Document [original]

88 E u r o p e a n J o u r n a l o f H o r t i c u l t u r a l S c i e n c e

Eur. J. Hortic. Sci. 86(1), 88–98 | ISSN 1611-4426 print, 1611-4434 online | https://doi.org/10.17660/eJHS.2021/86.1.10 | © ISHS 2021

Ǧϐ

Ǧϐ

M. Penzel1,2ǡǤϐ2, R. Gebbers2 and M. Zude-Sasse2

1 Technische Universität Berlin, Chair of Agromechatronics, Berlin, Germany

2 Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB), Potsdam, Germany

Original article

Summary

    

yield and fruit quality in the production of apple fruit.

    

ϐ

   Ǧ  ϐ

thinning. In the years 2011, 2014–2016, commercial

orchards of ‘Elstar’/M26 and ‘Gala’/M9 trained as

ǡϐ

set were mechanically thinned at balloon stage (BBCH

59) with the Darwin 250 device at constant vehicle

speed of 8 km h-1 with varying rotational frequency

ranging from 200 revolutions per minute (rpm) to

380 rpm. Rotational frequency of the thinning device

was translated to average kinetic energy (Ekin [J])

      Ǥ 

ͲǤͳͷͲǤ͸͸Ǥ

ϐǡ

production target of 119 fruit and, therefore, no

  Ǥ  εͲǤʹ͵  

yield loss by over-thinning without any positive effect

Ǥϐǡ

of 0.23 J and 0.33 J were adequate settings to reduce

crop load in ‘Elstar’ and 0.33 J in ‘Gala’ without yield

Ǥϐ

ϐ

the avoidance of yield loss by over-thinning of trees

ϐͳǤͶȂͶǤʹ-1 in

‘Elstar’ and 2.6–7.6 t ha-1 in ‘Gala’. Results indicate the

   ϐ   

ϐ

to the tree’s yield capacity.

Keywords

crop load, ‘Elstar’, fruit quality, fruit tree, ‘Gala’, precision

horticulture, yield

ϐ

What is already known on this subject?

• In former studies, the principle of mechanical thinning

with varying parameters was intensively investigated

in apple. The spatial variability of soil was shown for

ϐ

known in practise.

ϔǫ

• With the proposed method, the actual impact of

mechanical thinning becomes comparable by utilizing



treatment. It ϐ

ϐǡ

while tree-adapted thinning can optimize the process

considering fruit quality and yield.

What is the expected impact on horticulture?

• ϐ

managed more precisely resulting in enhanced yield.

Introduction

ϐ-

ceeding the fruit bearing capacity of the tree (Lakso, 2011).

This physiological characteristic is desirable regarding

yield reliability even in the case of negative environmental

impacts, e.g., late frost, hail and rain, which can cause un-

     ϐ 

fruit set (Heinicke, 1917; MacDaniels and Heinicke, 1929).

Consequently trees frequently bear more fruit than can be

supplied appropriately with carbohydrates (Lakso, 2011),

leading to minor fruit size and alternate bearing (Jonkers,

German Society for

Horticultural Science

1979; Looney, 1993). Endogenous regulation mechanisms of

the tree initiate the reduction of crop load by means of fruit

abscission. The fruit drop captures unfertilized and poorly

developed fruit, which are dominated by fruit with adequate

supply of exogenous and endogenous growth factors, visu-

ally measureable by common number of seeds (Luckwill,

1953; Bangerth, 2000). The endogenous regulation of fruit

 ǡ ǡ  ϐ     

the optimum level considering fruit quality and prevention

of alternate bearing. Therefore, management of surplus fruit

is required. The optimum crop load of a tree highly depends

on the target of fruit size and the availability of resources,

especially carbohydrates and water, per fruit. The supply

of carbohydrates is limited by the leaf area to fruit ratio of

the tree (Haller and Magness, 1933; Preston, 1954; Silberei-

sen, 1966; Hansen, 1969, 1977). Crop load management can

ǡǡ

increase fruit growth rate and advance maturity of apple

ȋǤǡͳͻͻ͹ǢòǤǡʹͲͲͲȌǤϐ

of the 20th century, hand thinning was the only production

measure of crop load management (Dennis, 2000). Mean-

while, chemical thinning has been widely accepted in prac-

ϐϐǤ-

ǡϐ

thinning intensity is frequently hardly predictable (Stover

and Greene, 2005). As an alternative, devices for mechani-

ϐ͵Ͳ years

with more reliable effects (Schröder, 1996; Bertschinger et

V o l u m e 8 6 | I s s u e 1 | F e b r u a r y 2 0 2 1

ǤȁǦϐ

al., 1998; Damerow et al., 2007). The principle is based on a

ǡϐ

ϐǤ

driven by a mechanical or hydraulic system, which allows the

-

ȋȌǤϐ

length of strings, the rotational frequency and the speed of

Ǥϐ-

er removal (Zoth, 2011; Kon et al., 2013; Penzel et al., 2020)

by the enhanced kinetic energy (Ekin), which is transferred

by the strings into the canopy. The speed of the tractor in-

ϐǤ

  ϐ   

and Blanke (2010) regards most of these factors and also in-

ǲǳ

ϐǤǡǯ

ϐ

ǡϐ

compared to the effect of the thinning intensity. The formula

developed by Zoth (2011) totalizes the kinetic energy, which

every centimeter of one string transfers into the canopy.

Apart from the intended treatment, the strings also hit

ϐȋ

et al., 2013). Therefore, it is assumed that mechanical thin-

ning has an effect on ethylene production and export from

the tissue (Kong et al., 2009), which might trigger fruit ab-

 ȋǡ ʹͲͲͲȌǤ  ǡ ϐ  -

ing a mechanical thinning device potentially improves fruit

quality and supports return bloom the same way as chemical

or mechanical hand thinning (Weibel et al., 2008; Hehnen,

ʹͲͳʹȌǤϐ-

duction of labor input (Schupp et al., 2008). Consequently,

ϐ

friendly way of crop load management (Solomakhin and

Blanke, 2010), which can also be applied in organic fruit

production (Weibel et al., 2008; Sinatsch, 2014) and at any

weather condition. In practice, mechanical thinning is per-

ϐǦ

ǡ    ǯ  ϐ Ǥ

This can potentially lead to yield losses by over-thinning of

ϐǡ

size and taste, as result of under-thinning trees with heavy

ϐǤ

by Zude-Sasse et al. (2016) would analyse the spatial het-

ϐǡ

thinning intensity according to these information. However,

there is still a lack of data on machine thinning and quan-

ϐ     

ϐǤ

ϐǦ-

form thinning intensity on (i) the yield considering trees

ǡǡϐǡȋȌ the fruit qual-

ity and percentage of marketable yield obtained from trees

ǡǡϐǤ

Materials and methods

Experimental sites

The study was carried out in two commercial orchards

of Malus × domestica Borkh. in the fruit production region

of Werder, Brandenburg, Germany. In 2002, trees of the cul-

tivars ‘Elstar’ and ‘Gala’, trained as slender spindle, were

planted with a spacing of 3.2 m × 1.2 m, which accounts for

2,604 trees ha-1. The rootstock was M26 for ‘Elstar’ and M9

for ‘Gala’. The soil type was a sandy loam. Trees were irri-

gated by drip irrigation. Management of the trees was per-

formed according to the national regulations of integrated

production.

For ‘Elstar’, 200 trees were selected for the years 2011,

2014, 2015, while in ‘Gala’ 100 trees were selected in 2014

and 200 trees in 2015 and 2016. The position of every tree

was recorded, using a real time kinematic global positioning

systems (RTK-GPS) (HiPer Pro, Topcon Corporation, Japan)

to investigate the spatial effect on yield parameters.

Trees were thinned mechanically when 50–80% of the

ϐ    ȋʹͲͳͳǣ 27.04; 2014: 22.04;

2015: 30.04; 2016: 06.05.), BBCH 59 (Meier, 1997). Mechan-

ical thinning was carried out using a rotating string thinner

(Darwin 250, FruitTec, Germany) equipped with 270 plastic

strings each with a length of 60 cm. Diameter of the internal

tube of the spindle was 50 mm. The tractor speed was con-

stant at 8 km h-1 during the treatment, while varying rota-

tional frequency ranging between 200 rpm and 380 rpm was

applied. Thinning at 210+10 rpm served as basic rotational

frequency (brf) and was considered as control, since no or

minimum thinning effect was found. No additional hand thin-

ning was conducted.

