 for previous
cone crop with female flowering in 1971 to r = 0.644 for previous cone
crop with 1978 female flowering.
The positive correlation probably originated solely as a result of
the positive correlation between successive flower crops, which was both
genetic and environmental, and the close correlation between a flower
crop and its subsequent cone crop. Positive correlation between cone
crops and flowering in the subsequent year does not prove, by itself,
that the drain on nutrients of developing cones would not affect flower-
ing. But it suggests that the size of the cone crop, in an operational
seed orchard, probably is not important in determining the next year's
flower crop or inducing periodicity of cone crops.
Effects of competition on growth and female flowering were weak but
significant. The correlation between competition index and 1976 female
flowering was r = -0.183, and that between competition index and 1977
D.B.H. was r = -0.255.
In the stepwise multiple regressions, D.B.H., elevation, and height
were significantly related to female flowering in 1973. Radial growth,
summerwood percent, and percent sand in the A and B horizons were not
significant. The equation
/1973 female flowering + 0.5 = 0.6 + 2.1 D.B.H. + 0.05 elevation
explained 29.2% of the variation in female flowering. The sign of the
coefficient for height suggests that if D.B.H. and elevation are held
constant, flowering was negatively related to height. The simple correla-
tion coefficient between 1973 female flowering and height was positive
but weak (r = 0.151).
Female flowering in 1977 was significantly related only to D.B.H.,
height, and competition index (Table 1-12). The equation
/1977 female strobili + 0.5 = 6.1 + 2.4 D.B.H. -0.5 height -2.4 CI
explained 28% of the variation in female flowering. Elevation and soil
texture did not enter into the equation, which agrees with the previous
conclusions that the ramets seem to be more independent of site variation
as they increase in size, but competition was important.
One important variable, moisture, was not measured in this experi-
ment and could possibly account for some of the variation not explained
in the multiple regressions. Drought, especially at critical times, can
increase flowering in loblolly pines (Dewers and Moehring 1970, Gallegos
1978). Casual observations during soil sampling in mid-summer indicated
that there was much less soil moisture at higher elevations in the
McLaurin soils than at lower elevations in the luka soils. This might
have accounted for much of the difference in flowering performance,
especially in the earlier years.
Stepwise regression relating 1977 female flowering as
a dependent variable with various independent variables.
Independent Regression Standard error Partial F test
Intercept 6.06 2.80 4.68
1977 ,LE 2.37 0.29 65.17
1977 height -0.51 0.10 25.05
Competition index -2.41 1.17 4.22
Independent variables not in the equation
Variable Partial correlation F to enter
Elevation 0.043 0.33
1976 DBH increment -0.134 3.22
Sand % A horizon -0.049 0.43
Sand % B horizon -0.093 1.55
R2 = 0.283
Overall F = 23.13
Male flowering. The pattern for male flowering was unlike that for
female flowering (Fig. 1-8). For most years, male flowering was best on
the luka soils, exactly opp-site the pattern for female flowering. Differ-
ences in flowering by soil type were significant only in 1976 and 1977.
Size of the grafts might have been a more important factor in male
flowering than in female flowering. Male strobilus production in Pinus
is usually confined to vegetatively less vigorous branches in the lower
part of the crown (Eggler 1961, Wareing 1957). It occurs on Southern
pines mainly on branches in the lower crown that produce only one cycle
of growth during the growing season (Eggler 1961, Greenwood 1979). Small
trees, whether seed grown or grafted ramets, have a relatively small
number of these less vigorous branches. Tree size would not be as
important for female flowering, as female strobili are produced on more
vigorous shoots, such as would occur over the entire crown of a smaller
graft or in the upper crown of larger grafts. This might explain the
lack of agreement between male and female flowering by soil type. The
fact that almost no male flowering occurred in the sample trees before
1976 (Fig. 1-8) supported this hypothesis, because none of these trees
would have had suppressed branches when young, regardless of soil type.
The only independent variables consistently correlated with male
flowering were associated with current size. The coefficients ranged
from r = 0.156 between 1973 male flowering and height to r = 0.537
between 1977 male flowering and D.B.H. In all years except 1976, site
elevation was not related to male flowering. In 1976 the correlation
was weakly negative (r = -0.154) which was opposite that for female
180 Sapl /
Ra rets / 3
1971 1972 1973 1974 1975
1976 1977 1978 1979
Male flowering of the 180 sample ramets and the
1964 grafts by soil series. Vertical axis is
square-root scale. Asterisk indicates differ-
ences were statistically significant.
1- 1: luka soils (lower elevation)
2- 2: Benndale soils (middle elevation)
3- 3: McLaurin soils (higher elevation)
In the stepwise regressions, only D.B.H. was correlated signifi-
cantly with male flowering in 1973. The regression equation derived for
1977 male flowering was similar to that derived for 1977 female flower-
ing, as the size variables were the first two entered into the equation.
/1977 male strobili clusters + 0.5 = 1.7 + 4.0 D.B.H. -0.6 height
-5.7 D.B.H. increment
explained 37% of the variation in male flowering (Table 1-13). Competi-
tion index was just below the significance level required to enter the
Cone and seed yields. It was anticipated that the drier soils might
have a deleterious effect on conelet survival and seed yields (Gallegos
1978). However, that was not true. If any trend was apparent, it was
opposite to that expected; i.e., yields tended to be poorer on the luka
soils where growth was more favorable. Sound seed per cone, for instance,
varied from 10.9 on the luka soils to 20.0 on the McLaurin soils in 1976.
Similar trends were noted for conelet survival. None of the statis-
tical tests indicated significance, however, even though some of the
means differed substantially. As mentioned previously, most of the
variation in these traits was caused by variability in insect predation.
Although there was fairly strong clonal variation in these traits (Tables
1-6, 1-7, and 1-8), the environmental variation seemed largely random,
at least with respect to soil type.
Cone counts were included as an independent variable in all the
analyses, with the expectation that large cone crops might have a
deleterious effect on cone and seed yields. However, this was not true.
