RETURNING MATERIALS:
IV1£31_} P1ace in book drop to
[JBRARJES remove this checkout from
“ your record. FINES will
be charged if book is
returned after the date
stamped below.

on E moi

 

 

 

 

3:35"? 5 eat.

 

 

 

AN INVESTIGATION OF YIELD COMPONENTS
IN 'MONTMORENCY' AND 'METEOR'

SOUR CHERRY

BY

LOONG-SHENG CHANG

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER of SCIENCE
Department of Horticulture

Plant Breeding and Genetics

1986

Abstract

An Investigation of Yield Components in 'Montmorency'
and 'Meteor' Sour Cherry

BY
LOONG-SHENG CHANG

Two sour cherry (Prunus cerasus L.) cultivars,
'Montmorency' and 'Meteor' were evaluated over 2 seasons to
determine the relative importance of different components.

A path coefficient analysis was performed to determined the
direct and indirect effects of primary, secondary, and tertiary
components on limb yield. Fruit number, fruit weight, the
number of lateral buds and spurs, and fruit set were found to
be the most important components affecting limb yield in both
cultivars. However, the fruiting habit of the 2 cultivars was
significantly different. 'Montmorency' produced 68% of its
fruit on lateral buds on one-year-old wood, while 'Meteor' had
70% of its fruit on two-year-old spurs. When the data was
standardized by dividing by limb cross-sectional area,
'Meteor' had a higher flower bud density and yield efficiency
(grams of fruit/cross-sectional area) than 'Montmorency'.

Although 'Meteor' had higher limb yields than
'Montmorency', the 'Montmorency' trees sampled had
approximately 4 times more limbs than 'Meteor' and therefore
higher tree yields.

Yield prediction equations for 'Montmorency' and 'Meteor'

were developed using the yield component factors included in
the study. With these equations, 'Montmorency' tree yield could
be predicted with 99.5% accuracy, while 'Meteor' tree yield

could only be predicted with an accuracy of 30%.

ACKNOWLEDGMENTS

I would like to express my sincere appreciation to my
advisor Dr. Amy Iezzoni for her guidence, and support. I
also thank the members of my advisory committee for their
helpful suggestions and corrections.

I also appreciate the helpful discussions concerning
the path diagram with Dr. M. Wayne Adams, and statistical
analysis with Dr. Marvin Pritts.

Finally, I would like to thank my parents for financial
support and my wife Ru-fen for her patient and unwavering

support.

TABLES OF CONTENTS

page

LIST OF TABLES

LIST OF FIGURES

CHAPTER ONE: AN INVESTIGATION OF YIELD COMPONENTS IN
'MONTMORENCY' AND 'METEOR' SOUR CHERRY --- 1
Introduction ----------------------------- 2
Literature Review ------------------------ 4
Materials and Methods -------------------- 12
Results and Discussion ------------------- 15
References ------------------------------- 28

CHAPTER TWO: YIELD PREDICTION FOR 'MONTMORENCY' AND
'METEOR' SOUR CHERRY USING STEPWISE
REGRESSION ------------------------------- 33
Introduction and Literature Review ------ 34
Materials and Methods -------------------- 39
Results and Discussion ------------------- 41

References -------------------------------- 46

LIST OF TABLES

Table page

CHAPTER ONE

Mean values (x) and coefficients of variation (cv)
for yield components from limbs of 'Montmorency' and
'Meteor' in 1983 and 1984 —— 16

 

Mean values (x) and coefficients of variation (cv)

for limb cross-sectional area (cm ), flower bud number
per limb, flower bud density, crop density, and yield
efficiency for 'Montmorency' and 'Meteor' sour cherry

in 1983 and 1984 ----------------------------------- 17

Mean values of primary yield components associated
with spurs and lateral buds for 'Montmorency' and
'Meteor' in 1983 and 1984. ------------------------- 20

Path coefficients showing direct and indirect effects

of fruit number and fruit weight on limb yields of
'Montmorencty' and 'Meteor' sour cherry. Yield is
divided into the yield from lateral buds and spurs -- 22

Path coefficients showing direct and indirect effects

of secondary yield components on fruit number of
'Montmorency' and 'Meteor' sour cherry in 1983 and

1984. Yield is divided into that from lateral buds

and spurs ----------------------------------------- 23

Path coefficients showing direct and indirect effects

of leaf number per limb and leaf size in 1984 on fruit
weight for 'Montmorency' and 'Meteor' sour cherry in

1984 ---------------------------------------------- 25

Path coefficients showing direct and indirect effects
of leaf number and size per limb in 1983 for
'Montmorency' and 'Meteor' sour cherry ------------ 26

CHAPTER TWO

Mean values (x) and coefficients of variation (cv)
for variables sampled from 'Montmorency' and
'Meteor' ------------------------------------------ 42

Validation of the prediction equation between
harvested yield and calculated yield from 14
'Montmorency' trees ranging from 7 to 23 years old -- 43

Validation of the prediction equation between
harvested yield and calculated yield for 5 'Meteor'
trees either 7 or 8 years old --------------------- 45

List of Figures

CHAPTER ONE
Figure page

1 Causal system of path-coefficient analysis used in
this study ---------------------------------------- 14

2 Yield for 8 and 9 years old trees of 'Montmorency'
and 'Meteor' in 1984 partitioned by the age of wood.
Yield is expressed as kg per tree. ---------------- 19

Chapter one
An Investigation of Yield Components in 'Montmorency' and

'Meteor' Sour Cherry

2

Introduction

Michigan's cherry industry produces about 70% of the
nation's sour cherry supply. In 1984, Michigan produced 210
million pounds of the approximately 271 million pounds of
sour cherries produced in the United States. However, almost
all the sour cherry production is based on one cultivar,
Montmorency, which represents 97% of the sour cherry acreage
in the United States (4).

'Montmorency', a cultivar which originated from France
about 300 years ago (24), was first studied at the Michigan
Agricultural Experimental Station in 1922 (18). Much of the
work focused on fruit set because of the possibility of
increasing fruit set by cultural factors and because research
indicated that increased fruit set might result in higher
yields (19,22,36,37,38,40).

Little work has been done in the United States on yield
component analysis comparing sour cherry cultivars. Roberts
(37) reported that the higher yield of 'Montmorency' compared
to 'Richmond' was the result of higher fruit set on
'Montmorency' shoots and spurs vs 'Richmond' shoots and
spurs. However, 'Richmond' has a similar French origin as
'Montmorency'. All of the additional literature pertains to

'Montmorency'.

3

Path analysis can be used in plant breeding to measure
each component affecting a complex trait such as yield.

Those characters can be identified which would be most
effective as selection criteria to improve crop productivity.
So far, no studies have been reported which discuss the
influence of individual yield components in sour cherry.
Approximately 70% of the fruit on mature 'Montmorency' trees
is produced on last year's shoots. However, sour cherry
cultivars which bear most of their fruit on spurs have been
reported (3,25). For ideotype breeding, it would be useful
to have a model of the most productive orchard canopy which
would include components ranging from the number of leaves
per tree to fruit size and number. This analysis could
ultimately be related to total yield per acre.

