Microsoft Word - 54catania.docx


 CHEMICAL ENGINEERING TRANSACTIONS  
 

VOL. 44, 2015 

A publication of 

The Italian Association 
of Chemical Engineering 
Online at www.aidic.it/cet 

Guest Editors: Riccardo Guidetti, Luigi Bodria, Stanley Best
Copyright © 2015, AIDIC Servizi S.r.l., 
ISBN 978-88-95608-35-8; ISSN 2283-9216                                                                               

 

 

Monitoring Strategies for Precise Production of High Quality 
Fruit and Yield in Apple in Emilia-Romagna 

Luigi L. Manfrini*a,c, Emanuele E. Pierpaolia,c, Marco M. Zibordia,c, Brunella B. 
Morandia,c, Enrico E. Muzzia, Pasquale P. Loscialeb,c, Luca L. Corelli Grappadellia,c 
aDipartimento Scienze Agrarie, University of Bologna, Bologna Italy 
bAgricultural Research Council, Research Unit for Agriculture in Dry Environments, Bari Italy 
cHK-Horticultural Knowledge srl, Bologna Italy 
luigi.manfrini@unibo.it 

Research has shown that pomefruit orchard production can be forecast. Basic information about orchard 
characteristics as tree density, fruit load and fruit size can translate into useful information on orchard 
production. In most cases the complete flow of this information is only available after harvest. Precision 
horticulture integrates this information in order to evaluate the management performed by the grower, and to 
assess whether it is in line to achieve the desired targets or not. The Department of Agricultural Sciences of 
the University of Bologna and HK-Horticultural Knowledge (HK, 2011) after several years of conducting 
experiences of precise fruit growing in commercial orchards cooperate to bring to fruition this novel approach. 
Precise measurement of fruit growth is justified on the concept that the fruit can be considered as the most 
relevant indicator of how well is the process progressing. A four year experiment will be presented. Fruit 
maximum diameter was recorded every other week during the cell expansion phase of growth. The data were 
processed and a diagnosis was released in real time to the grower, assessing the progress of the crop, and 
serving as a verification that the applied management techniques are yielding the expected production result. 
The real time forecast has been used to evaluate a comparison between predicted and the real breakdown of 
the total yield into size categories, which allows to release also a very accurate estimate of orchard yield for 
the packing house. 

1. Introduction 
Decisions on orchard management are in most cases based on visual and subjective observation, a result of 
experience rather than based on objective information (Manfrini et al. 2012). The monitoring strategies mostly 
used release information on environmental variation, which is not directly/easily referred to crop status, often 
making the interpretation of the seasonal production system difficult (Schueller et al. 1999). Horticultural 
systems contain a lot of unused or hidden information on the crop (Zadravec et al. 2013 ). These data are an 
important source of information for growers, technicians and pack-houses but often untapped. Parameters 
such as fruit diameter (Morandi et al. 2007), crop load (Manfrini et al.,2009) or calculated production 
information (Manfrini et al. 2012) can be recorded or calculated along the season. However real-time 
monitoring systems on fruit production are not common and management is usually performed using 
subjective practices (Ellis et al. 2010). Growers need real time information for many issues related to crop 
status. In apple production systems, the control of fruit size, and consequently apple quality (Link, 2000), can 
be achieved through the application of plant growth regulators (Marini et al. 2003) and other managing 
practices as irrigation scheduling for controlling both vegetative and fruit growth (Ebel et al., 1995). However, 
not many activities related to production and quality are the focus of monitoring or subjected to quality control 
processes to assess their usefulness before harvest (Manfrini et al. 2009). Although Decision Support 
Systems (DSS) has increased since the 1980s, and have been developed in similar fields as viticulture (Rossi 
et al. 2014) DSS for managing fruit growth are missing because of the absence of a quick and effective 

                               
 
 

 

 
   

                                                  
DOI: 10.3303/CET1544051

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Please cite this article as: Manfrini L., Pierpaoli E., Zibordi M., Morandi B., Muzzi E., Losciale P., Corelli Grappadelli L., 2015, Monitoring 
strategies for precise production of high quality fruit and yield in apple in emilia-romagna, Chemical Engineering Transactions, 44, 301-306  
DOI: 10.3303/CET1544051

301



 

 

indicator that can be easily implemented in commercial fruit production. Because they are busy and have little 
time, fruit growers need monitoring schemes that are both rapid and accurate. The main aim of this paper is to 
present an approach that may help fill this knowledge gap. The intention is to provide both the grower and the 
consultant with a real time decision support system at critical times of the season and to provide information to 
help packing-houses to better plan harvest and logistics. In this case study, fruit growing monitoring tools have 
been applied to an apple orchard in Italy along the four different years. 