Stepwise regression relating 1977 male flowering as a
dependent variable with various independent variables.
Independent Regression Standard error Partial F test
Intercept 1.69 3.978 0.18
1977 DBH 3.99 0.417 91.52
1977 height -0.57 0.145 15.60
1976 DBH increment -5.66 1.667 11.52
Independent variables not in the equation
Variable Partial correlation F to enter
Elevation -0.092 1.51
Sand % A horizon -0.004 0.00
Sand % B horizon 0.134 1.36
Competition index 0.134 3.20
R2 = 0.370
Overall F = 34.4
The size of the cone crop was always uncorrelated or positively corre-
lated to the cone and seed variables. The strongest example was the
positive correlation between 1978 seed weight and 1978 cone crop.
Variation in the 1978 cone crop was substantial, ranging from 0 to 425
cones per ramet. The overall correlation coefficient was r = 0.424. A
large part of this was genetic (r = 0.612), indicating that those clones
with large cone crops tended to have heavier seed.
Effects of rainfall. Summer rainfall varied considerably over the
10 years of the study (Fig. 1-9) and these variations might explain yearly
variations in flowering. Promotive effects of drought during strobilus
initiation were mentioned previously. Dewers and Moehring (1970) pro-
vide the best experimental evidence for the effects of moisture stress
on reproduction. They subjected loblolly pines to four moisture regimes:
(1) drought the entire growing season, (2) irrigated the entire growing
season, (3) drought April through June, irrigated thereafter, and (4)
irrigated April through June and drought thereafter. Treatments 1, 2,
and 3 were approximately equivalent; trees under treatment 4 produced
twice as many cones as the other treatments.
Figure 1-10 compares the timing of reproductive events in loblolly
pines with a graphical representation of Dewers and Moehring's treat-
ment 4. It was difficult to relate yearly increases in flowering to
climatic variables since flowering was expected to increase each year,
at least in the earlier years. However, some support for the hypothesis
that a rainfall pattern approximating Dewers and Moehring's treatment 4
increased female flowering can be found by comparing Figure l-9a with
Figure 1-9c. Since the effects of moisture on female flowering have
100 A. FEN,.LE FLOWERING
69 70 71 72 73 74 75 76 77 78 79
B. MALE FLOWERING
69 70 71 72 73 74 75 76 77 78 79
68 69 70 71 72 73 74 75 76 77 78
Yearly variation in flowering compared to early spring
and late growing season rainfall. A. Female Flowering.
B. Male Flowering. C. Rainfall
180 sample ramets
Average early growing season (April, May, June)
Average late growing, season (July, August, September)
been fairly well documented in the literature, it is more important to
consider the effects of rainfall on male flowering, for which there is no
Some workers (e.g., Giertych 1967) have claimed that treatments such
as fertilization and hormonal applications might have opposite effects on
male flowering from that on female flowering. However, since male and
female strobili initiate and differentiate at different tines, a treatment
applied at one time might affect their development differently. Consequent-
ly, Greenwood and Schmidtling (1980) reported that such treatments as
fertilization, subsoiling, or hormone application might promote both male
and female flowering if the treatments are properly timed.
Considering the fact that male flower initiation occurs in early
summer and female flower initiation occurs in late summer (Fig. 1-10), it
is reasonable to assume that if drought could influence both male and
female flowering, it should occur at different times. The opposite, or
at least unrelated yearly variation in male flowering compared to female
flowering, can be explained by differences in rainfall patterns. Rain-
fall in 1970 was characterized by low early summer rain and higher late
summer rain (Fig. 1-10c). This was followed in 1971 by a good pollen
crop (Fig. 1-10b, 1964 grafts) and a poor female flower crop (Fig. 1-10a).
The rainfall pattern in 1972 was opposite that of 1970, with higher early
summer rain and lower late summer rain. This pattern was followed by
a poor pollen crop (Fig. 1-10b) and a good female flower crop (Fig. 1-10a),
which was exactly opposite the 1971 pattern.
Pricordia i',it 'a i .
Anthesis Conelet Formaticn
Cone Grc-t": Fertitiization Deveio e Seed Fall
I 1 i 1 I
Irrigation .. ............. .
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Approximate timing of reproductive events in loblolly
pines (above) compared to Dewers and Moehring's (1970)
best cone inducing treatment (below). Timing of repro-
ductive events approximated from Greenwood (1979),
Greenwood and Schmidtling (1980), Eggler (1961), Dorman
and Barber (1956), and Sarvas (1962).
The greatest contrast, not only in rainfall pattern but also in
male and fer-le flowering, was between 1976 and 1978. There was a very
large female flower crop and a poor pollen crop in 1976, whereas 1978
was characterized by a very good pollen crop and poor female flower
crop (Figs. 1-10a and b). These two crops were preceded by rainfall
patternsthat were exactly opposite (Fig. 1-10c); i.e., higher early
summer rain followed by lower late summer rain in 1975, and lower early
summer rain followed by higher late summer rain in 1977.
Greenwood (1978) felt that the critical factor determining flowering
was the existence of a quiescent bud for a sufficient length of time
during the growing season to allow differentiation of primordia. Since
male strobili are formed earlier than female, growth in male flowering
buds would need to cease earlier than female flowering buds. This
normally occurs, as male flowers are often found on branches that had
only one cycle of growth the previous growing season (Eggler 1961,
Greenwood 1979). An early growing season drought might discourage
growth on buds normally expected to produce two or perhaps three cycles,
allowing them to differentiate male strobili. This would not necessarily
encourage female strobili if early drought is followed by abundant rain
in late season, as this often causes renewed growth in buds located in
the upper crown (Schmidtling and Scarbrough 1970) and suppresses female
flowering due to late season bud growth. This could explain the opposite
nature of the yearly variation in male flowering as compared with female
flowering in this study.
A drought early in the growing season appears to favor male flower-
ing, while a drought late in the growing season favors female flowering.