In this study, 2 cultivars of sour cherry, 'Montmorency'
and 'Meteor', were evaluated. 'Montmorency' was chosen because
it represents approximately 97% of the sour cherry acreage in
the United States, and 'Meteor' was chosen because
observations indicated that its fruiting habit is different
than 'MOntmorency'. The objectives of this study were: 1)
determine the relative importance of different yield
components for 'Montmorency' and 'Meteor', 2) measure the
effects of the individual components on yield, and 3)

describe the morphological basis of these differences.

4
Literature review

A. Factors affecting sour cherry yield.

Bradbury (7) reported that 30-50% of the flowers on
'Montmorency' sour cherry trees produced fruits. Flower and
fruit drop in sour cherry can occur at any of three stages.
1) One week after bloom, the flowers may drop due to a
defective pistil with the pedicel or peduncle attached.

2) Two weeks after full bloom, the fruit with the pedicel
attached may fall due to ovule degeneration. Pollination may
have occurred but not fertilization. 3) Three to five

weeks after full bloom, the endosperm may fail to develop and
the fruit abscises with the pedicel attached.

Gardner (19) identified fruit set as one of the factors
influencing 'Montmorency' tree yields. He concluded that the
low yields on 'Montmorency', obtained following profuse
blossoming, were due to poor fruit set. However, the
correlation between fruit set and yield, which was calculated
using data in the paper, was 0.468 and not significant at the
5% level. This lack of association between fruit set and
yield at the 5% significance level reflects the complex set
of factors which influence 'Montmorency' yields.

Numerous cultural and environmental factors have been
reported which affect sour cherry flower number and fruit
set. Roberts (36) stated that 'Montmorency' fruit set was
higher when 'Early Richmond' was used as a pollen source.

Shoemaker (40) found that 'Montmorency' trees caged with bees

5
had 20% higher fruit set than the control. However, Roberts
(37) concluded that poor yield was not due to poor
pollination and that sour cherry being self-fertile, commonly
set fruit without insect pollination. Gardner (l9) believed
that the difference in the abundance of pollinating insects
was the major reason for the difference in fruit set in his
experimental orchards. Redalen (35) investigated fruit set in
35 sour cherry cultivars and concluded that 'Montmorency' was
only partly self-compatible thereby suggesting a genetic
reason for the reduced fruit set in 'Montmorency'.

Roberts (37) and Roberts and Langord (38) reported that
shading cherry trees with burlap cages gave a significant
fruit set reduction compared to unshaded trees, 0.92% to
18.92% respectively. Gray (22) showed reduction in
'Montmorency' fruit set due to shading with screen, muslin,
and burlap. However, Langord (29) reported that poor light
conditions could reduce fruit set because of reduced
photosynthesis.

The effect of temperature on fruit set was also
apparent in Gray's work (22). Based on the data from Gray's
paper, I calculated the mean temperature during the 10-day
bloom period and related that to flower count and fruit set
in burlap shade and the unshaded control. The lower
temperature 10 days after bloom in 1931 compared to 1930 and

1932 may have resulted in the reduced fruit set. In Gray's

6
Year Mean temp.( C) Flower count Fruit set(%)

burlap control burlap control

1930 16 1236 3191 20.0 25.5
1931 11 1917 2392 10.0 20.1
1932 15 2040 2289 16.5 32.5

experiments, only one tree was sampled for each shading
observation, resulting in large variation due to tree and
year effects. Eisensmith et al (13) and Flore (15,16) studied
the light interception effect on cherry tree growth. They
found that shading begun during stage II of fruit development
significantly reduced flower bud differentiation when the
reduction was under 20% of full sunlight. This low level of
full sunlight frequently occurs in mature Montmorency trees
under field conditions.

Fruit set and yield in sour cherry is influenced by
pruning practices and fertilization. Kenworthy (27) showed
that a 1.8kg nitrogen application resulted in a 13.6kg
increase of fruit per tree. The nitrogen effect on
Montmorency yields may be due to the positive effect on
vegetative growth. Kesner et al (28) suggested that summer
hedging increased fruit set and yield but did not affect the
number of flowers per bud. However, Flore (15) did not
detect any change in canopy light pentration after 3
consecutive years of summer hedging. Bradbury (7) mentioned

that early thinning could increase sour cherry fruit set.

7
Diaz (9) thinned 'Montmorency' flowers to different numbers
and at different stages and found no evidence of competition
between and within flower clusters on fruit set. Also the
next year's fruit load was not affected by the thinning
treatments. He did not identify any effects influencing
fruit set after pistil differentiation; however, he observed
that at bud swell tetrad formation in the anthers occurs
consistently 2 days earlier on the most expanded flowers
while embryo-sac degeneration is significantly more frequent
in the least advanced flowers.

B. Statistical analysis of yield components.

Yield is a complex quantitative character influenced by
numerous factors. In 1923, Engledow and Wadham (14) proposed
that yield was the product of the number of grains per ear and
the average weight of a single grain. Additionally, they
suggested that selection for these components rather than yield
itself would accelerate the gain from selection in a breeding
program. However, yield is not a simple function. Instead it
is the product of complex factors occuring throughout plant
development. Therefore, simple correlation between each factor
and yield may not be the best approach to describe the relative
contribution of each component to yield.

In 1921, Wright (43) published his path coefficient
analysis which partition the correlation into direct and

indirect effects. The path coefficients are calculated as

8
standard partial regression coefficients. The direct effect is
the influence of one component on another component without
considering the interaction between components. The indirect
effect is the difference between the correlation and the direct
effect. Li (31) described the features and application of path
coefficient analysis. Since Li's introduction, path analysis
has been frequently used in crop plant breeding to measure the
relative contribution of each component to complex traits, such
as yield.

Grafius (20) developed the geometrical concept of
yield components in oats. Oat yield was described as the
products of the number of panicles per unit area, the
number of kernels per panicle, and the average weight of
each kernel. Leng (30) found the heritability of yield
components to be much higher than the heritability of yield
alone in maize. This concept enables the plant breeder to
separate yield into the product of its part and then choose
parents selected for their independent yield component
superiorities.

In 1967, Adams (1) introduced the concept of yield
component compensation to explain the common failure of
selection based on yield components. Yield compensation,
expressed as negative correlations between components, occurs
if the input of metabolic products to the component system is

limiting in the developmental sequence resulting in

9
competition for these metabolic products (1,2). Nickell and
Grafius (32) found that component compensation could retard
the genetic advance in winter barley selection, Grafius and
co—workers (21,41) developed approaches to the mathematical
analysis of yield component relationships by the standardized
the original data using log transformation since there were
large genotype x environment interactions associated with
these components.