ϐ,ϐ, yield and

average fruit mass

ϐͳʹ whole trees

of ‘Elstar’ in 2011 and 110 branches per cultivar in 2014 to

ϐǡ-

lated as arithmetic mean of counts. The factors found in 2014

were applied in 2015 and 2016. Prior to each thinning treat-

ment the number of clusters was counted manually for each

tree. All fruits were harvested during commercial harvest of

this orchard (2011: 17.08; 2014: 05. and 16.09; 2015: 10.-

11.09; 2016: 07.-09.09). Crop load, yield, fruit mass, and per-

centage of marketable yield considering fruit size >65 mm

was measured tree-individual with a commercial sorting line

ȋǡǤǤǡȌǤϐ-

ϐȋ

ȏΨȐȌͳͲͲϐǤ



In every year, at harvest fruit samples from each treat-

ment (2011: 787; 2014: 63; 2015: 2118; 2016: 165 fruits)

were analyzed regarding fruit quality. Soluble solids content

ȋȏΨȐȌ

refractometer (DR-301-95, Krüss, Germany). Starch break-

down was analyzed by staining the fruit halves with Lugol’s

iodine solution and interpreting the starch hydrolysis on a

scale between 1 ͳͲǤϐϐȋ cm-2) was

measured with motor-driven, digital penetrometer (Texture-

Analyser, TA.XT, Stable Micro Systems Ltd., UK). A handheld

spectrophotometer (Pigment Analyzer PA1101, CP, Germa-

ny) was used for non-destructive measurement of chloro-

phyll-related normalized difference vegetation index (NDVI

[-1; 1]; Zude, 2003).

ϐ

ϐ  

kinetic energy (Ekin [J]) that one string transfers into the can-

opy. The Ekin calculated for the middle of the string is affect-

ed by the mass, m [kg], and the speed, w [m s-1] of the string

(Eq. 1).





ȋȏȐȌ

Ǥ

ǡȏȐǡǡȏȉǦͳȐȋǤͳȌǤ

ܧ ൌଵ

ଶή݉ήݓ; ȋǤͳȌ



ǡǡǡǡǡ

ȋǤʹǢͳȌǤ

Ǥ

;ൌ;൅; ȋǤʹȌ



ȋǤ͵ȌǤǡȏȉǦͳȐǡ

Ǥ

ൌ൅ ȋǤ͵Ȍ

ǡ

Ͳǡ

ȋǤͶȌǤ

ൌ൅ ȋǤͶȌ



ȏȉǦͳȐǡǡȽȏȐǡ

ǤǡͻͲι

ȽǤȋǤͷ͸ȌǤ

ൌήሺȽ൅ͻͲሻ ȋǤͷȌ

ൌήሺȽ൅ͻͲሻ ȋǤ͸Ȍ

ȏȉǦͳȐǡȏȐǡ

ȏȉǦͳȐȋǤ͹ȌǤ

ǡȏȐǡǡȏȐǡ

ǡǦǡȏȐǡ

ȋǤͺȌǤȏȉǦͳȐ

ȋȌǦͳǤͻǤ

ݑ ൌή ȋǤ͹Ȍ

ݎ ൌௗ

ଶ൅௟

ଶ ȋǤͺȌ

݊ൌ ௦

଺଴ήʹήɎ ȋǤͻȌ

ǤͳͲͳǤ

୩୧୬ൌͳ

ʹήሺ൬൅ሺ݀

ʹ൅݈

ʹሻή ݏ

͸ͲήʹήɎήሺȽ൅ͻͲሻ൰ʹ

൅ቀௗ

ଶ൅௟

ଶሻή ௦

଺଴ήʹήɎήሺȽ൅ͻͲሻቁʹሻ ȋǤͳͲȌ



Ǥ͵ǤͶǤͳȋǡʹͲͳͺȌ

ǮǯȋǤǡʹͲͳͺȌǡǮǯȋǡʹͲͲͶȌǡǮǯȋǡ

ʹͲͲͷȌǡǮǯȋǤǡʹͲͲͺȌǤ

ȋȌ

ȋǡǡǡ

ȌȋʹͲͳ͵ȌǤ

Ǥ

ǡǯȋǡʹͲͳ͹Ȍ

ͷΨǤ





(Eq. 1)

90 E u r o p e a n J o u r n a l o f H o r t i c u l t u r a l S c i e n c e

The magnitude of the velocity vector w of the string

consists from the translational movement, wx, from the

tractor in direction x and the rotary movement, wy, in

direction y, from the spindle (Eq. 2; Figure 1). It is assumed

that the rotation is conducted in parallel to the ground and

no movement in direction z occurs.

w² = wx² + wy² (Eq. 2)

The speed wx results from the addition of the vector

components vx and ux in the driving direction of the tractor

x (Eq. 3). The vector vx equates the speed, v [m s-1], of the

tractor because the motion is rectilinear in one direction.

wx = vx + ux (Eq. 3)

The speed wy results from the addition of the vector

components vy in direction y, which equates 0 because the

movement of the tractor occurs only in direction x, and uy in

direction y upright to the direction x (Eq. 4).

wy = vy + uy (Eq. 4)

The vector components ux and uy follow from multiplying

the circumferential speed u [m s-1] of the string and the cosine,

ǡȽȏȐǡ

wx and wy. Because the string hits the tree vertically to the

ǡͻͲιȽǤȋǤ 5 and 6).





ǡȏȐǡǡȏȉǦͳȐȋǤͳȌǤ

ܧ ൌଵ

ଶή݉ήݓ; ȋǤͳȌ

Ǥ

;ൌ;൅; ȋǤʹȌ

Ǥ

ൌ൅ ȋǤ͵Ȍ

ȋǤͶȌǤ

ൌ൅ ȋǤͶȌ

ȽǤȋǤͷ͸ȌǤ

ൌήሺȽ൅ͻͲሻ ȋǤͷȌ

ൌήሺȽ൅ͻͲሻ ȋǤ͸Ȍ

ȋȌǦͳǤͻǤ

ݑ ൌή ȋǤ͹Ȍ

ݎ ൌௗ

ଶ൅௟

ଶ ȋǤͺȌ

݊ൌ ௦

଺଴ήʹήɎ ȋǤͻȌ

୩୧୬ൌͳ

ʹήሺ൬൅ሺ݀

ʹ൅݈

ʹሻή ݏ

͸ͲήʹήɎήሺȽ൅ͻͲሻ൰ʹ

൅ቀௗ

ଶ൅௟

ଶሻή ௦

଺଴ήʹήɎήሺȽ൅ͻͲሻቁʹሻ ȋǤͳͲȌ



ͷΨǤ





(Eq. 5)





ǡȏȐǡǡȏȉǦͳȐȋǤͳȌǤ

ܧ ൌଵ

ଶή݉ήݓ; ȋǤͳȌ

Ǥ

;ൌ;൅; ȋǤʹȌ

Ǥ

ൌ൅ ȋǤ͵Ȍ

ȋǤͶȌǤ

ൌ൅ ȋǤͶȌ

ȽǤȋǤͷ͸ȌǤ

ൌήሺȽ൅ͻͲሻ ȋǤͷȌ

ൌήሺȽ൅ͻͲሻ ȋǤ͸Ȍ

ȋȌǦͳǤͻǤ

ݑ ൌή ȋǤ͹Ȍ

ݎ ൌௗ

ଶ൅௟

ଶ ȋǤͺȌ

݊ൌ ௦

଺଴ήʹήɎ ȋǤͻȌ

୩୧୬ൌͳ

ʹήሺ൬൅ሺ݀

ʹ൅݈

ʹሻή ݏ

͸ͲήʹήɎήሺȽ൅ͻͲሻ൰ʹ

൅ቀௗ

ଶ൅௟

ଶሻή ௦

଺଴ήʹήɎήሺȽ൅ͻͲሻቁʹሻ ȋǤͳͲȌ



ͷΨǤ





(Eq. 6)

The circumferential speed u [m s-1] results from

multiplying the radius, r [m], at the middle of the string with

the rotational frequency n [rad s-1] of the rotating spindle

(Eq. 7). The radius at the middle of the string, r [m], captures

the half diameter, d [m], of the inner tube of the rotating

spindle, where the string is attached and the half-length,

l [m], of the string (Eq. 8). The rotational frequency n [rad s-1]

results from converting the revolutions per minute (rpm)

into the unit rad s-1 by using Eq. 9.