Moisture stress might exert its effect simply by limiting vegetative
growth. Because of changes in rainfall patterns, this would cause not
only yearly variation in flowering, but would also affect site variation
caused by water availability.
Large genetic variation exists in fruitfulness in loblolly pines,
with heritabilities ranging from 0.4 to 0.6, indicating that about half
the observed variation is inherent. The biological significance of the
substantial clonal variation is not clear. It seems to contradict the
hypothesized advantage of sexual reproduction under natural forest condi-
tions (i.e., high genetic variability through recombination), since only
a few genotypes would produce a major portion of the reproduction.
Year x clone interactions in male and female flowering would return
some of this variability to the system, because individual trees do not
consistently produce a major part of the seed or pollen every year.
The asynchrony in yearly variation between male and female flowering of
clones provides further yearly variation. As a consequence, large genetic
variation occurs among seed crops.
Evidence from Figure 1-2 shows that this large genetic variation
might not be very important under natural forest conditions. Under
increasing competition, environmental effects seem to increase. However,
high genetic variation might be obtained under some natural conditions
where competition is minimal, such as after a fire, storm, or insect
epidemic where only a few trees are left standing.
,eh-ner or not the high genetic variability is an artifact of
orchard management, it is surely an important factor in tree improvement
for two reasons. First, the number of clones included in many orchards
might be -uch too small. In a 20-clone orchard, there is a good chance
that only four or five of them make a significant contribution to orchard
production. This provides a very narrow genetic base, and also increases
the probability of inbreeding. Second, many clones never produce enough
seed or pollen to justify the expense of maintaining them in an orchard.
If pollen production is used as a criterion, the prospect for identifying
the poorest clones at an early stage is not promising. Clones which are
fruitful early continue to be fruitful. But some that show little
reproductive activity in early years become fruitful later on. Strong
year x clone interactions also make it imperative that fruitfulness be
assessed over a period of several years. This information would normally
be available for cones and seed, but not for pollen which is infrequently
measured. If controlled pollinations for progeny tests are carried out
in the orchard, these data could be easily obtained by assessing relative
production when pollen is collected.
Another consequence of year x clone interactions is the unreliability
of open-pollinated tests of orchard clones, if seeds from different crops
are used in different tests. What might appear as family x site or
family x planting year interactions could be partly due to pollen crop
x year interactions, which cause large differences in male parentage
from year zo year. The same kind of variation could be expected in
various kinds of "check" lots collected in different years.
The original observation that environmental conditions which
favored growth did not necessarily favor flowering was reinforced upon
examination of the data. Tree size and overall vigor were important in
determining the size of the flower crop, but continuous growth in the
year of initiation was unfavorable for large flower crops. Limited
quantities of moisture, especially at critical times during the growing
season, seemed important, and probably caused cessation of vegetative
growth and allowed reproductive development.
There seemed to be more underlying similarities between male and
female flowering than previously supposed. Genetic and environmental
factors favoring female flowering appeared to influence male flowering
as well. However, pollen production seems to be inadequate during early
stages of an orchard.
The difference in the timing of initiation, however, might make it
difficult to manage a seed orchard for optimum production of both male
and female strobili. Continuous drought stress would affect the over-
all vigor of the tree and reduce flowering. May through June stress
appeared to increase male flowering, while July through August stress
increased female flowering. May through August stress might increase
both but would be optimum for neither. If an orchard is to be managed
(including irrigation) for both male and female flowering, it might be
necessary to subdivide the orchard. The perimeter, or perhaps a
portion windward to the prevailing winds at time of pollination, could
be managed for pollen production and the rest for female flowering.
The results of this research indicate that the ideal orchard should
be situated on a well-drained to drought site, with an irrigation system.
In the establishment phase and for 3 or 4 years thereafter, water, ferti-
lizers, etc., should be provided to maximize growth to produce a tree
large enough to provide good crops of female and male strobili. After
the establishment phase, vigorous growth will no longer be necessary or
desirable. In the long run, larger trees would make harvesting more
difficult. In the short run, vigorous growth would take place at the
expense of flowering.
When orchard ramets are 4 or 5 years old, irrigation should be
provided as needed, except for July and August, in that portion to be
managed for female flower production. After the orchard ramets are 6 to
7 years old, the part of the orchard managed for pollen production should
be irrigated except during May and June. The additional period of
optimum growth allows larger trees, which would be necessary for pollen
production. Application of fertilizers should also be timed appropriately:
June for the pollen production portion and August for the seed production
portion. Other management practices such as mowing, subsoiling, and
thinning may also be necessary, as will good insect control.
INHERITANCE OF PRECOCITY IN LOBLOLLY PINE
AND ITS RELATION TO GROWTH
Precocious flowering as well as abundant fruiting has obvious
advantages in breeding programs and in the production of improved seed
(Green and Porterfield 1962, Matthews 1963). It is possible the pre-
cocious flowering has already been selected for unintentionally, as
tree breeders develop cultivated varieties from wild strains. Controlled
mating and artificial regeneration may tend to multiply genes that favor
seed production at an early age. This seems to have occurred in Scots
pine, as the "Nye Branch" variety (Gerhold 1966) flowers precociously.
It is important to distinguish between fruitfulness and "ripeness
to flower." Fruitfulness, which was treated at length in the first part
of this dissertation, is a quantitative trait, as it consists of counts
of reproductive structures. "Ripeness to flower," and precocity, the
early expression of "ripeness to flower," are qualitative traits; i.e.,
either the tree has flowered or it has not. This concept also supposes
that once flowering occurs, it is not reversible, even though flowering
may not occur in subsequent years for various reasons.
Trees that direct a great deal of photosynthetic energy into repro-
ductive development at an early age probably do so at the expense of
growth (Ronberger 1967). The study of the inheritance of fruitfulness
and precocity is, therefore, important not only because of its obvious
involvement with seed production and breeding strategy, but also because
of its possible involvement with .growth.