Path analysis has been used in agronomic studies to
identify components having strong direct effects on yields.
Dewey and Lu (8) used path analysis to determine components
affecting crested wheatgrass seed yields. They found that
fertility and plant size were major factors. Duarte and Adams
(10) used path analysis to study the effect of the primary
components on dry bean yield and the secondary components
(leaf size and leaf number) on the primary components. They
found that leaf size contributed to seed weight and that leaf
number was correlated with pod number; however, the number of
pods per plant was the most important factor affecting yield.
Bhatt (5) showed that spike number and kernel weight exerted a
predominent effect on spring wheat yield, but the residual
factors in the path analysis were relatively large indicating
that other traits should be considered. Pandley and Torrie
(33) concluded that selection for high pod and seed number
were the most critical factor for increasing soybean yield.

They also found negative correlations existing among yield

10
components. In gggsgca, Thurling (42) also reported yield
component compensation with seed number per pod negatively
correlated with pod number per plant. Borojesvic and Williams
(6) stated that the spike number was the most important
contribution to wheat yields, but the environmental
variability for spike number was greater than the genetic
variability for spike number. Kang et al. (26) concluded that
stalk number and stalk diameter were more important than plant
height in determining sugarcane yield.

Path coefficient analysis has been applied less often to
horticultural crops so other types of analysis have been
employed to measure yield component interaction. For example,
Eaton and MacPherson (11) used stepwise regression to estimate
the relative contribution of several variables to cranberry
yield. They found that the number of flowering uprights per
unit area made a major contribution to yield. Using stepwise
regression the variation of the number of flowering uprights
per unit area contributed about 80% to cranberry yields.
However, if the number of berries per flower was included in
their analysis, the relationships changed and the number of
berries per flower contributed about 43% to yield as compared
to 29% for the number of flowering uprights per unit area.
Stepwise regression could not precisely determine the direct
effect of these components. Later Eaton and Kyte (12) using
multiple regression concluded, that not flowering uprights but

fruit set was also a major factors influencing cranberry

11

yield. In their paper, fruit set and flowering upright number
compensated for each other; however, the correlations for
those negative significant relationships between components
were not presented.

Hancock et al.(23), Pritts and Hancock (34), and Siefker
(39) used path coefficient analysis to determine the factors
contributing to strawberry and bluebery yields. Crown and
fruit number played a significant role in strawberry
production. The number of buds per cane exerted the most
significant effect on blueberry yields. Canes per plant and

fruit set also played an important role in blueberry yields.

12

Materials and Methods

In 1983 and 1984, data was taken from three
representative trees each of 'Montmorency and 'Meteor' at 2
locations, Clarksville Horticultural Experiment Station,
Clarksville, Mich., and Hilltop Orchards and Nurseries,
Hartford, Mich.. The trees at Clarksville and Hartford were
7 and 8 years old in 1983, respectively. The two cultivars

were grafted on P. mahaleb seedling rootstocks and

trained to a modified central leader.

One 3 or 4-year-old representative limb was randomly
selected from each tree and the following totals were
counted in both 1983 and 1984: number of flowers, number of
flower clusters, number of lateral flower buds, number of
spurs, number of flowering branches, and limb diameter.
Three limbs within a tree were also selected and the number
of leaves along the main branch of each limb were counted.
Five leaves from each limb were randomly selected for leaf
area measurement in cm, using a L1-3000 leaf area meter(cm ).
Fruit at maturity was counted and weighed for each limb
and kept separate within the limb by age of wood. Total
tree yield were taken at both locations in 1984.

Path analysis (23,34) was used to calculated the
relationships between yield components using the causative

relationships diagrammed an Fig. 1. The path coefficient

analysis partitions the correlation into the direct and

13
indirect effects. The direct effect is the influence of one
component on another component without considering the
interaction between components. The indirect effect is the
difference between the correlation and the direct effect.

The significance of each path coefficient was analyzed with

an F-test.

14

 

moo. em aEm
EV \ou_u woo
.
_
:
r moo—:_ ASL.
3!! \.0c cued

n.cocon60u
>.o_.50h

A!!!! !!!!!!

.uoaeo .uo.16c_

‘lllrlll .uoto .0010

c. on_n wood

I
!
!
!
I
W’
V
m
o

woo. :_
aE__\.0c mood

r----- -----I

..3 ._=.m

.1IIV toeua.u\

 

 

   

V n.039_* *0 .02

6.0.» ne_z

 

.l..! Uafi _O..0~O- LO 5:Qu\

.uuV «Lots: .030: e0 .02

.0: ._:.m

ne__\.v:a .oro.o_
pea «teen *0 .oz

necocanOu
xeoE_ee

n.cocoe600
xeoucOuow

Spa.» at... .: u_»>_uco .comumemoOung.on e0 Emir: .onaoU .— .omu

15

Results and Discussion

Sour cherry buds are simple, producing either floral or
vegetative growth. Therefore the number of flower buds on 1-
year-old-wood reduces the number of nodes which become spurs
or lateral branches the following season. In 1983, an
average of 152 lateral buds flowered on the sampled limbs of
'Montmorency' (Table 1). In 1984, the mean number of
flowering spurs on those limbs was only 27. In contrast, the
mean number of flowering lateral buds on the sampled limbs of
'Meteor' was 26 in 1983 and the following year the mean
number of flowering spurs was 168. The number of flowering
lateral buds and spurs on 'Montmorency' has been reported to
be influenced by vigor (l9,27,36,37). If tree vigor is moderate
to low, where shoot growth is less than 25.4 cm., then the
majority of the lateral buds are floral buds. As vigor
increases to more than 45.7cm,, more buds on the shoot remain
vegetative producing spurs at the basal portion of the shoot.
Although very few flowers are produced, an increased bearing
surface is formed for the next year. Fruiting spurs develop
on these 1-year-old branches. For the 2 years, 1983 and 1984,
the average terminal shoot growth for 'Montmorency' and
'Meteor' was 38 cm and 47 cm respectively. Therefore,
'Montmorency' tree vigor would be classified as moderate.

In 1984, 'Meteor' had a significantly larger number of

flower clusters, flowers, and fruits per limb and a greater

16

.Pm>oF am .ummu «acme mpnwppss m.=muc=o s5 mzon :_ covpmgmamm cwmz~

 

 

 

 

 

 

 

 

mm mmemm as name mm mesa NN news Image v.8?» new;
Hm mmm Hm nus em msm me sass QE_F\mm;u=mta m=_rmzo_t to .02
em meme mm ames om nmem km nomm nE_F\.o= pearl
we mflmHN es ammm ON nmmm Nu nmum as_1\.o= Lugosi
as was“ Hm gems oe seam km AMHN seep\mrmpmzpu razost to .02
Hm owes me no“ ow new He saw aswp\me=am acreage,» to .02
mes mmms on new mm mmms me Nmmms asep\me=n sermons s=_rmzopt to .oz

>u m >o m >o m >u m pcmcoasoo u_m_>

cams mesh «was mass
tempo: aocmg05ucoz

 

mass? sore mucmcoaeou quwx Low A>uv cowamegm> eo mucmrowmmmou new Amy mwzpm> memo:

.emms new mass cw .romumz. ace .sucwroepcoz_ to

.H mPQMF

17

Table 2. Mean values (32) and coefficients of variation (cv) for limb cross
sectional area, flower bud number per limb and, flower bud number,
fruit number, and yield per cross sectional area (cm2 ) for 'Montmorency'
and 'Meteor' in 1983 and 1984.