ǡȏȐǡǡȏȉǦͳȐȋǤͳȌǤ

ܧ ൌଵ

ଶή݉ήݓ; ȋǤͳȌ

Ǥ

;ൌ;൅; ȋǤʹȌ

Ǥ

ൌ൅ ȋǤ͵Ȍ

ȋǤͶȌǤ

ൌ൅ ȋǤͶȌ

ȽǤȋǤͷ͸ȌǤ

ൌήሺȽ൅ͻͲሻ ȋǤͷȌ

ൌήሺȽ൅ͻͲሻ ȋǤ͸Ȍ

ȋȌǦͳǤͻǤ

ݑ ൌή ȋǤ͹Ȍ

ݎ ൌௗ

ଶ൅௟

ଶ ȋǤͺȌ

݊ൌ ௦

଺଴ήʹήɎ ȋǤͻȌ

୩୧୬ൌͳ

ʹήሺ൬൅ሺ݀

ʹ൅݈

ʹሻή ݏ

͸ͲήʹήɎήሺȽ൅ͻͲሻ൰ʹ

൅ቀௗ

ଶ൅௟

ଶሻή ௦

଺଴ήʹήɎήሺȽ൅ͻͲሻቁʹሻ ȋǤͳͲȌ



ͷΨǤ





(Eq. 7)





ǡȏȐǡǡȏȉǦͳȐȋǤͳȌǤ

ܧ ൌଵ

ଶή݉ήݓ; ȋǤͳȌ

Ǥ

;ൌ;൅; ȋǤʹȌ

Ǥ

ൌ൅ ȋǤ͵Ȍ

ȋǤͶȌǤ

ൌ൅ ȋǤͶȌ

ȽǤȋǤͷ͸ȌǤ

ൌήሺȽ൅ͻͲሻ ȋǤͷȌ

ൌήሺȽ൅ͻͲሻ ȋǤ͸Ȍ

ȋȌǦͳǤͻǤ

ݑ ൌή ȋǤ͹Ȍ

ݎ ൌௗ

ଶ൅௟

ଶ ȋǤͺȌ

݊ൌ ௦

଺଴ήʹήɎ ȋǤͻȌ

୩୧୬ൌͳ

ʹήሺ൬൅ሺ݀

ʹ൅݈

ʹሻή ݏ

͸ͲήʹήɎήሺȽ൅ͻͲሻ൰ʹ

൅ቀௗ

ଶ൅௟

ଶሻή ௦

଺଴ήʹήɎήሺȽ൅ͻͲሻቁʹሻ ȋǤͳͲȌ



ͷΨǤ





(Eq. 8)





ǡȏȐǡǡȏȉǦͳȐȋǤͳȌǤ

ܧ ൌଵ

ଶή݉ήݓ; ȋǤͳȌ

Ǥ

;ൌ;൅; ȋǤʹȌ

Ǥ

ൌ൅ ȋǤ͵Ȍ

ȋǤͶȌǤ

ൌ൅ ȋǤͶȌ

ȽǤȋǤͷ͸ȌǤ

ൌήሺȽ൅ͻͲሻ ȋǤͷȌ

ൌήሺȽ൅ͻͲሻ ȋǤ͸Ȍ

ȋȌǦͳǤͻǤ

ݑ ൌή ȋǤ͹Ȍ

ݎ ൌௗ

ଶ൅௟

ଶ ȋǤͺȌ

݊ൌ ௦

଺଴ήʹήɎ ȋǤͻȌ

୩୧୬ൌͳ

ʹήሺ൬൅ሺ݀

ʹ൅݈

ʹሻή ݏ

͸ͲήʹήɎήሺȽ൅ͻͲሻ൰ʹ

൅ቀௗ

ଶ൅௟

ଶሻή ௦

଺଴ήʹήɎήሺȽ൅ͻͲሻቁʹሻ ȋǤͳͲȌ



ͷΨǤ





(Eq. 9)

Eq. 10 displays the resulting overall formula for Ekin

illustrated in Figure 1.





ǡȏȐǡǡȏȉǦͳȐȋǤͳȌǤ

ܧ ൌଵ

ଶή݉ήݓ; ȋǤͳȌ

Ǥ

;ൌ;൅; ȋǤʹȌ

Ǥ

ൌ൅ ȋǤ͵Ȍ

ȋǤͶȌǤ

ൌ൅ ȋǤͶȌ

ȽǤȋǤͷ͸ȌǤ

ൌήሺȽ൅ͻͲሻ ȋǤͷȌ

ൌήሺȽ൅ͻͲሻ ȋǤ͸Ȍ

ȋȌǦͳǤͻǤ

ݑ ൌή ȋǤ͹Ȍ

ݎ ൌௗ

ଶ൅௟

ଶ ȋǤͺȌ

݊ൌ ௦

଺଴ήʹήɎ ȋǤͻȌ

୩୧୬ൌͳ

ʹήሺ൬൅ሺ݀

ʹ൅݈

ʹሻή ݏ

͸ͲήʹήɎήሺȽ൅ͻͲሻ൰ʹ

൅ቀௗ

ଶ൅௟

ଶሻή ௦

଺଴ήʹήɎήሺȽ൅ͻͲሻቁʹሻ ȋǤͳͲȌ



ͷΨǤ









ǡȏȐǡǡȏȉǦͳȐȋǤͳȌǤ

ܧ ൌଵ

ଶή݉ήݓ; ȋǤͳȌ

Ǥ

;ൌ;൅; ȋǤʹȌ

Ǥ

ൌ൅ ȋǤ͵Ȍ

ȋǤͶȌǤ

ൌ൅ ȋǤͶȌ

ȽǤȋǤͷ͸ȌǤ

ൌήሺȽ൅ͻͲሻ ȋǤͷȌ

ൌήሺȽ൅ͻͲሻ ȋǤ͸Ȍ

ȋȌǦͳǤͻǤ

ݑ ൌή ȋǤ͹Ȍ

ݎ ൌௗ

ଶ൅௟

ଶ ȋǤͺȌ

݊ൌ ௦

଺଴ήʹήɎ ȋǤͻȌ

୩୧୬ൌͳ

ʹήሺ൬൅ሺ݀

ʹ൅݈

ʹሻή ݏ

͸ͲήʹήɎήሺȽ൅ͻͲሻ൰ʹ

൅ቀௗ

ଶ൅௟

ଶሻή ௦

଺଴ήʹήɎήሺȽ൅ͻͲሻቁʹሻ ȋǤͳͲȌ



ͷΨǤ





(Eq. 10)

Statistical analysis

Statistical analyses were carried out in software R v. 3.4.1

(R Core Team, 2018) using the packages ‘nlme’ (Pinheiro

et al., 2018), ‘gstat’ (Pebesma, 2004), ‘sp’ (Pebesma and

Bivand, 2005), and ‘multcomp’ (Hothorn et al., 2008). The

spatial covariance of latitude and longitude was included in

the analysis of variance (ANOVA) when analyzing the effect

of thinning treatment on the yield parameters (crop load,

ϐ  ǡ   ǡ    

marketable yield) as described by Crawley (2013). ANOVA

results are expressed as p values in the text. In the case that

ϐ

traits was found, Dunn’s Test (Dinno, 2017) was used for

ͷΨϐǤ







ͳǤȋα

ǡǢαǢȽα

ȌǤ

 

ĎČĚėĊͳǤ Scheme of the operating mode of the mechanical thinning device and velocity vectors (d = diameter of the inner

tube of the rotating spindle, where the string is attached; l αǢȽ = the angle between wx and wy).