The present study was undertaken with two objectives in mind:
(1 to determine the heritability of precocity and fruitfulness in
loblolly pine seedlings, and (2) to explore the genetic relationship
between fruitfulness, precocity, and growth.
Most woody plants are unable to flower until they attain a stage
or condition known as "ripeness to flower" (Klebs 1918). Normally a
mature tree is considered as having attained this stage and, in most
cases, does not revert to the juvenile condition when propagated
vegetatively (Schaffalitzky de Muckadell 1959). Vegetative propagation
is the favored reproductive method for establishing seed orchards,
because propagules from mature trees normally flower much sooner than
seedlings of the same species (Barber and Dorman 1964).
All of the treatments and conditions which promote flowering in
mature trees, such as fertilization, crown release, subsoiling, and
drought stress, are effective in young trees as long as they have
attained this "ripeness to flowering." However, these treatments do
not seem to be effective prior'to this stage of development (Robinson
and Wareing 1969).
The stimulation of fruitfulness in woody plants therefore appears
to be divided into two areas: increasing the number of flowers in
mature trees, and shortening the juvenile phase to bring about "ripeness
to flower." Precocity and fruitfulness may very well be related, though
the former is a qualitative trait and the latter quantitative. No
information seems to be available as to the relationship between these
two traits in pines, but in pear and apple seedlings precocity and
eventual productivity were unrelated (Visser et al. 1976).
Ability to shorten the juvenile phase could have great utility in
tree breeding programs, since controlled crosses could be made early in
the life cycle of a tree. This would allow a rapid turnover of genera-
tions and an increase in genetic gain. For instance, Greene (1969)
reported potentially large volume gains by breeding precocious loblolly
Processes governing the phase change from purely vegetative growth
to reproductive growth have been the subject of a great deal of experi-
mentation. Aside from being a feature of advancing age, phase change
seems to be correlated with height in Virginia pines (Bramlett 1971) and
loblolly pines (Schmidtling 1969). Robinson and Wareing (1969) concluded
that the primary factor governing phase change was not dependent upon the
plants passing through a certain number of growing seasons or attaining
a certain size. They experimented with cuttings taken from Ribes nigrum
L. seedlings of various ages, none of which had flowered. They found that
the minimum size for flowering was less for cuttings from older seedlings
than from younger seedlings. They then concluded that phase change
occurred after the plant had passed through a certain minimum number of
mitoses following embryo formation. This was correlated with, but not
determined by, attaining a certain size.
Whether phase change is governed by age, size, or mitotic divisions,
there does seem to be large genetic variation in this trait both between
and within species. Mergen and Koerting (1957) found that flowering
normally begins after about the 5th year in slash pine, but Smith and
Konar (1969) found female strobili in cotyledon stage seedlings of that
There are several accounts of 1-year-old pine seedlings producing
reproductive structures. Nursery-grown seedlings of P. tabulaeformis
Carr. and P. mugo Turra produced male strobili at 1 year (Righter 1939,
Mergen and Cutting 1957), and a P. rigida D. Don seedling of the same age
produced female strobili (Namkoong 1960). Johnson and Critchfield (1978)
found functional male and female strobili on P. contorta x banksiana
hybrids at 1 year of age and presented evidence for the inheritance of
Precocious flowering was found to be inherited in Scots pine by
Gerhold (1966) and in jack pine by Jeffers and Nienstaedt (1972), but
they did not determine the mode of inheritance or its relationship to
growth. Teich and Holst (1969) found that one form of precocity was
related to a cone cluster trait, and that a simple dominant gene was
involved in its inheritance, with a possible involvement of cytoplasmic
inheritance. Wright et al. (1966) found substantial variation between
provenances in precocity, but low family (within provenance) variation,
and no relationship between precocity and growth.
In Virginia pine, which normally flowers at an early age, Bramlett
(1971) found high narrow-sense heritability of fruitfulness in seedlings,
and that flowering trees averaged taller than nonflowering trees. In
another study, Bramlett and Belanger (1976) found that fruitfulness was
highly heritable, but negatively correlated with height in parent-progeny
correlations. In both cases, however, precocity was not measured
directly. Flower counts in young seedlings, really more a measure of
fruitfulness, were used.
In contrast to Bramlett's work, Varnell et al. (1967) found fruit-
fulness to have low narrow-sense heritability (h2 = 0.13). Again their
study measured fruitfulness and not precocity. Unfortunately, most of
the authors of papers dealing with the heritability of fruitfulness
(Bramlett 1971, Bramlett and Belanger 1976, Varnell et al. 1967) relied
on counts of strobili which, even in young trees, would contain com-
ponents of both ripeness (contrast between zero and one or more flowers)
and fruitfulness (number of flowers in flowering trees). The two traits
are obviously related, in the sense that one must occur before the other
can be expressed. But they may be governed by separate genetic and
environmental factors. At the other extreme, the two traits may be one
and the same, the precocious trees being those with genomes for heavy
Different answers to the question of how precocity is inherited may
be obtained if counts of reproductive structures are used as opposed to
proportion of trees flowering. In an analogous situation dealing with
fusiform rust in loblolly pine, somewhat different results were obtained
when proportion infected was used rather than number of galls (Blair 1970).
Visser et al. (1976) were able to deal with this problem by using
age at first flowering as their precocity criterion. Their precocity
trait was not related to the eventual fruitfulness of the seedlings,
suggesting that genetic control was separate for the two traits. This
kind of analysis is not possible in pines, or most other conifers, since
flowering may never occur in many of the trees under plantation conditions
in a relatively long (10 years) experiment.
This necessitated treating "ripeness to flower" as a threshold trait,
and using measures of number of trees flowering rather than age at first
flowering. Relationships among precocity, fruitfulness, and growth are
important in tree breeding programs and have not been adequately determined.
Materials and Methods
Data from two experiments are included in this study, a 10-parent
diallel and a half-sib progeny test.