 

 

 

 

Montmorency Meteor
Trait 2 cv 2 cv
Limb cross-sectional area 6.7l ll 7.5l l0
Flower bud no./limb 212.25 b2 38 493.9 a l6
Flower bud no./cross-sectional area 32.94 b 14 60 a 8
Fruit no./cross-sectional area 37.56 l6 46 13
Yield (gms)/cross-sectional area l34.08 b l9 2ll a l2

 

zMean separation in rows by LSD, 5% level.

18
limb yield than 'Montmorency' (Table 1). However, in 1983
there were no differences between the cultivars for these
traits. The cultivar x year interaction was highly
significant and 'Meteor' exhibited large differences between
the 2 years. In general, 'Meteor' also had greater variation
within years than 'Montmorency' as indicated by the
coefficients of variation.

When several of the yield components were standardized by
dividing by the limb cross-sectional areas, the year effect
was largely eliminated and there were significant differences
between 'Montmorency' and 'Meteor' (Table 2). 'Meteor'
had a higher flower bud density than 'Montmorency'; however,
the difference in crop density was not significant. This lack
of difference in fruit load resulted because 'Montmorency' had
a higher fruit set (46%) than 'Meteor' (28%). However, there
was a significant difference in crop density between the
Clarksville and Hartford orchards, presumably resulting from
different weather conditions influencing fruit set. Even
though 'Montmorency' and 'Meteor' had similar crop densities,
'Meteor' had a higher yield efficiency (grams of fruit/cross-
sectional area) than 'Montmorency', because of differences in
fruit weight. Individual fruit weight for 'Meteor' was
approximately 4.7 grams compared to 3.6 grams for
'Montmorency'.

The average limb yields for 'Montmorency' and 'Meteor'

were 882 grams and 1504 grams, respectively. However, the

0003 m0 mm<
mmcp c» L» c» c»

e m m H

OH

19

md

.sxceu::>
om

 

Hound: mm
mocmuoaucoz __H

 

 

 

 

 

.mmcu can mx.mo ummmmcaxm we upmw> .uooz co man an»
an cmcowuwpcma ewmfi cw .compmz_ new .xucmgosucoz. to moor» upoucmmx m use m com u_mw> .N .mwm

2!)

Table 3. Mean values of primary yield components associated with spurs
and lateral buds for 'Montmorency' and 'Meteor' in l983 and l984.

 

 

 

 

. 'Montmorency' 'Meteor'

Y1eld component Lateral bud Spur Lateral bud Spur
Total no. of lateral

buds or spurs l44 a 29 b 89 a l20 a
No. of flower clusters l.0l c 2.42 b l.04 c 2.99 a
No. of flowers/cluster 2.63 a 2.47 a 2.54 a 2.66 a
% fruit set 4l ab 52 a 28 ab 27 b
Individual fruit

weight (gms) 3.86 b 3.38 b 4.7 a 4.6 a
Yield gms 609 b 272 b 376 b ll27 a

 

zMean separation in rows by Duncan's multiple range test, 5% level.

21
'Montmorency' trees sampled had approximately 4 times as many
4-year-old-limbs than 'Meteor' (33.3 compared to 8.5) and
therefore a considerably higher tree yield.

The average tree yields for 'Montmorency' and 'Meteor'
were 28 and 13 kg, respectively (Fig 2). The fruiting habit
of the 2 cultivars also differed significantly. 'Montmorency'
had 19 kg or 68% of its fruit on one-year-old wood while
'Meteor' had 9 kg or 70% of its fruits on 2-year-old spurs.

Lateral bud and spur reproduction was considered
separately because there was a highly significant interaction
between cultivar and reproduction location. Although the
cultivar x year interaction was significant, the age of wood
for reproductive type x year interaction was not significant
for all parameters. 'Meteor' had significantly more spurs,
flower clusters per spur, and spur yield than 'Montmorency'
(Table 3). However, 'Meteor' spurs had significantly lower
fruit set than 'Montmorency' spurs (52% and 27%, respectively).
Individual fruit weight for 'Meteor' was approximately 4.75 gms
compared to 3.62 gms for 'Montmorency'. The fruit set on
'Montmorency' spurs was higher than the 30% fruit set reported
by Diaz (9) on 'Montmorency' limbs. It is most likely that
differences between years contribute to the differences in
fruit set.

Limb yield of the 2 sour cherry cultivars was designated
as the product of fruit number and fruit weight (Fig 1).

Fruit number had a larger direct effect on yield than fruit

22

Table 4. Path coefficients showing direct and indirect effects of fruit
number and fruit weight onliufiinelds of 'Montmorency' and
'Meteor' sour cherry. Yield is divided into the yield from
lateral buds and spurs.

 

 

 

 

'Montmorency' 'Meteor'
Type Of effect Lateral buds Spurs Lateral buds Spurs
Fruit no. (x)
Direct effect (sz) l.04** l.02** l.00** 0.99**
Indirect effect via fruit wt. -0.04 -0.02 0.00 0.00
Fruit wt. (y)
Direct effect (Ply) 0.l3** 0.08** 0.03** 0.02

 

**Indicates significance at the l% level.

23
Table 5. Path coefficients showing direct and indirect effects of secondary
yield components on fruit number of 'Montmorency' and 'Meteor' sour
cherry in l983 and l984. Yield is divided into that from lateral
buds and spurs.

 

 

 

 

'Montmorency' 'Meteor'
Type of effect Lateral buds Spurs Lateral buds Spurs
No. of spurs and lateral
buds per limb (a)
Direct effect (Pxé) 0.8l** o.92** 1.00** 0.78**
Indirect effect via:
No. of flower clusters per
spur or lateral bud 0.00 -0.13 -0.0l -0.l9
No. of flowers per cluster 0.18 -0.03 -0.00 -0.02
% of fruit set 0.02 -0.ll -0.0l -0.l5
No. of flower clusters per
spur or lateral bud (b)
Direct effect (be) 0.03 0.37** 0.03 O.34**
Indirect effect via:
No. of flowers per cluster -0.03 -0.l0 0.00 0.04
% fruit set -0.05 -0.l6 -0.08 -0.04
No. of flowers per cluster (c)
Direct effect (ch) 0.2l* 0.35** 0.00 -0.l7
Indirect effect via:
% fruit set -0.05 0.0l 0.05 0.44
% fruit set (d) .
Direct effect (de) 0.2l* 0.7l** 0.l6** 0.63**

 

**, * Indicates significance at the l and 5% level, respectively.

24
weight for both cultivars (Table 4), although fruit weight
also had a very significant positive effect on 'Montmorency'
yield. The indirect effect of fruit number via fruit weight
on yield was small and insignificant.