V o l u m e 8 6 | I s s u e 1 | F e b r u a r y 2 0 2 1

Results

ϐ

ϐͶǤ͵άͲǤ͸

for ‘Elstar’ in 2011 and 4.7±0.3 for ‘Elstar’ and 5.2±0.3 for

‘Gala’ in 2014. The absolute number and the distribution of

ϐǡ

investigated trees, varied considerably between cultivars

ȋʹȌǤǮǯϐ

arithmetic mean of 249, 246, 646 in 2011, 2014, 2015, re-

ǡǮǯϐͷͶͳǡ

532, 1,434 in 2014, 2015, 2016, respectively. In ‘Elstar’,

ϐǡ

ȋͳȌǤϐǮǯ

was 2,162 observed on one single tree in 2015, however the

95thϐͳǡʹͲ͸ǡ

ǮǯϐǤǮǯǡ

ϐ͵ǡʹͶͷʹͲͳ͸ǡ

extreme value regarding 95th quantile of 2,149 in the same

Ǥϐ

as residues to avoid overestimation of the effect of thinning

intensity. Trees were grouped in 3 classes according to low,

ǡϐȋͳʹȌǤ

ϐ

The angle of the string to the driving direction has an im-

pact on Ekin (Figure 3). Considering the settings of the Dar-

win device used in the present trials, the angle of the string

generated a difference in Ekin of 0.29 J at 380 rpm, 8 km h-1

due to the possible angles between 0 and 180 degrees, with

0 degree and 180 degrees resulting in maximum and 90° in

minimum Ekin for each rotation.

The speed of the tractor resulted in a reduced effect on

Ekin considering the operation speeds described earlier and,









ʹǤ

ǮǯȀʹ͸ǮǯȀͻͳͷͲǤ

Ǥ

 

ĎČĚėĊʹǤ ϐϐ

clusters in all years in ‘Elstar’/M26 and ‘Gala’/M9 providing a class width of 150. Dashed line marks the mean value.

92 E u r o p e a n J o u r n a l o f H o r t i c u l t u r a l S c i e n c e

to our knowledge, applied in practice ranging from 2.5 km h-1

to 12.0 km h-1 (Kon et al., 2013). At rotational frequency of

380 rpm, this difference in tractor speed caused a maximum

spread of 0.19 J, regarding the angle of the string at 135° to

the driving direction. However, the effects of rotational fre-



with quadratic effect on Ekin per hit calculated for the mid-

dle of the rotating string (Eq. 10). For the 60-cm strings and

8 km h-1 used in the present study, the average Ekin per hit cal-

culated at the middle of the string at an angle of 135 degree,

ranged from 0.15 J at 200 rpm to 0.66 J at 380 rpm.

ϐǡǡ

1. Spatial autocorrelation. Variograms (21 ϐǡ

not shown) were used for selecting post hoc or adapted post

ǤϐȋȌ-

ketable yield no spatial effect was found in any trial. In ‘El-

star’, an adapted post hoc test was applied for yield, crop

load, and average fruit mass where a spatial effect was found

until a distance of 30 m between the trees. For ‘Gala’ spatial

effects were found until a distance of 15 m in 2014 and until

20 m in 2015 and 2016, considering the variables yield, crop

load, and average fruit mass.

ʹǤϐϐǤThe

FFS, representing the percentage of fruit developed from

100 ϐǡǡ-

ϐ

 ȋ ͳ  ʹȌǤ ǡ   ϐ 

thinning treatment in both cultivars in every year. FFS ranged

ʹǤ͹ΨϐʹͲͳ͸

ͶͺǤͶΨϐʹͲͳͷǤ

  ϐ ǡ  Ǯǯ ʹͲͳͳ ȋ ͶȌ 

2014, thinning treatment of 0.33 J and higher reduced FFS

(2011: p<0.01; 2014: p=0.03) in comparison to the basic ro-

tational frequency (brf) of 0.15 J. In 2015, treatment above

or equal to 0.28 J had an effect (2015: p<0.001) on FFS in

comparison to brf of 0.19 J (Table 3). In ‘Gala’ 2014, no effect

of thinning treatment on FFS was observed for trees with low

ϐȋͷȌǤʹͲͳͷǡ

ϐϐ

within this class. In ‘Gala’ 2016, treatments of 0.39 J and high-

ȋδͲǤͲͳȌϐ

set from 8.2% to 3.8% and lower.

  ϐ ǡ  Ǯǯ ʹͲͳͳ  

0.33 J and 0.45 J reduced FFS (p<0.01) in comparison to brf

(Figure 4), while in 2014 only 0.45 ϐ-

ĆćđĊͳǤϐϐǡǡǮǯȀʹ͸Ǥ

Flower set Class 2011 2014 Flower set Class 2015

n x¯ n ¯x n ¯x

0 – – – 18 0 0 – 2 0

1–200 Low 99 84.5 89 57.5 1–350 Low 57 195.9

201–400 Medium 64 292.8 52 296.5 351–700 Medium 67 519.4

401–800 High 44 529.3 46 556.6 701–1,400 High 82 962.5

Residue – 2 – 8 – Residue – 9 –

ĆćđĊʹǤϐϐǡǡǮǯȀͻǤ

Flower set Class 2014 2015 Flower set Class 2016

n x¯ n x¯ n x¯

50–350 Low 22 242.5 30 269.2 650–1,150 Low 45 989.6

350–600 Medium 45 510.5 105 479.6 1,151–1,650 Medium 75 1,412.7

600–1,000 High 29 740.9 58 742.7 1,651–2,150 High 56 1,849.5

Residue – 4 – 7 – Residue – 24 –









͵Ǥ



ǡ͸Ͳ

ȋʹǤͷ

Ǧͳ

αǡͺǤͲ

Ǧͳ

αǡͳʹǤͲ

Ǧͳ

αȌͻͲιȋ

ȌͳͺͲιȋȌǤ

 

ĎČĚėĊ ͵Ǥ Calculated Ekin occurring

during mechanical thinning, consider-

ing one hit of in the middle of the

60 cm string at varying tractor speed

(2.5 km h-1 = square, 8.0 km h-1 = cir-

cle, and 12.0 km h-1 = triangle) at 90°

(open symbol) and 180° (closed sym-

bol) angle of the string to the driving

direction.

V o l u m e 8 6 | I s s u e 1 | F e b r u a r y 2 0 2 1









ͶǤȋαͲǤͳͷǡαͲǤʹ͵ǡαͲǤ͵͵ǡ

ǦαͲǤͶͷȌȋȌǡȋȌǡ

ȋȌǡ      ȋǡΨȌ     ǣ ȋȌǣ ͳȂʹͲͲǡ

ȋȌǣʹͲͳȂͶͲͲǡȋȌǣͶͲͳȂͺͲͲȋȌǮǯȀʹ͸ʹͲͳͳǤ



 

ĎČĚėĊ ͶǤ Effect of mechanical thinning

treatment (solid lines = 0.15 J, dashed lines =

0.23 J, dotted lines = 0.33 J, dash-dotted

lines = 0.45 J) on crop load (CL as number of

fruit per tree), fresh mass (FM in g), absolute

yield (Y in kg), and percentage of marketable

yield (MY, ΨȌ  ϐ

set: (A) low: 1–200, (B) medium: 201–400,

ȋȌǣͶͲͳȂͺͲͲϐȋȌǮǯȀ

M26 in 2011.

ĆćđĊ͵ǤȋȽαͷΨȌȋϐȌϐȋȌ

‘Elstar’ in two years.

Flower set

Thinning

treatment Ekin

(J)

Crop load

(fruit/tree)

Final fruit set

(fruit 10-2ÀRZHUV

Yield

(kg)

Fruit mass

(g)

Marketable yield

(%)