Diallel. The 10 parents of the diallel were randomly selected
loblolly pines located on the Harrison Experimental Forest in south
Mississippi. These parent trees were crossed in all possible combina-
tions, excluding selfs. The resulting seed were sown in the nursery
in the spring of 1966, and the 1-year-old seedlings were bar-planted in
January 1967 on the Harrison Experimental Forest. Spacing was triangu-
lar, with 9 feet (2.74 m) between rows and 10.39 feet (3.17 m) between
trees in the rows. The experimental design was a randomized complete
block of eight replications with eight trees per plot. Scions from the
parent trees were grafted on potted seedling rootstocks in January of
1967, and the surviving grafts were included in the planting. The grafts
did not grow as well as the seedlings initially, averaging only 1.9 meters
tall at age 3 versus 2.2 meters for the seedlings. In the spring of the
2nd year, each graft was mulched with pine straw and fertilized with 7
pounds (3 kg) of 8-8-8 N-P205-K20 fertilizer. As a result, the grafts
averaged 5.5 meters tall at age 5 versus 3.8 meters for the seedlings.
They were 11.3 meters tall at age 10 versus 10.6 meters for the seedlings.
Female flowers were counted on seedlings and grafts each spring
from 1969 through 1975 (ages 2 through 8 years from planting) except
1973, when measurements on the grafts were not made. Heights were
measured at ages 3, 5, and 10 years, diameters at 5 and 10 years, and
crown width at age 5.
Flowering data were analyzed as binomial traits (zero or one, "ripe-
ness to flower") where a tree was considered "ripe" if ic had flowered
in a given year or in any previous year. These traits were analyzed
untransformed, both on an individual tree basis and on a plot mean basis.
Analysis of a similar trait, fusiform rust, was carried out satisfactori-
ly in this manner (Sohn 1977). Average number of flowers, square-root
transformed, was also analyzed.
The flowering data from the grafts were used only for parent-progeny
regressions. The data from the seedlings were analyzed by a general
least squares analysis program for diallels, "DIALL" (Schaffer and Usanis
1969), on an individual tree basis and a plot mean basis.
The assumptions for the analysis of variance and the genetic variance
expectations are the usual ones for nonrelated, random parents from a
diploid population (Cockerham 1963). Effects accounted for with degrees
of freedom (d.f.) adjusted for missing plots are shown in Table 2-1.
Negative components of variance were handled as recommended by
Thompson and Moore (1963); i.e., a mean square smaller than a prede-
cessor mean square, and whose component was included in it, was pooled
with the predecessor and the result equated to both expectations.
Heritability (h2) estimates were calculated by the formula:
h2 4 o2GCA where the phenotypic variance =
2GCA + 2 + 02
GCA SCA wp
When plot means were analyzed, heritability was computed as:
0GCA where phenotypic variance = GCA +
O SCA + 2p/8.
Table 2-1. Form of analysis of variance and covariance for the diallel
Source D.F. Expected mean squares
ability (GCA) 9 o2 + C2 CA + C3 2GCA
ability (SCA) 35 2 p + C1 2SCA
Error 1361 a2wp
1Error degrees of freedom are adjusted for missing data.
02wp = Variance component due to error. DIALL computed this by sub-
traction, so it was a pooled error term containing both within-
and among-plot variance.
02 = Variance component due to among-plot error in analyses based
on plot means.
O2SCA= Variance component due to specific combining ability.
o2GCA= Variance component due to general combining ability.
The variance component coefficients Cl, C2, C3, determined by DIALL,
were C1 = 30.8, C2 = 32.8, C3 = 249.1 for individual tree analysis,
and CI = 7.2, C2 = 7.4, C3 = 58.2 for plot mean analysis.
The DIALL program computes the standard deviation of the variance
S2 a (MSi)2
S.D. = -
DFi + 2
where the ai are the coefficients of the linear combination of the mean
squares used to estimate the component (Anderson and Bancroft 1952).
Genetic correlations based on GCA components were calculated also.
Half-sib test. This experiment is of interest primarily because
three of the families were selected for precocity. The test consisted
of open-pollinated progeny of 24 trees, 21 of which were selected for
the National Forest System's Southern Region Tree Improvement Program.
Seed from the select trees were collected from the ortets,which were
located in Mississippi, Alabama, Texas, North Carolina, and South
The three trees selected for precocity were from a fertilizer study
located on the Harrison Experimental Forest. Out of 4,000 loblolly pines
in the study, 12 flowered 2 years after outplanting and 63 flowered after
3 years. All were on fertilized plots (Schmidtling 1971) and the enhanced
flowering appeared to be related to attaining a certain height (Schmidtling
1969). The trees used in the present study were subjected to further selec-
tion, as only 3 of the total of 75 precocious trees produced enough seed
in the 9th year to be included in this study. One of these (Pre-l) had
flowered at age 2, the other two (Pre-2 and Pre-3) at age 3. All three
precocious trees were larger than the plot means at 2 years of age as
well as at 9 years of age when seed were collected (Table 2-2). The fact
that they were taller than average at age 9 might have had an important
role in determining their fruitfulness, as there was intense competition
in the original stands at the 10 x 10 foot (3 x 3 m) spacing.
Table 2-2. Height and diameter of the three precocious parent
trees used in the half-sib study compared to the means
of the 100-tree plots where they were located.
Tree Second year Ninth year Ninth year
height height D.B.H.
m m cm
Pre-1 Select tree 0.94 12.2 20.3
Plot mean 0.77 11.7 16.7
Pre-2 Select tree 0.76 -1 19.1
Plot mean 0.64 10.1 14.9
Pre-3 Select tree 0.73 11.7 15.5
Plot mean 0.68 11.1 15.4
Seed collected from the 24 families were sown in the nursery in the
spring of 1970, and the 1-year-old seedlings bar-planted in the winter
of 1970-1971 at two locations on the Harrison Experimental Forest. Four
replications of a randomized complete block were planted at each location,
with four trees per plot and 10 x 10 foot (3 x 3 m) spacing. In spring
of 1977, female flowers, conelets (1-year-old strobili), and cones were
counted and height and D.B.H. were measured. At 4 and 5 years, fruitful-
ness was assessed by cone and conelet counts, respectively. This was
conservative, because some trees without cones or conelets might have had
flowers which aborted.