Fruit number was expressed as the product of the
following secondary components: the number of spurs or
lateral buds, the number of flower clusters per spur or
lateral bud, the number of flowers per cluster, and % fruit
set (Fig 1). The number of spurs or lateral buds was the
most important secondary yield component influencing fruit
number in both cultivars (Table 5). Fruit set also had a
significant direct effect on fruit number which was more
important for spur fruit production. 'Montmorency' spur
fruit number was significantly associated with the number of
flower clusters per spur and flower number per cluster. For
'Meteors' accumulated spur fruit number, the number of
flowers per cluster also had a positive indirect effect via
fruit set indicating that flower competition within clusters
had little effect on fruit set. These results are similar to
those of Diaz (9) who concluded that in 'Montmorency'
competition between and within flower clusters did not affect
fruit set.

There were no significant effects of leaf number and
leaf size on that year's fruit weight in either cultivars
(Table 6). Possibly this is because leaf number was above

the threshold value which would affect fruit development.

25

Table 6. Path coefficients showing direct and indirect effects of leaf
number per limb and leaf size in l984 on fruit weight for
'Montmorency' and 'Meteor' sour cherry in 1984.

 

Type of effect 'Montmorency' 'Meteor'

 

Leaf no. (n)
Direct effect (Pyn) 0.l2 -0.35

Indirect effect via leaf size 0.09 0.05

Leaf size (5)
Direct effect (Pys) 0.44 -0.26

 

26

._m>mp am as» as mucou.e_cmpm mmumumvcwm

 

eo.o- No.8 eem e_=ee a Lucas eeeeee eeeeps
me.o mo.o cmumapu eon mgmzoFe ea .62 Aueav «bowed uuogwo
_o.o- 40.0 use _eeeeep
to seem can mcmumspu Lazope mo .62 Aneav uuweem uumcpn
o~.o- “v.01 muse ~meoamp use mgaem mo .02 Amwmv uumwmm Hume’s
«new mom;
eo.o- ec.c- me.” eee_ e_> eeeeee eeeeeeea
so.o- em.o- eem e.=ee u Isaac eeeeee ”eaten
.m.o- mo.o o~wm camp a_> uommem uuog_c:H
e~.o pn.o- gonna—u can mgozope mo .02 Aumav aomeem Homewn
~o.o mm.o . . m~wm mew. a.> uumeem “curves“
eee .eeeee_ ,
-.o1 m~.o- co team can mgmumapo Logo—m mo .oz Agony uumeeo poocpa
oo.o- m~.o e~_e temp er, eeeeee eeee.eea
~m.o ._o._ mesa _eeeeep eee mesa” ee .ez Aeeav eeeeeo eeee.a
Amy .0: eumm
.Lomumz. .AucmcoEucoz. aeoeoqsou sgevcoumm uumeeo co wasp

newcoesou seepage»

 

.sggmgu caom combos. new .zucmtosucoz. so; emap a. mucmeoeeou use?» zuuueoumm as» so mwmp cw

ase— can w~vm use guess: eam~ we muooeem uuogevcr was powers mcwzozm mpcmeuwemmou game. .m opens

27
Flore (17) reported that a minimum of 2 leaves per fruit are
necessary for optimum fruit size and development for
'Montmorency'. However, there was a significant positive
effect of 'Montmorency' leaf number in 1983 on the number of
spurs and lateral buds in 1984 (Table 7). Presumably more
lateral buds were vegetative in 1983 resulting in more spurs
the next year. 'Meteor' also had a similar positive effect
for leaf number, however it was not significant. Leaf number
and size had no other significant effects on the secondary
yield components in 1984.

Fruit number, fruit weight, the number of reproductive
buds, and fruit set appear to be the most important
components influencing limb yields. For 'Montmorency',
cultural or environmental factors which increase fruit weight
may increase yields. Alternatively for 'Meteor', increasing
the fruit set ratio may result in a yield increase. For
maximizing the yield per acre, it may be of value to consider
the spur fruiting habit because of the higher yield
efficiency. However, it must be emphasized that the yield
efficiency of 'Meteor' was on a per limb basis. Yield
evaluations of sour cherry clones with different fruiting
habits must include the number of limbs per tree since this
may be one of the most crucial factors influencing the

productivity of the orchard canopy.

28

REFERENCES

10.

11.

12.

29
References

Adams, M. W. 1967. Basis of yield component compensation
in crop plants with special reference to the field bean,
ghaseolus vulgaris. CrOp Sci. 7: 505-510.

Adams, M. W. and J. E. Grafius. 1971. Yield component
compensation-alternative interpretation. Crop Sci. 11:
33-55.

Alderman, W. H. et a1. 1952. Meteor cherry. Minn. Agric.
Expt. Sta. Misc. Rept. 16.

Anderson, R. L. 1981. Cherry cultivar situation. Fruit
Var. J. 32: 82-92.

Bhatt, G. M. 1973. Significance of path coefficient
analysis in determining the nature of character
association. Euphytica. 22: 338-343.

Borojevic, S. and W. A. Williams. 1982. Genotype x
environment interaction for leaf area parameters and
yield components and their effects on wheat yield.Crop
Sci. 22: 1020-1025.

Bradbury, D. 1929. A comparative study of the developing
and aborting fruits of Prunus gerasus. Amer. J. Bot. 16:
525-542.

Dewey, D. R. and K. H. Lu. 1959. A correlation and path-
coefficient analysis of component of crested wheatgrass
seed productiuon. Agron. J. 51: 515-518.

Diaz, D. H. 1979. Development of inflorence in
'Montmorency' sour cherry (Prunus cerasus L.) and effects
of competition between and within flower clusters on
fruit set and abscission. Ph D. Dissertation. Mich. State
Univ. East Lansing., Mi.

Duarte, R. A. and M. W. Adams. 1972. A path coefficient
analysis of some yield component interrelations in field
bean (Phaseolus vulgaris L.). Crop 801. 12: 579-582.

Eaton, G. W. and E. A. MacPherson. 1978. Morphological
components of yield in cranberry. Hort. Res. 17: 73-82.

Eaton, G. W. and T. R. Kyte. 1978. Yield component
analysis in the cranberry. J. Amer. Soc. Hort. Sci. 103:
578-583.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

30

Eisensmith, S. P., A. L. Jones, and J. A. Flore. 1980.
Predicting leaf emergence of 'MOntmorency' sour cherry
tree degree-day accumulatioms. J. Amer. Soc. Hort. Sci.
105: 75-78.

Engledow, F. L. and M. A. Wadham. 1923. Investigation on
yield of the cereal. Part I. J. Agr. Sci. 13: 390-439.