2014

Low 0.15 21.8 BC, a 38.9 B, b 4.0 BC, a 196.9 A, b 97.7 A, b

0.23 33.3 C, a 39.6 B, b 5.3 C, a 182.8 A, b 95.5 A, b

0.33 14.6 AB, a 26.4 A, a 2.7 AB, a 182.1 A, b 96.5 A, b

0.45 16.2 A, a 27.8 A, a 2.7 A, a 171.1 A, b 93.8 A, b

Medium 0.15 100.5 A, b 35.4 B, b 13.1 A, b 130.8 A, a 77.7 A, a

0.23 110.3 A, b 39.4 B, ab 16.0 A, b 145.8 A, a 86.2 A, a

0.33 101.4 A, b 35.7 B, b 15.5 A, b 153.4 A, b 90.3 A, b

0.45 87.4 A, b 26.8 A, a 12.4 A, b 136.0 A, a 84.6 A, ab

High 0.15 131.9 A, c 25.3 B, a 16.0 A, b 119.9 A, a 67.9 A, a

0.23 126.5 A, b 24.5 AB, a 14.7 A, b 116.6 A, a 63.3 A, a

0.33 125.2 A, b 22.8 AB, a 14.7 A, b 117.1 A, a 69.7 A, a

0.45 107.3 A, b 19.0 A, a 12.9 A, b 118.6 A, a 72.5 A, a

2015

Low 0.19 85.5 C, a 48.4 C, b 13.4 C, a 168.4 A, b 91.2 A, b

0.23 58.1 BC, a 36.1 BC, a 10.0 BC, a 176.1 A, c 95.7 A, b

0.28 40.3 AB, a 21.7 AB, a 6.9 AB, a 159.0 A, b 87.4 A, b

0.45 26.7 A, a 11.8 A, a 4.3 A, a 146.8 A, a 89.4 A, a

Medium 0.19 123.4 B, ab 23.2 B, a 17.8 B, a 144.9 A, b 85.4 A, b

0.23 113.2 B, b 22.5 B, a 16.4 B, ab 141.6 A, b 80.8 A, ab

0.28 110.8 B, b 21.5 B, a 13.6 AB, b 131.7 A, ab 76.0 A, ab

0.45 73.5 A, b 14.3 A, a 10.7 A, b 149.7 A, a 87.8 A, a

High 0.19 162.8 B,b 17.1 B, a 16.1 A, a 100.7 A, a 55.3 A, a

0.23 172.9 B,c 19.1 B, a 18.4 A, b 106.0 A, a 60.7 A, a

0.28 142.3 AB, c 15.9 AB, a 16.0 A, b 113.3 A, ab 65.9 A, a

0.45 108.5 A, b 10.8 A, a 15.3 A, b 146.9 B, b 87.1 B, b

94 E u r o p e a n J o u r n a l o f H o r t i c u l t u r a l S c i e n c e

duction in ‘Elstar’ (p=0.05) and in ‘Gala’ (p<0.01) (Figure

5). In 2015, treatments of 0.45 J and higher reduced FFS

(2015: p=0.02) in ‘Elstar’, while 0.28 J and higher resulted

in a reduction of FFS in ‘Gala’ 2015 (p=0.01) and 0.39 J and

higher in 2016 (p<0.001).

ϐǡǮǯʹͲͳͳȋͶȌǡ

of 0.33 J and 0.45 J caused again reduction of FFS (p=0.04;

p<0.01), while in 2014 and 2015 (Table 3) only 0.45 J result-

ed in the same effect (2014: p=0.04; 2015: p=0.02). In ‘Gala’

2014 (Figure 5), at ϐͲǤͶͷ

had a reducing effect on FFS in comparison to brf (p=0.05).

In 2015, treatments of 0.28 J and higher Ekinϐ

set at p=0.01 and p<0.01, respectively.

͵ǤϐǤThe sampling

date for laboratory analyses of fruit quality was set at the

commercial harvest date for direct marketing from the same

orchards. No correlation was found between crop load and

quality parameters except for fruit mass (Figures 4 and 5). In

Ǯǯǡ    ϐ    

equal or higher than 0.33 J in 2011, 2014 and 0.28 J in 2015

had a reduced crop load compared to brf, resulting in yield

losses (Figure 4; Table 3). In none of the trials thinning

showed a positive effect on fruit mass or percentage of mar-

ȋͶȌǤǮǯǡ ϐ

showed no reduction in yield in 2014, however, at 0.45 J av-

erage fruit mass was enhanced by 35.8 g (p=0.01) to 166.4 g

with an effect on percentage of marketable yield (p=0.01)

which reached 99.0%. In 2016, treatment with 0.39 J caused

reduction (p=0.02) of marketable yield by 5.4 kg per tree in

comparison to brf.









ͷǤȋαͲǤͳͷǡαͲǤʹ͵ǡαͲǤ͵͵ǡǦ

αͲǤͶͷȌȋȌǡȋȌǡȋȌǡ

ȋǡΨȌǣȋȌǣͷͲȂ͵ͷͲǡȋȌǣ͵ͷͳȂ͸ͲͲǡ

ȋȌǣ͸ͲͳȂͳǡͲͲͲȋȌǮǯȀͻʹͲͳͶǤ



ĎČĚėĊͷǤ Effect of thinning intensity (solid

lines = 0.15 J, dashed lines = 0.23 J, dotted

lines = 0.33 J, dash-dotted lines = 0.45 J) on

absolute crop load (CL in number of fruit per

tree), fresh mass (FM in g), total yield

(Y in kg), and percentage of marketable yield

(MY, ΨȌ    ϐ ǣ

(A) low: 50–350, (B) medium: 351–600, and

in 2014.

ĆćđĊͶǤ Change of marketable yield (fruit size >65 mm) caused by mechanical thinning treatment in comparison to basic

ȋȌǮǯȀʹ͸ϐǤ

Year

Marketable yield

at brf

(kg)

Lowest treatment

causing yield reduction*

(J)

Reduction* of marketable

yield per tree

(kg)

Reduction of marketable

yield per ha

(t)

/RZÀRZHUVHW

2011 4.2 0.33 0.9 1.1

2014 3.9 0.33 1.3 1.4

2015 12.2 0.33 6.2 4.2

0HGLXPÀRZHUVHW

2011 7.4 No reduction 0 0

2014 10.2 No reduction 0 0

2015 14.5 0.45 5.1 5

+LJKÀRZHUVHW

2011 12.3 0.33 4 2.2

2014 10.9 0.45 1.5 1.5

2015 8.9 No reduction 0 0

*: Reduction of marketable yield in comparison to brf calculated as mean value from the total yield and percentage of marketable yield >65 mm

per tree.

V o l u m e 8 6 | I s s u e 1 | F e b r u a r y 2 0 2 1

 Ǯǯ    ϐ ǡ 

treatment with 0.33 J reduced crop load in 2011 and 0.45 J

in 2015, hence yield was reduced. In 2011, thinning at 0.33

increased fruit mass (p=0.03) by 14.6 g in comparison to

brf, however without effect on percentage of marketable

ȋͶȌǤʹͲͳͶǡǮǯϐ

responded to the treatment with 0.45 J (p<0.001) (Figure

5). Already at 0.33 J, fruit mass (p=0.04) was enhanced. At

0.45 J, average yield per tree was reduced with positive effect

on percentage of marketable yield (p<0.001), which was

enhanced by 20.7% to 97.5% (Figure 5). No effect was found

in 2015 and 2016.

  ϐ ǡ  ʹͲͳͳ     

‘Elstar’ at 0.33 J (p<0.01) and 0.45 J (p=0.02) reducing

yield per tree with no effect on average fruit mass (Figure 4;

Table 4). At ϐǮǯʹͲͳͷ

0.45 J reduced crop load (p=0.02), while gaining fruit mass

(p=0.02) and increasing the marketable yield (p=0.004)

(Table 3). In ‘Gala’ 2014, only treatment with 0.45 J reduced

crop load (p=0.03) and percentage of marketable yield was

enhanced (p=0.001) to 99.3% (Figure 5). In 2015 and 2016,

ϐǡͲǤʹͺ J reduced crop load and yield per tree

in comparison to brf. Average fruit mass was not affected

by thinning treatment, as well as percentage of marketable

yield, which was already close to 100% at brf. Due to lack

of changes in the marketable yield, which was already high

in years 2015 and 2016, Tables 3 and 4 weren’t reproduced

for ‘Gala’.