Mean number of trees flowering per plot was analyzed, and the analysis
took the form:
Source of variation D.F.
Blocks in location 6
Families x location 23
The fixed model was assumed, and statistical significance was tested at
the 0.05 level of probability.
Results and Discussion
Diallel. Approximately 5% of the seedlings flowered in 1969, or
after 2 years in the field. Those are the ones that were considered
precocious. Number flowering increased to over 30% in 1970, and 80% of
the trees had shown some signs of reproduction (age 8) by 1975.
Seedlings actually started flowering before the grafts (Fig. 2-1),
as there was almost no flowering among the grafts in 1969. Undoubtedly
because of physiological maturity and special treatments the grafts
received, they subsequently flowered better than the seedlings. They
averaged over 10 female strobili each in 1974, whereas the seedlings
averaged only about 2.0 each. Yearly variation in flowering of the seed-
lings was approximately parallel to that of the grafts.
Seedlings flowering in 1969 were taller than average at age 3 (Fig.
2-2a). The difference, although statistically significant, was very
small (2.3 m versus 2.2 m for the overall average). Precocious individuals
were still slightly taller than average at age 5, but by age 10 they were
slightly shorter. Differences at ages 5 and 10 were not statistically
significant. Precocious trees continued to flower better than the
average (Fig. 2-2b). In 1973, the difference was largest, when the
precocious trees averaged 12 flowers per tree and the others averaged
only 6. Though the absolute difference varied widely, precocious trees
had about twice as many flowers in all years measured. Possibly the
precocious trees lost their initial height advantage because their energy
was diverted into reproduction.
2 3 4 5 6 7 8
Flowering of the parent grafts and their seedling
progeny in the diallel experiment. Vertical axis
is square-root scale.
0 0: Grafts
* -- *: Seedlings
T,- .>.. 1* '
3 4 5 6 7 8 9 10
3 4 5 6 7
8 9 10
Growth and flowering of the precocious trees
(flowering at age two) compared to the non-
precocious trees in the diallel experiment.
A. Height. B. Flowering. Vertical axis of
B is square-root scale.
- : Precocious
The diallel analysis showed that all flowering characters exhibited
some degree of heritability (Tables 2-3 and 2-4). These heritabilities
are rather limited in application, and probably could not be used to
accurately predict gain in many breeding situations. They were undoubted-
ly biased upwards because genotype x environment interaction was not
estimated. Exact values are not important, but their relative magnitude
is of interest.
General combining ability (GCA) values for flowering were all more
than two standard deviations above zero, except for 8th-year ripeness on
an individual tree basis (Tables 2-3 and 2-4). Specific combining ability
(SCA) values for flowering traits were generally much smaller than GCA
values. They differed from zero by two standard deviations only in
three instances: 8th-year "ripeness," average number of flowers from
individual tree data (Table 2-3), and average number of flowers from plot
mean data (Table 2-4). Heritabilities ranged from 0.134 for 2nd-year
"ripeness" to 0.609 for average number of flowers on individual tree data.
Values for plot mean data were higher, ranging from 0.409 for 8th-year
"ripeness" to 0.630 for average number of flowers. Thus, average number
of flowers, a quantitative trait, was highly heritable both on an individ-
ual tree basis and on a family basis. This agrees with findings from the
previous chapter. Precocity was much less heritable on an individual
basis, but was moderately heritable on a family (plot mean) basis.
Diallel statistics based on individual tree data for
flowering and growth traits. Standard deviations
are shown below general combining ability (GCA) and
specific combining ability (SCA) statistics.
Trait GCA SCA Error h2
Average number of
Having flowered by the year indicated
2 Over the eight years measured
Table 2-4. Diallel statistics based on plot mean data for
flowering and growth characters. Standard deviations
are shown below General combining ability (GCA) and
Specific combining ability (SCA) statistics.
Trait XCA SCA Error h2
Average number of
SHaving flowered by the year indicated
2 Over the eight years measured
Heritabilities of the growth variables were much lower than those
for flowering (Tables 2-3 and 2-4). Only the GCA for crown width
differed from zero by more than two standard deviations. Heritabilities
for height seemed to increase :.ith time, however. They reached 0.299 by
the 10th year for individual tree height. In contrast to flowering,
SCA values differed from zero by two standard deviations for several
characters: 5th-year D.B.H., crown width, 10th-year height, and 10th-
year D.B.H. on an individual basis, and for 10th-year height and 10th-
year D.B.H. on a family basis. These results agree fairly well with
those of Snyder and Namkoong (1978) in a longleaf pine diallel. Their
heritabilities for growth traits were in the same range as those found
here. They also noted significant SCA effects for many traits.
Genetic correlations among flowering traits were uniformly high
and positive (Table 2-5), ranging from a low of 0.780 between average
number of flowers and 2nd-year "ripeness" to a high of 1.0 between
average number of flowers and 3th-year "ripeness." Individual tree and
family correlations for each pair of traits were very close, as they
Genetic correlations between all growth variables and flowering
variables were negative except between 8th-year ripeness and 10th-year
D.B.H., which was essentially zero on a family basis (r = 0.065). Corre-
lations were generally not very strong. But uniformly negative signs
indicate that selection based on fruitfulness or precocity alone would
result in some loss in growth. The strongest genetic correlations were
between 2nd-year "ripeness" (precocity) and various growth traits. They
ranged from r = -0.127 between precocity and height at 3 years for in-
dividuals to r = -0.654 between precocity and crown width for plot means.