Flore, J. A. 1980. The effect of light on cherry trees.
Mich. Hort. Ann. Rept. 110.

Flore, J. A. 1981. Influence of light interception on
cherry production and orchard design. Mich. State Hort.
Ann. Rept. 111.

Flore, J. A. 1985. The effect of carbohydrate supply on
sour cherry fruit size and maturity. HortScience 20:90.

Gardner, V. R. 1933. Field studies of bud sports in
Michigan tree fruits. Mich. State Coll. Tech. Bull. 130.

Gardner, V. R. 1935. Factors influencing the yield of
Montmorency cherry cultivars in Michigan. Mich. Agr.
Expt. Sta. Spec. Bull. 275.

Grafius, J. E. 1956. Components of yield in oats: a
geometrical interpretation. Agron. J. 48: 419-423.

Grafius, J. E. and R. L. Thomas. 1971. The case for
indirect genetics control of sequential traits and
strategy of development of environmental resources by the
plant. Heredity. 26: 433-442.

Gray, G. F. 1943. Relation of light intensity to fruit
setting in the sour cherry. Mich. Agr. Expt. Sta. Bull.
136.

Hancock, J. F., J. H. Siefker, and N. L. Schulte. 1983.
Cultivar variation in yield component of strawberries.
HortSCi. 18: 312-313.

Hedrick, U. F. 1915. The cherry of New York. 22nd Ann.
Rept. State of N. Y. Dept. Agr. part II.

Iezzoni, A. F. 1984. Sour cherry breeding in Eastern
Europe. Fruit Var. J. 38: 121-125.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

31

Kang, M. S., J. D. Miller, and P. Y. D. Tai. 1983.
Genetic and phenotypic path analysis and heritability in
sugarcane. Crop Sci. 23: 643-647.

Kenworthy, A. C. 1974. Sour cherry tree vigour as related
to higher yields and better fruit quality. Farm Sci. 223:
1-4.

Kesner, C. D., C. H. Hanson, and S. B. Fouch. 1978.
Mechanical summer tipping as a method of controlling sour
cherry tree size: a preliminary report. Fruit Var. J. 32:
26-28.

Langord, L. R. 1935. Seasonal influence upon the effect
of shading in regard to setting of sour cherry fruits.
Proc. Amer. Soc. Hort. Sci. 23: 234-236.

Leng, E. R. 1962. Component analysis in inheritance
studies of grain yield in maize. Crop Sci. 3: 186-190.

Li. C. C. 1956. The concept of path coefficients and its
impact on population genetics. Biometrics. 12: 190-210.

Nickell, C. D. and J. E. Grafius. 1967. Analysis of
negative response to selection for high yield in winter
barley, flordeum vulgare L. Crop Sci. 9: 447-451.

Pandley, J. P. and J. H. Torrie. 1973. Path coefficient
analysis of seed yield components in soybean (kaflne
g@g(C) Crop Sci. 13: 505-507.

Pritts, M. P. and J. F. Hancock. 1985. Lifetime biomass
partitioning and yield component relationships in the
highbush blueberry, Vaccinium corymbosumL. (Eriaceae) .
Amer. J. Bot. 72: 446-452.

Redalen, G. 1984. Fertility in sour cherry.
Gartenbauwissenschaft. 49: 212-217.

Roberts, R. H. 1922. Better cherry yields. Wisc. Agr.
Expt. Sta. Bull. 344.

Roberts, R. H. 1930. Souw cherry fruiting . Wisc. Agr.
Expt. Sta. Bull. 415.

38.

39.

40.

41.

42.

43.

32

Roberts, R. H. and L. Langord. 1932. Faulty nutrition
responsible for poor set of cherries during cloudy
weather. Ann. Rept. Wisc. Agr. Expt. Sta. Bull. 421.

Siefker, J. H. 1985. Stability and components of yield in
highbush blueberry cultivars. Ph. D. Dissertation. Mich.
State Univ. East Lansing, MI.

Shoemaker, J. S. 1928. Cherry pollination studies. Ohio
Agr. Expt. Sta. Bull. 422.

Thomas, R. L., J. E. Grafius, and S. K. Halm. 1971b.
Transformation of sequential quantitative characters.
Heredity. 26: 189-193.

Thurling, N. 1974. Morphophysiological determinants of
yiel in rapeseed (Brassica campestris and Qrassica

nagus), II. yield componernts. Aust. J. Agr. Res.

25: 711-721.

Wright, S. 1921. Correlation and causation. J. Agron.
Res. 20: 557-585.

Chapter two
Yield Prediction for 'Montmorency' and 'Meteor' Sour Cherry

Using Stepwise Regression

34

Introduction and literature review

Reliable yield prediction is critical to the marketing
of agricultural products. The most common method of
predicting yield is correlation. Waring (14) reported that
the correlation coefficient was 0.30 to 0.75 between apple
tree yield and trunk circumference, however, the coefficient
of variability for yield and trunk circumference was
relatively large. Sudds and Anthony (12) confirmed Waring's
reports but found the trunk growth of one year was more highly
correlated with the following year's load than the load in the
same year. Reed (11) investigating fruit load in apricot
showed that negative correlation values existed between yield
and the current growth whereas positive values existed between
yield and the preceding years' wood growth. Cummings and
Jenkins (2) indicated a close correlation between wood growth
and sweet cherry yields. Hofmann (7) reported a significant
association between terminal growth in the preceding year and
apple yields. Wilcox (16) used a growth index on apples
obtained by multiplying the mean terminal growth and trunk
circumference increase. He found a positive correlation
between terminal growth and increased trunk circumference, but
negative relationships between growth index and percentage
bloom and between both trunk circumference and percentage
bloom. Overholser et al (8,9) confirmed the negative

relationship between growth index and fruit load in apple, but

35
found little association between terminal growth and trunk
circumference. However, simple correlation values r = 0.35 to
0.75 between trunk circumference and fruit load were
relatively low. Negative or positive associations might exist
between growth and yield depending on environmental
circumstances, and the cultivars investigated.

In an attempt to reduce variability in 'Elberta' peach
performance, Proebsting (10) provided constant growing
conditions and found a linear relationship between trunk
cross-sectional area and crop load in 8— to 10-year-old peach
trees. He calculated that each kilogram of peach fruit
produced exhausted a 0.128 cm trunk cross-sectional area
increment. However, Proebsting did not show the correlation
between growth and yield. Webster and Brown (15) found no
association between trunk cross-sectional area increase and
mean yields in 'McIntosh' apple trees ranging from 8- to 17-
years-old. However, they found a linear relationship between
crop load and the change in trunk circumference using the
concept developed by Proebsting for estimating 'Elberta' peach
yields.

The existing correlation between limb growth and bearing
potential has been used as an aid in the evaluation of pome
fruit thinning (5,6,13). Forshey (4) tried to predict the
'McIntosh' apple crop and found a higher correlation between
fruit number and yield than between fruit size and yield.