Discussion



ϐ

and was described earlier (Bukovac et al., 2010; Stopar, 2010;

«ǤǡʹͲͳͶǢǤǡʹͲͳͶǢǤǡʹͲͳ͸ȌǤ

 ǡ ϐ

ϐ

order to investigate the effect of thinning intensity for each

class separately. The size of the classes resulted from the

pragmatic approach to meet the requirements for statistical

ȋͳȌǤϐ

ϐȋǡʹͲͳͲȌǤ

  ϐ       

explained with the light terrain gradient within the orchard

causing spatial distribution of water availability (Moore et al.,

1993), nutrients supply (Aandahl, 1948), and occurrence of

frost (Weise, 1978). Since these factors have a known effect on

crop load (Powell, 1974; Hansen, 1980; Heinicke, 1917), they

ϐϐǡ

 ϐ Ǥ   ǡ     

ϐ

slightly lower elevation showing stronger vegetative growth

assumingly due to late frost events. In cultivar ‘Elstar’, a higher

ϐ

to ‘Gala’, which was expected because ‘Elstar’ is known in

practice as biannual bearing cultivar. In order to achieve the

growth capacity, trees should bear approximately one apple

per 14–42 leaves (Haller and Magness, 1933; Preston, 1954;

Silbereisen, 1966; Hansen, 1969) depending on cultivar and

growth factors. Unpublished data of 2008 from the same

‘Elstar’ orchard showed that the leaf number per trees

ranged between 2,600 and 4,540. Under the assumption

that the necessary leaf number per fruit to produce a

marketable size is 30, the trees can hypothetically support

87–151 fruit. This huge variation points to spatial variation

in growth production target (PT) which was in average 117

Ǥϐ

crop load per tree should be slightly enhanced to the PT due

  ϐǤ 

‘Elstar’, the fruit drop resulted in crop loads slightly below

the PT. In ‘Gala’, on the other hand, a high percentage of

Ǥϐ

practice is time consuming, but necessary to conclude on

Ǥϐ

is required and can be performed with optical sensors and

  ȋ«  Ǥǡ ʹͲͳͶǢ 

 Ǥǡ ʹͲͳ͹ȌǤ    ϐ  

ϐ

ȋϐǤǡʹͲͳ͸Ȍǡ

the thinning requests of individual trees are still rare.

ϐ

For comparing the effect of thinning treatments of

 ǡ   ϐ  

regarding the variables of the devices and treatments. Most

ǡϐǡǡ

mass of the strings, the tractor speed, and the rotational

frequency. The formula developed by Zoth (2011) was

ϐ

to the driving direction as a separate factor, which also has an

effect on the kinetic energy, Ekinȋ͵ȌǤϐǡ

the average Ekin is emphasized, which is achieved in the

middle of the string. Ekin at different points of the string

varies, consistently according to the distance to the rotating

spindle (Figure 3). The Ekin is limited because it ignores the

frequency of hits per string in the turbulent application,

which is affected by the distance between the device and the

tree and the amount of strings of the device. Particularly, the

ϐϐ

(Kon et al., 2013), but more experiments are needed to test

linearity of various numbers of strings and resulting Ekin. For

this purpose the distance of the tractor to the tree row needs

to be included to calculate the size of the section of tree row

which one string possibly hits at one rotation and the depth

one string penetrates the canopy. This could be a helpful

tool for further optimization of the settings of the device

and additionally the shape of the canopy. Clearly visible

already with the present simple approach is the effect of the

length of one string, which explains the frequently observed

phenomenon in practice that used, worn-out strings demand

ϐ

new strings.

ϐǡǡ

Mechanical thinning is an effective measure for crop load

management in practice (Schröder, 1996; Bertschinger et

al., 1998; Damerow et al., 2007; Weibel et al., 2008; Schupp

et al., 2008; Kong et al., 2009; Solomakhin and Blanke,

2010; Hehnen et al., 2012; Kon et al., 2013; Sinatsch et al.,

2014; McClure and Cline, 2015; Beber, 2016; Lordan et al.,

2018). An overview about the settings of the devices and

ϐ

al. (2013). A model was developed by Lordan et al. (2018)

ϐ

ϐ

considering vehicle speed and rotational frequency. All

previous studies report a reduction of fruit set by means

ϐǦ

control. Few work was published on thinning intensity

ϐȋǡʹͲͳͲǢ

96 E u r o p e a n J o u r n a l o f H o r t i c u l t u r a l S c i e n c e

et al., 2016) although spatial variation within orchards was

proved earlier (Manfrini et al., 2009; Aggelopoulou et al.,

ʹͲͳͲǡʹͲͳͳǢǤǡʹͲͳͲȌǤȋʹͲͳͲȌϐ

ϐ

assess the effect of varying thinning treatments. In the

present study, the spatial correlation was recognized by

means of semi-variograms (Crawley, 2013).

Results indicate an expected close relationship between

ϐϐǤϐǡ

drop was apparently reduced as pointed out earlier (Penzel

ǤǡʹͲʹͲȌǤǡϐ

(FFS) after varying thinning treatment should be carried

ϐǡ

which is a factor with a strong effect on FFS. Furthermore,

a positive correlation exists between mechanical thinning

       ϐ  

ϐǤ ϐǡ

was not necessary, because in none of the trials the

production target of 119 was realized. The 0.23 J treatment

showed no effect on FFS in comparison to brf. Consistently,

the fruit mass and percentage of marketable yield was hardly

ǡϐ

literature on mechanical thinning (Beber et al., 2016) and

chemical thinning (Greene, 1989; Stophar, 2010). Beber et al.

(2016) pointed out that thinning is not necessary to achieve

δͺͲϐǡ

δͶͲͲϐ

ϐϐǮǯǤ

thinning treatment further reduced FFS leading to absolute

yield losses. The cortex cell number is positively correlated

with fruit mass and thinning at full bloom is the optimum

     ȋϐ  Ǥǡ ͳͻͻͷȌǤ

However, in the present trials effect on fruit mass was

limited, since it can be assumed that the remaining fruit had a

ϐǡǤǡ

no effect of thinning treatments on fruit maturity was found

for trees with low crop load as previous studies suggested

(Volz et al., 1993; Wünsche and Ferguson, 2005) considering

    ǡ ϐǡ 

starch index.

ϐǡ

as PT of 119 in ‘Elstar’ was obtained in 2015 and ‘Gala’ in 2014

Ǥǡϐ

treatment was below the optimum and further reduced by

ηͲǤ͵͵ǮǯηͲǤ͵͵ ηͲǤͶͷ J,

depending on the year, in ‘Elstar’. However, the percentage of

marketable yield showed no increase in ‘Elstar’ considering

any year, pointing out that obviously a crop load equal

or lower than the PT will have no effect on percentage of

marketable yield. Though, average fruit mass of ‘Elstar’ 2011

was further enhanced by thinning treatments, which may

indeed have a positive effect on market value. In ‘Gala’, fruit

ϐǡ

percentage of marketable yield was enhanced only in 2014.

In the subsequent years the low fruit set resulted in high

percentage of marketable yield of >90% in all treatments.

ǡ    ϐ   

yield losses as the effect on quality was marginal. Maximum

ϐ

should be 0.23 J for the two cultivars.

ϐ  

at brf (Sinatsch et al., 2014; McClure and Cline, 2015). For

‘Gala’ in 2014 treatment of 0.33 J was appropriate to reach

PT. Enhanced thinning treatment reduced fruit set heavily,

and fruit mass was increased with every fruit having a

marketable size. Lower thinning treatments resulted in crop

load exceeding PT with negative effect on fruit mass. Kon

(2013) reported for similar EkinǡηͲǤʹ͹ǡ

ϐǡǤ

ϐ

than in the present study because of the lower vehicle speed.

ǡ ȋʹͲͳͷȌ 

vehicle speed was low. As compromise of thinning response

and tree damage, Kon (2013) suggested thinning at lower

rotational frequencies equaling 0.14 J or 0.20 J for ‘Gala’.

However, it was emphasized that these intensities may not

ϐ

the combination with other thinning methods. McClure and

Cline (2015) also reported no effect on fruit mass when

thinned up to 0.29 J at low vehicle speed, though marketable

yield was adequate already in the control.

In ‘Elstar’ 2011, treatment of 0.23 J reduced crop load

close to PT, though no effect on fruit mass was achieved.

This may have resulted from damage on spur leaves, which

was not further evaluated. In 2014, treatments of 0.23 J and

0.33 J provided the optimum thinning treatment in terms of

crop load, while in 2015 treatments >0.28 J and <0.45 J have

caused best results regarding crop load level and fruit mass

as crop load at applied treatments exceeded or underrun

Ǥǡϐ

‘Elstar’, 0.23 J and 0.33 J are adequate treatments to reduce

crop load. Sinatsch et al. (2014) suggested treatments of

0.19 J or 0.23 Ǯǯϐ

affecting fruit size and reducing biannual bearing.