Genetic (GCA) correlations among flowering variables
and flowering and growth variables from the diallel
analysis. Individual tree correlations are shown
with correlations based on plot mean data below
individual tree correlations in parenthesis.
2nd year 3rd year 8th year number
"ripeness" "ripeness" "ripeness" of flowers
1Having flowered by the year indicated
2 Over the eight years measured
3 Correlation based on individual tree data
4 Correlation based on plot mean data
So even though precocious trees were significantly taller at age 3,
this correlation was environmental since the genetic portion of the
overall correlation was negative. Negative correlations between flower-
ing traits and crown width were surprising, because it seems logical to
assume that trees with larger crowns would have more cones. In this
study, however, there was no genetic evidence for such an assumption.
Parent-progeny relations. Regression between mid-parent average
flowering (average of the two parents for each cross) and progeny average
flowering yielded a heritability of 0.518 (Fig. 2-3). This was slightly
less than the heritability found in the diallel analysis, h2 = 0.609
(Table 2-3). However, the most fruitful parents, in a quantitative sense,
were not the ones which were producing the most precocious progeny. This
was evident in Table 2-6. Parent one ranked fourth out of 10 for average
flowering, but produced the most precocious progeny, 14.7%. The correla-
tion between mid-parent average flowering of the grafts and precocity of
the progeny was only r = 0.355. This increased sharply with subsequent
measures of "ripeness" to a peak of r = 0.680 the 5th year (Fig. 2-4).
Correlations between mid-parent fruitfulness of the grafts and
various growth measures of the progeny were all negative except for the
correlation between 10th-year D.B.H. and parent fruitfulness (r = 0.134).
Correlations between parent fruitfulness and 3rd-year height, 5th-year
height, 5th-year diameter, and 10th-year height were all negative though
weak, ranging from r = -0.241 for 3rd-year height to r = -0.020 for 5th-
year diameter. Though probably not very important, correlations of that
magnitude reinforce the findings of the diallel analysis in cautioning
against selection for fruitfulness without considering other aspects.
* b = h2 = 0.518
1 5 10 12
Average Female Strobili of M.d-parent o.
Parent-progeny regression for average flowering,
for seven years in the diallel experiment. Both
axes are square-root scale.
Average flowering of the grafts compared to precocity
and average flowering of the progeny in the Diallel
Parents (grafts) Progeny
Family Average Flowering Precocious Average Flowering
No. % No.
7 14.62 12.7 2.54
9 12.04 3.2 2.05
4 4.09 5.2 1.5
1 3.52 14.7 2.35
8 3.40 6.1 1.77
6 1.93 4.4 .95
3 1.73 6.3 .91
5 1.22 1.4 .77
10 0.83 2.0 .80
2 0.65 2.9 .89
1Flowered at age two
Flowered at age two
2 3 4 5 6 7 3 k:g.
Change in correlation with time between average
fruitfulness of mid-parent and "ripeness to
flower" of progeny. Last point marked "avg" is
the correlation between mid-parent fruitfulness
and progeny average fruitfulness.
Half-sib test. About 12.5% of the trees in the half-sib test flower-
ed the 4th year (Table 2-7). Flowering increased to 15.6% the 5th year
and to 26.3% the 6th year. Results of selection for precocity were evident.
An average of 32.6% of individuals from the three precocious families
flowered by 4 years compared with an average of 9.6% of the other families.
Over 40% of Pre-1 (precocious family number one) progeny flowered by age 4.
Only one family, Bud 12, equalled flowering of the poorest precocious
family, Pre-3. Differences were statistically significant at all years
and indicated that selection for precocity will result in early flowering
Growth of precocious selections was good, contrary to what one would
expect from the negative correlations between flowering and growth in the
diallel analysis. Precocious trees were all larger than average when
selected, however (Table 2-2). Their progeny averaged 5.7 meters tall
compared with 5.4 for the others, and 9.3 centimeters in diameter compared
with 8.6 for the others. One precocious family, Pre-2, ranked first in
diameter out of 24 families. This ranking was remarkable since the 21
nonprecocious families were intensively selected, and size was an important
selection criterion. But geographic variation may be an important factor
in this study, as 11 of the select trees were from areas of slower growing
provenance; i.e., the Tal, Ban, Tex, and Sum sources from north and central
Alabama, Texas, and South Carolina piedmont, respectively (Wells 1969, Wells
personal communication). By excluding the latter, overall mean height of
the remaining select trees was 5.88 meters and D.B.H. was 9.4 centimeters,
slightly larger than precocious selections. But dropping these from the
data does not change differences in flowering between precocious and
Flowering and growth of the 24 families in the half-sib
Tree Flowering By Age Sixth Year Sixth Year
Family 4 years 5 years 6 years Height D.B.H.
Precocious trees performed better than expected, however, considering
negative relationships that were found between growth and flowering in the
diallel. As stated previously, the two-stage selection process probably
favored fruitfulness as well as growth, since only dominant trees would be
expected to flower under the very close crown competition in which they
grew at age 9. In fact, the three precocious trees used were larger than
their neighbors at age 9 (Table 2-2).
Precocious flowering appears to be moderately heritable, and selec-
tion for this trait can reduce the age of first flowering. Fruitfulness
appears to be highly heritable, and the broad-sense heritability estimates
of the previous chapter for fruitfulness appear to be on the low side.
They probably are biased downward by rootstock effects. Estimates of
broad-sense heritability include all genetic variation and should there-
fore be the upper limit for narrow-sense heritability. The narrow-sense
heritability estimate of 0.609 from the diallel analysis for average
flowering on an individual tree basis is higher than many of the
estimates for broad-sense heritability from the previous chapter. A
broad-sense estimate from Table 2-3 would be:
h2 = 4(GCA + SCA)/phenotypic variance = 0.715
which is greater than any of the estimates from the clonal data.