2. 2. Materials and Methods 
2.1  Location and Plant Material  

The study was conducted between 2010 and 2013 growing season in a commercial apple orchard located in 
the province of Cesena, Italy. Data were collected from a 1.36 ha block of Pink Lady® apple trees planted in 
2004, grafted on M9 rootstock and trained as slender spindle at the density of 2381 trees ha-1 (4.2 X 1.0  m); 
the trees were under a standard management protocol. 

2.2 Data Collection and Rational Use of Tools 

In the four years of the experiment, beginning after completion of cytokinesis, i.e. the final week of June or the 
first of July, the maximum equatorial diameter of 20 fruit on 14 randomly selected trees among the rows for a 
total of 280 fruit/orchard was recorded as described in figure 1. 

 

Figure 1. Tree selection strategy for data collection. The trees and rows selection have been randomly 
executed every measurements. On each tree 20 maximum diameter measurements have been performed. 

The measurements normally took place every other week for a total of 5/6 times along the growing season. 
Data were recorded by a Calibit (Calibit, 2013) a digital calliper fitted with an onboard datalogger (figure 2). 
This calliper allows a single person to record all 280 measurements in less than 30 minutes each time and a 
total of a less than 3 hours in a season. Because the growth rate of apple fruit has been shown to be linear in 
the second part of the season (Goffinet et al. 1995) when the fruit enlargement is driven by cell expansion, an 
accurate prediction of fruit diameter (size) at a nominal harvest date can be made as reported in table 1 
(Lakso et al. 1995). In this experiment the harvest date was set on October 20th. A linear regression has been 
calculated on all the predicted vs. actual data and an R2 value >0.98 was found. Since the diameter of the fruit 
has also been shown to be strongly related to the mass of the fruit (De Silva et al. 1997) this information can 
be used to predict fruit weight by the equation (1): 

W = a * D(mm)b (1) 

302



 

 

where W is the fruit weight in grams, D is the fruit diameter expressed in millimetre and a and b are 
parameters specific to Pink Lady®. These parameters were obtained in 2007 by regressing diameter and 
weight data from a large number of fruit picked during the whole season; the R2 of the relationship was > 0.99. 
The same equation allows forecasting fruit distribution in size class categories both in percentage (table 2) 
and t/ha (table 3). This was done according to the commercial standards, which increase by 5 mm, starting 
from 65 mm to 90 mm, was also assessed (table 2 and 3). The data for this paper was analysed comparing 
predicted values with real orchard production data using as a variate the different years applying respectively 
the glimmix procedure and a Mixed model analysis. Crop load was assessed by fruit counts on the last 
measurement performed in August, following the protocol described by Manfrini et al. (2009). The crop load 
(number fruit/tree) was 106, 160, 67, 161, from 2010 to 2013 respectively. The multiplication between the 
expected fruit size at harvest obtained in the last measurement prediction (table 1), the crop load and the tree 
density, permit predicting the orchard production of the current year (table 2).  
 

 

Figure 2. Digital Caliper with onboard datalogger (Calibit). 1Datalogger; 2Recording Button; 3 Data display 

Table 1. Growth measurement among the years and predicted diameter . First day of measurement do not 
include a predicted diameter because two subsequent dates are needed to obtain the forecast 

Year 2013 
Day of Measurement  3/7 12/7 26/7 9/8 2/10 

Average diameter measured (mm)  42.7 46.0 51.8 55.0 67.0 
Predicted diameter at harvest (mm)   69.9 73.7 72.0 71.8 

Year 2012 
Day of Measurement  22/6 10/7 26/7 10/8 25/9 

Average diameter measured (mm)  40.1 47.6 54.1 60.1 69.6 
Predicted diameter at harvest (mm)   71.2 73.7 75.9 74.6 

Year 2011 
Day of Measurement  15/6 30/6 13/7 27/7 9/8 

Average diameter measured (mm)  38.9 43.8 49.6 54.0 59.1 
Predicted diameter at harvest (mm)   64.7 69.8 71.0 73.1 

Year 2010 
Day of Measurement 22/6 6/7 20/7 3/8 16/8 10/9 

Average diameter measured (mm) 41.3 48.3 53.5 59.8 62.8 67.3 
Predicted diameter at harvest (mm)  75.4 75.2 77.7 77.4 75.8 

3. Result and Discussion 
Pink Lady® is a cultivar with a high selling value for growers, but only if fruit size at harvest (not considering 
the blush color) reaches between 70 and 85 mm. In 2011, the forecast average fruit diameter on June 30th 

303



 