In summary all treatments indicate that different levels

ϐ

for best thinning results. A model for thinning response

 ϐ 

high R2 as Lordan and co-workers (Lordan et al., 2018)

developed for two cultivars was not possible from the present

data. However, data indicate that tree adapted mechanical

thinning can decrease possible yield losses by over-thinning

ϐǤϐ

requires enhanced thinning intensity as crop loads possibly

ǤϐǦ

uniform mechanical thinning ranged from 1.4 t ha-1 – 4.2 t ha-1

in ‘Elstar’ in years 2011, 2014, 2015, and 2.6 t ha-1 – 7.6 t ha-1

in ‘Gala’ in 2014. Consequently, tree-adapted thinning is a

promising method to increase production volume without

increasing land consumption. When commercial systems

for tree-adapted thinning become available, the economic

aspect is crucial for the farmer.

Conclusion

ϐ

ϐ

      ϐ 

intensity. The concept of mechanical thinning considering the

ϐ

to reduce yield losses by over-thinning of trees with low

ϐǡͳǤͶȂͶǤʹ ha-1

in ‘Elstar’ and 2.6–7.6 t ha-1 in ‘Gala’. When commercially

available, this approach of precise management can

potentially balance spatial heterogeneity within orchards.

Acknowledgments

The authors acknowledge Dr. K. Gottschalk for his

expertise on developing the formula for calculating kinetic

energy of rotating strings and Dr. M. Schirrmann for his

support on the application of spatial statistic.

V o l u m e 8 6 | I s s u e 1 | F e b r u a r y 2 0 2 1

References

Aandahl, A.R. (1948). The characterization of slope positions and

ϐ

Iowa. Soil Sci. Soc. Am. Proc. 13, 449–454. https://doi.org/10.2136/

ͳͻͶͻǤͲ͵͸ͳͷͻͻͷͲͲͳ͵ͲͲͲͲͲͺͳǤ

Aggelopoulou, K.D., Wulfsohn, D., Fountas, S., Gemtos, T.A., Nanos,

G.D., and Blackmore, S. (2010). Spatial variation in yield and quality

in a small apple orchard. Precision Agric. 11, 538–556. https://doi.

org/10.1007/s11119-009-9146-9.

Aggelopoulou, K.D., Pateras, D., Fountas, S., Gemtos, T.A., and Nanos,

ǤǤ ȋʹͲͳͳȌǤ     Ǧϐ 

maps in an apple orchard. Precision Agric. 12, 118–129. https://doi.

org/10.1007/s11119-010-9161-x.

Bangerth, F. (2000). Abscission and thinning of young fruit and

their regulation by plant hormones and bioregulators. Plant Growth

Regul. 31, 43–49. https://doi.org/10.1023/A:1006398513703.

ǡ Ǥǡ  ǡ Ǥǡ  «ǡ Ǥ ȋʹͲͳ͸ȌǤ  ϐ

of mechanical thinning on fruit quality and constant bearing

of ‘Jonagold’ apples. Acta Hortic. 1139, 513–518. https://doi.

org/10.17660/ActaHortic.2016.1139.88.

Bertschinger, L., Stadler, L., Stadler, W., Weibel, F., and Schumacher, R.

ȋͳͻͻͺȌǤϐ

and fruit set and of alternate bearing of apple crop. Acta Hortic. 466,

65–70. http://dx.doi.org/10.17660/ActaHortic.1998.466.11.

Bukovac, M.J., Sabbatini, P., Zucconi, F., and Schwallier, P.G. (2010).

Ǧϐ

spur-type ‘Delicious’ apple. HortScience 45, 22–29. https://doi.

org/10.21273/HORTSCI.45.1.22.

Crawley, M.J. (2013). The R book, 2nd edn. (Chichester, U.K.: Wiley),

p. 860–866.

Damerow, L., Kunz, A., and Blanke, M. (2007). Mechanische

Fruchtbehangsregulierung. Erwerbs-Obstbau 49, 1–9. https://doi.

org/10.1007/s10341-007-0029-9.

Dennis, F.G. (2000). The history of fruit thinning. Plant Growth Regul.

31, 1–16. https://doi.org/10.1023/A:1006330009160.

Dinno, A. (2017). Dunn’s test of multiple comparisons using rank

sums. Ǧ ǮǤǯǤ ǣȀȀǤǦǤȀȀ

packages/dunn.test/dunn.test.pdf.

ϐǡǤǤǡǡǤǤǡǡǤǤȋͳͻͻͷȌǤ

of ‘Empire’ apple fruit size and anatomy in unthinned and hand-

thinned trees. J. Hortic. Sci. 70, 375–387. https://doi.org/10.1080/

14620316.1995.11515307.

Greene, D.W. (1989). Regulation of fruit set in tree fruits with

plant growth regulators. Acta Hortic. 239, 323–334. https://doi.

org/10.17660/ActaHortic.1989.239.51.

Haller, M.H., and Magness, J.R. (1933). Relation of leaf area and

position to quality of fruit and to bud differentiation in apples. U.S.

Dept. Agric. Techn. Bull. 338.

Hansen, P. (1969). 14C-studies on apple trees. IV. Photosynthate

consumption in fruits in relation to the leaf-fruit ratio and to the

leaf-fruit position. Physiologia Plantarum 22, 186–198. https://doi.

ȀͳͲǤͳͳͳͳȀǤͳ͵ͻͻǦ͵ͲͷͶǤͳͻ͸ͻǤͲ͹ͺͷͷǤǤ

Hansen, P. (1977). The relative importance of fruits and leaves for

ǦϐǤǤǤǤ52, 501–

508. https://doi.org/10.1080/00221589.1977.11514780.

Hansen, P. (1980). Yield components and fruit development in

‘Golden Delicious’ apples as affected by the timing of nitrogen

supply. Sci. Hortic. 12, 243–257. https://doi.org/10.1016/0304-

4238(80)90005-9.

Hehnen, D., Hanrahan, I., Lewis, K., McFerson, J., and Blanke, M.

ȋʹͲͳʹȌǤϐ

and promotes consistent bearing. Sci. Hortic. 134, 241–244. https://

ǤȀͳͲǤͳͲͳ͸ȀǤǤʹͲͳͳǤͳͳǤͲͳͳǤ

ǡǤǤȋͳͻͳ͹ȌǤϐϐ

and partially developed fruits of apple (Pyrus malus L.). Cornell Univ.

Experim. Station Bull. 393, 41–114. https://doi.org/10.5962/bhl.

title.36986.

«ǡǤǡeǡǤǡæǡǤǡǡǤȋʹͲͳͶȌǤ

estimation in apple orchards by image analysis. Precision Agric. 15,

466–478. https://doi.org/10.1007/s11119-013-9341-6.

Hothorn, T., Bretz, F., and Westfall, P. (2008). Simultaneous inference

in general parametric models. Biometrical J. 50, 346–363. https://

ǤȀͳͲǤͳͲͲʹȀǤʹͲͲͺͳͲͶʹͷǤ

Jonkers, H. (1979). Biennial bearing in apple and pear: A literature

survey. Sci. Hortic. 11, 303–317. https://doi.org/10.1016/0304-

4238(79)90015-3.

Kon, T.M., Schupp, J.R., Winzeler, H.E., and Marini, R.P. (2013).

ϐ

reproductive tissues, fruit set, yield, and fruit quality of ‘Gala’ apple.

HortScience 48, 40–46. https://doi.org/10.21273/HORTSCI.48.1.40.

Kong, T., Damerow, L., and Blanke, M. (2009). ϐ 

mechanischer Fruchtbehangsregulierung auf Ethylensynthese

als Stressindikator sowie Ertrag und Fruchtqualität bei Kernobst.

Erwerbs-Obstbau 51, 39–53. https://doi.org/10.1007/s10341-009-

0080-9.

Krikeb, O., Alchanatis, V., Crane, O., and Naor, A. (2017). Evaluation

ϐ

ϐǤǤǤ8, 466–470. https://

doi.org/10.1017/S2040470017001406.

Lakso, A.N. (2011). Early fruit growth and drop – The role of carbon

balance in the apple tree. Acta Hortic. 903, 733–742. https://doi.

org/10.17660/ActaHortic.2011.903.102.

Looney, N.E. (1993). Improving fruit size, appearance, and

other effects of fruit crop ‘quality’ with plant bioregulating

chemicals. Acta Hortic. 329, 120–127. https://doi.org/10.17660/

ActaHortic.1993.329.21.