However, he did not present the yield prediction equation

36
which used the following data: bloom count, fruit size, and
fruit weight, nor did he consider leaf fruit ratio for apple
yield prediction.

Several methods have been developed for cherry yield
prediction. Chaplin and Westwood (1) proposed the equation Y=
-240.7 + 2039.6X where X was the actual yield and Y was the
yield index equal to[(2:WiFi/Li)/n]*Ti. To calculate their
yield index requires the following data: (a) fruit weight from
2 liters of harvested fruit (Wi), (b) fruit numbers from 10
limbs per tree (Fi), (c) limb diameter for conversion to
cross-sectional area (Li), (d) trunk cross-sectional area
(Ti), and (e) limb number (n). They obtained a coefficient of
determination equal to 0.846 in the prediction model. Several
problems occurred in this formula: (a) tree yield was treated
as an independent variable in the formula: however, tree yield
was a dependent variable which responded to the growth index,
(b) the equation could only be used in the harvesting season:
it would be advantageous to be able to predict yield in the
early season, (c) sampling procedures were cubersome requiring
data collected from 10 sampled limbs of each tree, (d) yield
was also reduced by the amount collected to provide data for
the yield index, and (e) the coefficient of determination
(0.846) was good but no validation of the equation was
presented in the paper.

For the sour cherry yield survey in Michigan, one method

has been developed based on the computer simulation (personal

37
communication with Don Fedewa, Michigan Agricultural Reporting
Service). The estimated fruit per tree = fruit count * 1/Pi *
1/Pi+1 * ---- * 1/Pi+n, where Pi is the ratio between prime
limb cross sectional area and the summation of all limb cross
sectional area at the specific notch. For collecting data, 2
terminal limbs and one prime limb of each tree are selected
and the fruit is counted twice, once in mid-June and than
again 3 days before harvest. This equation has been used in
the sour cherry market survey for 15 years and the differences
between the survey predictions and actual yields in Michigan
range from 5 to 15%. However, since yield prediction requires
a ripe-fruit count, the yield data needed for marketing
purposes is only available shortly before harvest. However,
no data validating the equation or showing a correlation
between the equation and the yield has been presented based on
individual trees.

Several regression equations for prediction have been
demonstrated and widely used (3). One is refered to as the
'all possible' regression (3). However, in 'all possible'
regression the computation is cubersome involving many factors
and the choice between equations with large numbers of factors
is subjective. The other method is refered to as 'backward
elimination' regression. However, the problems with backward
elimination are: (a) Once the variable has been deleted, it is
never considered again. (b) The matrix singularity would

cause a computation problem. (c) There is no guarantee that

38

the model achieved is a reliable model or the best fitted.
With stepwise regression, these problems do not exist. In
stepwise regression, if a predictor has been eliminated once,
it still has the possibility of being considered again.
However in backward elimination, once a predictor has been
eliminated it can not be considered in the model again.
Stepwise regression, including addition and deletion
procedures, may present an alternative model using different
predictive characters. However, the characters used in the
predictive equation are objective, since selection is
automatically performed in the computer by the computer
program eliminating the involvement of the researchers.

The objective of this paper is to develop and validate a
yield prediction formula for yield of 'Montmorency' and

'Meteor' sour cherry using stepwise regression.

39

Materials and Methods

In 1984, data was taken from (I) 3 trees of 23-year-old
and 2 trees of 16-year-old 'Montmorency' trees at the Botany
Farm, East Lansing, Mich., and (II) 3 trees each of
'Montmorency' and 'Meteor' at 2 locations, Clarksville
Horticultural Experimental Station, Clarksville, Mich. and
Hilltop Orchards and Nurseries, Hartford, Mich..

'Montmorency' tree yield at Clarksville were also collected in
1983. The trees at Clarksville and Hilltop were originally 7
and 8-years-old, respectively. The two cultivars were grafted

on g. mahaleb seedling rootstocks and trained to a modified

 

central leader.

One representative 3 or 4-year-old limb was randomly
selected from each tree at Clarksville and Hartford. Three 4-
year-old limbs were randomly selected from each tree at East
Lansing. The following totals were counted: (I) total tree
yield, (II) number of lateral buds per limb, (III) number of
spurs per limb, (IV) flower number per limb, (V) fruit set
(total fruit count/total flower count), (VI) average fruit
weight of 50 fruits, (VII) flowering branch number, (VIII)
limb circumference (cm) (IX) trunk circumference (cm), (X)
crown area (m2) calculated from wr2 , (XI) leaf number along
the terminal growth of the limb, (XII) leaf area measured with
a L1-3000 leaf area meter of 5 randomly selected leaves (cmz),

(XIII) average flower cluster number (total cluster

number)/(spur number + lateral bud number).

40

Predictors wre added to the multiple regression equation
in the order in which they were weighted to the partial F test
( STAT 4 program by Dr. Charles Cress, M. S. U., East
Lansing, Mich.). The procedures for stepwise regression are
as follow:

(I) The variable most highly correlated with yield was
selected by partial F criterion rather than its fit to the
regression model 9 = f(Xi).

(II) The partial F values for variables not in the model
were computed and the variable with the highest F values was
ntered if it was significant.

(III) The partial F values of the variables were checked
for possible deletion at each step for making decisions to
either reject or retain the corresponding prediction.

(IV) Procedures were stoped when no variable could be
added or deleted that significantly improved the prediction

equation.

41

Results and Discussion

'Meteor' tree yield was about half of 'Montmorency' tree
yield (Table 1). 'Meteor' leaf number along the terminal
growth of the limb was about 3 times that of 'Montmorency',
which may reflect an increased number of spur leaves retained
on the main branches in 'Meteor'. The coefficients of
variation for most characters from both sour cherry cultivars
were relative high. A detailed discussion of the flowering
and fruiting habits of 'Montmorency' and 'Meteor' is presented
in Chapter I. The variables in Table 1 were used to develop
the prediction equation for the healthy and mature
'Montmorency' and 'Meteor' sour cherry cultivars.

The equations developed by stepwise regression for
'Montmorency' and 'Meteor' tree yields were: 'Montmorency'
tree yield,

Y = 104.6237 - 12.449 X1 - 0.381X2 + 0.015X3 - 8.268X4

where X1: average flower cluster number per spur and

lateral bud, X2: leaf number per prime terminal shoot of the
limb, X3: flower number per limb, X4: limb circumference (cm);
'Meteor' tree yield: Y = -9.3119 + 0.3102X where X: leaf
number per prime terminal shoot of the limb.

To develop a useful and reliable model, 2 criteria have to
be achieved: (I) Involve as many as predictors as possible so

that reliable fitted values can be determined. (II) Include

Table 1. Mean values (i) and coefficients of variation (c.v.) for

42

variables sampled from 'Montmorency' and 'Meteor'.