The previous chapter concluded that size and overall vigor were
positively related to flowering, but that prolonged vegetative growth
discouraged reproductive development. Reproductive biologists tend to
think in terms of "a transition" from vegetative growth to reproductive
growth as part of an overall reproductive strategy (Cohen 1976). Obviously
this does not happen in most forest trees, as growth and reproduction occur
within the same growing season. There is no complete transition from
vegetative to reproductive growth as occurs in most herbaceous annuals.
Greenwood (1978) feels that the primary reason seedling pines do not
flower is that they grow almost continuously during the growing season.
He feels that a "quiescent" bud must be formed early in the season to
allow strobilus initials to form.
Thus, whether one views the commencement of reproduction as a change-
over from vegetative growth or as a process where a reduction in vegetative
growth allows reproduction, a negative correlation between reproductive
growth and vegetative growth should be expected. Although this study was
limited by the fact that the conclusions are mainly based on an intensive
analysis of only 10 parents, the expected negative relationship between
growth and repoduction does seem to have a genetic basis in loblolly pines,
which would seem to mitigate against selection based on flowering alone.
The greater height of the precocious trees at age 3 was apparently an
environmental relationship. The genetic correlations between precocity
and all growth variables, including height at 3 years, though weak, were
The fact that precocity and growth are only weakly correlated suggests
that one could select for both of these traits simultaneously in a breeding
program. The use of genetically induced precocity for obtaining large
gains in growth by shortening generation time, however, hardly seems
SU2MARY AND CONCLUSIONS
Genetic variability in fruitfulness is certainly an important
consideration in planning loblolly pine tree improvement programs. It
is evident from this study that a relatively small proportion of orchard
clones produce most of the progeny. Most of the genetic variation is
additive, indicating that the use of seedling seed orchards would do
little to solve the problem. In terms of genetic variability, the
effective number of clones in a seed orchard may be less than half the
actual number and could be further reduced by differences in reproductive
phenology. In addition, the negative relationship between cone production
in the parent trees and growth of the progeny indicates not only that few
clones will be producing most of the seed, but that these are not the
best clones with regard to progeny growth.
Under these conditions, early progeny testing assumes even more
importance than previously. Including more selections in breeding
programs also appear warranted. The ideal solution to the problem
would be to find a way to increase flowering in the unfruitful clones.
Unfortunately, the more fruitful clones are the ones which respond best
to treatments such as fertilization. Beers' (1974) suggestion for
fertilizing clones individually by "prescription" based on their previous
response in seed production should, perhaps,be modified to include progeny
test information in the form of an index similar to a selection index.
Relative response to fertilizers, in terms of increase in pounds of
seed produced by each clone, would be included as well as relative
growth of their progeny, to optimize economic gain.
Using precocious individuals to shorten generation time and in-
crease genetic gain should be carried out with caution. In some breeding
programs, selection as early as age 7 is being practiced. It would be
tempting, but risky, to choose trees above average in growth from progeny
tests primarily because they flower early. If precocious individuals are
selected which are very much above average in growth, however, the risks
are probably not great, since the heritability of precocity is only
moderate to weak. Also, the genetic correlation with growth is not
strong. Selection based on precocity alone seems to be clearly un-
There is ample evidence from previous studies showing that abundant
production of seed reduces vegetative growth. The converse also appears
to be true; i.e., reduced vegetative growth enhances the production of
seed. The best conditions for vegetative growth appear to be less than
optimum for reproductive growth. Some sites which have been carefully
chosen for seed orchards may be too good. Previously, it was assumed
that the best sites for vegetative growth would be the best for seed
production, but it is apparent that conditions which favor cessation of
growth early in the season are best for flowering.
Silvicultural treatments such as fertilization and irrigation need
to be optimized to favor vegetative growth in the establishment phase,
as size is very important in determining fruitfulness. After the orchard
is in production, these silvicultural treatments need to be changed to
optimize reproductive growth. This may require an adjustment in the
attitude of seed orchard managers.
Anders:n, R. L., and T. A. Bancroft. 1952. Statistical theory in
research. McGraw-Hill, New York. 399 p.
Andersson, E., and H. H. Hattemer. 1975. Growth and flowering of pri-
mary and secondary grafts of Scotch pine. Silvae Genet. 24:49-54.
Barber, J. C., and K. W. Dorman. 1964. Clonal or seedling seed orchard?
Sil-ae Genet. 8:11-17.
Barnes, B. V. 1969. Effects of thinning and fertilizing on production
of western white pine seed. U.S.D.A. For. Serv., Intermt. For. and
Range Exp. Stn. Res. Pap. INT-58, 14 p.
Barnes, R. L., and G. W. Bengtson. 1968. Effects of fertilization,
irrigation, and cover cropping on flowering and on nitrogen and
soluble sugar composition of slash pine. For. Sci. 14:172-180.
Becker, W. A. 1967. Manual of procedures in quantitative genetics.
2nd ad. Wash. State Univ., Pullman. 130 p.
Beers, W. L., Jr. 1974. Industry's analysis of operational problems
and research in increasing cone and seed yields. In Seed Yield
from South. Pine Seed Orchards. J. Kraus, ed. Proc. Colloq. Ga.
For. Cent., p. 86-96.
Bergman, A. 1968. Variation in flowering and its effect on seed cost.
N.C. State Univ., Sch. For. Resour. Tech. Rep. 38, 63 p.
Blair, R. L. 1970. Quantitative inheritance of resistance to fusiform
rust in loblolly pine. Ph.D. Thesis, N.C. State Univ., Raleigh, 87 p.
Bradstreet, R. B. 1965. The Kjeldahl method for organic nitrogen.
Academic Press, New York and London.
Bramlett, D. L. 1971. Correlations between reproductive and vegeta-
tive growth in a 6-year-old Virginia pine plantation. U.S.D.A.
For. Serv., Southeast. For. Exp. Stn. Res. Pap. SE-88, 6 p.
Bramlett, D. L., and R. P. Belanger. 1976. Fertilizer and phenotypic
selection increase growth and flowering in young Virginia pine.
For. Sci. 22:461-467