Lordan, J., Alins, G., Àvila, G., Torres, E., Carbó, J., Bonany, J., and

Alegre, S. (2018). Screening of eco-friendly thinning agents and

ǮǯǡǮǯǮǯ

apple trees. Sci. Hortic. 239ǡ ͳͶͳȂͳͷͷǤ ǣȀȀǤȀͳͲǤͳͲͳ͸ȀǤ

scienta.2018.05.027.

Luckwill, L.C. (1953). Studies of fruit development in relation to

plant hormones. I. Hormone production by the developing apple

seed in relation to fruit drop. J. Hortic. Sci. 28, 14–24. https://doi.org

/10.1080/00221589.1953.11513767.

MacDaniels, L.H., and Heinicke, A.J. (1929). Pollination and other

factors affecting the set of fruit, with special reference to the Apple.

Cornell Univ. Experim. Station Bull. 497.

Manfrini, L., Taylor, J.A., and Grappadelli, L.C. (2009). Spatial analysis

of the effect of fruit thinning on apple crop load. Eur. J. Hortic. Sci.

74, 54–60.

McClure, K.A., and Cline, J.A. (2015). Mechanical blossom thinning

   ϐ  ǡ      Ǥ

Canadian J. Plant Sci. 95ǡ ͺͺ͹Ȃͺͻ͸Ǥ ǣȀȀǤȀͳͲǤͶͳͶͳȀǦ

2014-421.

Meier, U. (1997). BBCH-Monograph. Growth Stages of Mono- and

Dicotyledonous Plants (Berlin, Wien: Blackwell).

Moore, I.D., Gessler, P.E., Nielsen, G.A., and Peterson, G.A.

(1993). Soil attribute prediction using terrain analysis.

Soil Sci. Soc. Am. J. 57, 443–452. https://doi.org/10.2136/

ͳͻͻ͵ǤͲ͵͸ͳͷͻͻͷͲͲͷ͹ͲͲͲʹͲͲʹ͸Ǥ

98 E u r o p e a n J o u r n a l o f H o r t i c u l t u r a l S c i e n c e

Palmer, J.W., Giuliani, R., and Adams, H.M. (1997). Effect of crop

load on fruiting and leaf photosynthesis of ‘Braeburn’/M.26

apple trees. Tree Physiol. 17, 741–746. https://doi.org/10.1093/

treephys/17.11.741.

Pebesma, E.J. (2004). Multivariable geostatistics in S: the

gstat package. Computers & Geosci. 30, 683–691. https://doi.

ȀͳͲǤͳͲͳ͸ȀǤǤʹͲͲͶǤͲ͵ǤͲͳʹǤ

Pebesma, E.J., and Bivand, R.S. (2005). Classes and methods for spatial

data in R. R News 5(2). ǣȀȀǤǦǤȀȀȀǤ

ǡ Ǥǡ ϐǡ Ǥǡ ǡ Ǥǡ  Ǧǡ Ǥ ȋʹͲʹͲȌǤ

Mechanical thinning of apples reduces fruit drop. Acta Hortic. 1281,

533–538. https://doi.org/10.17660/ActaHortic.2020.1281.70.

ϐǡǤǡǡǤǡǡǤȋʹͲͳ͸ȌǤϐǦ

ϐ

    ϐǤ  Ǥ 1130,

605–612. https://doi.org/10.17660/ActaHortic.2016.1130.90.

Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., and R Core Team (2018).

nlme: Linear and Nonlinear Mixed Effects Models. R Package v. 3.1-

ͳ͵͹ǤǣȀȀǤǦǤȀαǤ

Powell, D.B.B. (1974). Some effects of water stress in late spring on

apple trees. J. Hortic. Sci. 49, 257–272. https://doi.org/10.1080/002

21589.1974.11514578.

Preston, A.P. (1954). Effects of fruit thinning by the leaf count method

on yield, size and biennial bearing of the apple Duchess Favorite.

J. Hortic. Sci. 43, 373–381.

R Core Team (2018). R: A language and environment for statistical

computing (Vienna, Austria: R Foundation for Statistical Computing).

ǣǣȀȀǤǦǤȀǤ

Schröder, M. (1996). Maschinell ausdünnen. Obst und Garten 3,

116–117.

Schupp, J.R., Baugher, T.A., Miller, S.S., Harsh, R.M., and Lesser, K.M.

(2008). Mechanical thinning of peach and apple trees reduces labor

input and increases fruit size. Hort Technology 18, 660–670. https://

doi.org/10.21273/HORTTECH.18.4.660.

Silbereisen, R. (1966). Beziehungen zwischen Fruchtausbildung,

Blatt-Frucht-Verhältnis und Wärmeklima bei ‘Golden Delicious’:

I. Abhängigkeit des Fruchtgewichtes vom Blatt-Frucht-Verhältnis

und vom Wärmeklima. Gartenbauwissenschaft 3, 267–295.

Sinatsch, S., Pfeiffer, B., Schult, I., Zimmer, J., and Brockamp, L. (2014).

Crop regulation in organic grown apples – Results of different

trials on three different sites. Proc. 16th Int. Conf. on Organic Fruit-

Growing, Feb. 17–19, Hohenheim, Germany (Weinsberg: Föko e.V.),

p. 142–152.

ǡ ǤǤǡ  ǡ ǤǤ ȋʹͲͳͲȌǤ  ϐ

thinning improves the fruit quality of apples. J. Sci. Food Agric. 90,

͹͵ͷȂ͹ͶͳǤǣȀȀǤȀͳͲǤͳͲͲʹȀǤ͵ͺ͹ͷǤ

Stopar, M. (2010). Fruit set and return bloom of light, medium and

ϐǤǤ884,

351–356. https://doi.org/10.17660/ActaHortic.2010.884.41.

Stover, E.W., and Greene, D.W. (2005). Environmental effects

on the performance of foliar applied plant growth regulators.

HortTechnology 15, 214–221. https://doi.org/10.21273/

HORTTECH.15.2.0214.

Volz, R.K., Ferguson, I.B., Bowen, J.H., and Watkins, C.B. (1993). Crop

load effects on mineral nutrition, maturity, fruiting and tree growth

of ‘Cox’s Orange Pippin’ apple. J. Hortic. Sci. 68, 127–137. https://doi.

org/10.1080/00221589.1993.11516336.

Weibel, F.P., Chevillat, V.S., Rios, E., Tschabold, J.-L., and Stadler, W.

(2008). Fruit thinning in organic apple growing with optimised

strategies including natural spray products and rope-device. Eur. J.

Hortic. Sci. 73, 145–154.

Weise, A. (1978). Zum Problem der Frostgefährdung im Obstbau

unter Berücksichtigung von Gebieten mit geringer Reliefenergie.

Archiv für Gartenbau 26, 349–359.

Wünsche, J.N., and Ferguson, I.B. (2005). Crop load interactions in

apple. Hortic. Rev. 31, 233–292.

Wünsche, J.N., Palmer, J.W., and Greer, D.H. (2000). Effects of crop

load on fruiting and gas exchange characteristics of ‘Braeburn’/M.26

apple trees at full canopy. J. Am. Soc. Hortic. Sci. 125, 93–99. https://

doi.org/10.21273/JASHS.125.1.93.

Zoth, M. (2011). Untersuchung zur abgestuften Ausdünnungswirkung

der ‘DARWIN’-Fadenmaschine mittels Staffelung der kinetischen

Rotationsenergie. DGG-Proceedings 17, 1–5. DOI: 10.5288/dgg-pr-

01-17-mz-2011.

Zude-Sasse, M., Fountas, S., Gemtos, T.A., and Abu-Khalaf, N. (2016).

Applications of precision agriculture in horticultural crops. Eur. J.

Hortic. Sci. 81, 78–90. https://doi.org/10.17660/eJHS.2016/81.2.2.

Received: May 23, 2019

Accepted: Apr. 6, 2020

Address of authors:

Martin Penzel1,2,ȗǡϐ2, Robin Gebbers2 and

Manuela Zude-Sasse2

1 Technische Universität Berlin, Chair of Agromechatronics,

Straße des 17. Juni 135, 10623 Berlin, Germany

2 Leibniz Institute for Agricultural Engineering and

Bioeconomy (ATB), Potsdam, Germany

* Corresponding author;

Tel.: +49-331-5699-915; Fax: +49-331-5699-849