 

 

 

 

'Montmorency' 'Meteor'

Variable x c.v. x c.v.
Yield (kg) 28. 26.0 13.1 67.
No. of lateral buds/limb 175. 41.4 152.2 142.
No. of spurs/limb 37. 53.8 168.2 53.
No. of flowers/limb 784. 40.2 2138.2 70.
Fruit set (%) 38. 33.5 22.7 27.
Fruit weight (gms) 3. 15.7 4.5 8.
No. of branches/limb 43. 32.1 53.8 52.
Limb circumference (cm) 7. 12.5 3.3 27.
Trunk circumference (cm) 45. 39.4 28.3 18.
Projected crown area (m2) 11. 19.1 9.0 15.
Leaf no. 26. 32.0 75.6 40.
Leaf area (cm2) 48. 19.1 54.4 11.
Average flower cluster no. 1. 21.4 2.8 17.

 

43

Table 2. Validation of the prediction equation between harvested yield and calculated
yield for 14 'Montmorency' trees ranging from 7 to 23 years old.

 

Harvested Yield Variables in the prediction equation

 

 

from individual Average Limb Calculated Yield
trees (y) flower Leaf Flower circumference yield difference X2
(kg) cluster no. no. no. (cm) (y) (y - y) values
29.58 1.662 31.67 566 6.08 30.08 -0.50 0.00845
23.65 1.303 28.33 420 7.19 24.46 -0.81 0.02774
27.29 1.636 36.89 533 6.18 27.10 0.19 0.00132
37.64 1.193 25.00 706 5.87 42.30 -4.66 0.57692
30.55 1.121 29.22 464 6.88 29.61 0.94 0.02892
20.30 1.107 24.75 449 8.04 21.67 -1.37 0.09245
20.48 1.295 25.89 1004 8.78 21.11 -0.63 0.01937
19.68 1.838 19.44 1123 8.73 19.00 0.68 0.02349
19.80 1.867 18.44 1186 8.30 23.52 -3.72 0.69890
36.94 1.647 10.14 1007 7.26 35.33 1.61 0.07017
42.59 1.061 15.55 1223 7.94 38.18 4.41 0.45663
28.37 1.754 36.70 438 6.13 24.69 3.68 0.47734
31.66 1.265 33.70 709 6.84 29.73 1.93 0.11765
35.72 1.040 39.40 1.61 6.96 36.53 -0.81 0.01836

2.61778

 

44
as few predictors as possible in the resulting equation to
simplify data collection. The model for 'Montmorency' sour
cherry tree yield appears to be meet both requirements.

Coefficient of determination (R?=0.892) and F values (18.6)
(degree of freedom = 12, and P = 0.001) for the prediction
equation for 'Montmorency' tree yield indicate that the
stepwise regression equation is highly reliable. The
chi-square values between predicted yield for 'Montmorency'
tree and actual yields equaled 2.62 resulting in a validation
wehich was greater than 99.5% (Table 2).

Coefficient of determination (R?=0.951) and F (59.02)
values for the prediction model for 'Meteor' tree yields were
high, however, the chi-square values for validation between
tree yield and predictive yield was 4.88 which was about 30%
of validation. This low value may be due to the reduced
number of trees sampled for 'Meteor', or more complex
interrelationships in 'Meteor'. Since the number of
predictors was larger than the number of 'Meteor' trees
harvested, there was a matrix singularity problem existing
in the deletion procedures which terminated the computer

performed calculations in the early stage.

45

Table 3. Variation of the prediction equation between harvested yield and
calculated yield for five 'Meteor' trees either seven or eight

 

 

 

years old.

Harvested

yield from

individual Variable in the Calculated Yield

trees (Y) prediction equation yield difference X2
(kg) Leaf no. (9) (y - y) values
11.34 70.3 12.49 -1.15 0.11662
12.02 58.1 8.71 3.31 0.91148
22.45 99.5 21.54 0.91 0.03688
14.57 100.5 21.86 ~7.29 3.64750
5.11 49.5 6.04 -0.93 0.16925

4.88175

 

46

REFERENCES

10.

11.

47

References

Chaplin, M. H. and M. N. Westwood. 1972. A method for
estimating the yields of sweet cherry. HortSci. 7: 508-
509.

Cummings, M. B. et al. 1933. Rootstock effects with
cherries. Vt. Agr. Expt. Sta. Bull. 351: 1-33.

Draper, N. R. and H. Smith. 1981. Applied Regression
Analysis, 2nd edition, John Wiley & Sons. New York.
P. 294-352.

Forshey, C. G. 1975. McIntosh apple crop prediction
studies. Proc. N. Y. State Hort. Sci. Bull. 120: 139-
144.

Forshey, C. G. and D. C. Elfving. 1977. Fruit number,
fruit size, and yield relationships in 'McIntosh apple'.
J. Amer. Soc. Hort. Sci. 102: 399-402.

Forshey, C. G. and D. C. Elfving. 1979. Branch samples
for yield and fruit size comparisons in apple. HortSci.
14: 143-144.

Hofmann, F. W. 1933. Yield relationships on terminal
growth in York Imperial apple. Proc. Amer. Soc. Hort.
Sci. 30: 299-302.

Overholser, E. L, F. L. Overley, and L. M. Barnhill.
1938. Correlations of trunk circumference increase and
length of terminal growth with yield of apple. Proc.
Amer. Soc. Hort. Sci. 35: 263-268.

Overholser, E. L., F. L. Overley, L. M. Barnhill, and J.
C. Wilcox. 1941. Some correlations between growth and
yield of the apple in central Washington. Proc. Amer.
Soc. Hort. Sci. 39: 11-15.

Proesbting, E. L. Jr.. 1958. A quantitative evaluation of
the effect of fruiting on growth of Elberta peach trees.
Proc. Amer. Soc. Hort. Sci. 71: 103-109.

Reed, H. S.. 1928. Correlations between growth and fruit
production of apricot. Proc. Amer. Soc. Hort. Sci. 25:
247-249.

12.

13.

14.

15.

16.

48

Sudds, R. H. and R. D. Anthony. 1928. The correlation of
trunk measurement with tree performance in apple. Proc.
Amer. Soc. Hort. Sci. 25: 244-246.

Uriu, K. and O. Lilleland. 1959. Limb circumference
measurements as an aid in the better evaluation of apple
fruit thinning data. Proc. Amer. Soc. Hort. Sci. 74: 67-
71.

Waring, J. H.. 1920. The probable value for trunk circumference
as an adjacent to fruit yield in interpreting apple

orchard experiments. Proc. Amer. Soc. Hort. Sci. 17: 179-

185.

Wester, D. H. and G. L. Brown. 1980. Trunk growth of
apple trees as affected by crop load. J. P1. Sci. 60:
1383-1391.

Wilcox, J. C.. 1937. Field studies of apple tree growth
and fruiting. Sci. Agr. 17: 575-586.

HI CHIGRN STRTE UNIV.

IllIlnulliiitlliiilmmlnluullulnulmnlmlnlmll