 

falls in a smaller category with a size equal to 64.7 mm (table 1). Because in this period fruit growth is still very 
sensitive to crop load, additional fruit thinning (a key management practice) was suggested. A reduction of the 
carbon sinks on the tree and a smaller number of fruit among which to partition tree resources would have a 
positive effect to the fruit growth (Morandi et al. 2011). The response to thinning is apparent in the June 13th 
forecast, where the predicted diameter at harvest was close to 70 mm (table 1). This indicates the usefulness 
of the thinning treatment, even more so in this case, since the subsequent August fruit count revealed a still 
quite high crop load (average of 160 fruit/tree). The effects of excessive crop load can be noticed in 2013, 
when 161 fruit/tree were estimated and the expected diameter was the smallest (71.8 mm) of the 4 years 
experiment (table 1). Also the predicted and harvested breakdown in size categories, both in percentage and 
in quantity (t/ha), confirm a higher occurrence of the smaller fruit categories, with more than 50 of the total 
production smaller than the 75 mm class, even though the total production was the highest (table 3). This 
result could be undesired because profit from a high quality apple, such as Pink Lady®, often does not 
correlate with the quantity but the quality (size) of production. Very few anomalies were recorded in fruit 
growth in the remaining measurements of the four seasons, indicating a prediction relatively constant and 
always above the optimum size range. This could have been used for providing feedback for other managing 
activities such as irrigation scheduling. As well as generating a mean diameter forecast, the last August 
measurement was used to forecast a harvest size for every single fruit monitored (280 fruit in total), with the 
goal of building up a total fruit size class distribution (table 2 and 3).  

Table 2. Size classes distribution of the harvest and the predicted values 

  Size Classes (%)* 
Year Data <65 65/70 70/75 75/80 80/85 85/90 >90 

2010 
Forecast 0.0 4.2 36.7 50.8 8.3 0.0 0.0 

Real 0.0 16.5 47.3 31.5 4.4 0.2 0.0 

2011 
Forecast 0.0 8.1 38.1 43.1 7.5 2.5 0.6 

Real 0.0 12.3 44.1 35.0 8.1 0.5 0.0 

2012 
Forecast 0.0 4.6 12.9 37.5 30.7 11.8 2.5 

Real 0.0 5.4 22.8 33.8 27.8 9.0 1.2 

2013 
Forecast 7.1 26.1 43.2 21.8 1.8 0.0 0.0 

Real 0.0 11.8 53.3 31.7 3.0 0.1 0.0 
*Means separation within columns by glimmix procedure  

Table 3. Total yield and size classes distribution of the harvest and the predicted values. The expected fruit 
production per ha was calculated considering a tree density of 2381 tree/ha 

   Size Classes (t/ha)* 

Year Data Total Production (t/ha)* <65 65/70 70/75 75/80 80/85 85/90 >90 

2010 
Forecast 45.15 0.00 1.88 16.56 22.95 3.76 0.00 0.00 

Real 54.92 0.00 9.09 25.96 17.32 2.40 0.13 0.02 

2011 
Forecast 67.35 0.00 5.47 25.68 29.05 5.05 1.68 0.42 

Real 63.78 0.00 7.85 28.15 22.31 5.17 0.29 0.01 

2012 
Forecast 28.95 0.00 1.34 3.72 10.86 8.89 3.41 0.72 

Real 38.07 0.00 2.04 8.68 12.88 10.58 3.44 0.45 

2013 
Forecast 62.02 4.43 16.17 26.80 13.51 1.11 0.00 0.00 

Real 66.56 0.00 7.89 35.47 21.09 2.02 0.08 0.01 
*Means separation within columns by a Mixed model analysis. 
 
The forecast production expressed both in percentage and t/ha was plotted against actual values and no 
differences were found. This attests to the effectiveness of the protocol implemented and the reliability of the 
forecasting performed along the season. This information is also highly valuable for the sales department of 
the packing house, which can thus have a very good idea of the product they have to sell, much earlier than 
they are currently used to (2 months before harvest time). This provides them with the opportunity to start 

304



 

 

contacting perspective clients sooner in the season, and with a much clearer picture of the upcoming 
marketing scenarios. 

4. Conclusion 
This paper shows a four year case study illustrating the possibility and the advantages provided by adopting 
precise orchard management, if this is based on sound assessment of orchard performance, and robust 
modelling tools. It has been possible to provide timely feedback to the grower about a  problem of excessive 
crop load, which would have caused insufficient fruit size at harvest, at a time when it was possible, and 
practical, to correct it. The removal of fruit carried out in mid-July was effective and allowed to increase fruit 
growth enough to reach a satisfactory fruit size at harvest. Moreover, due to this real-time decision support, it 
has been possible to deliver the packing house crucial information on their crop status in terms of size 
distribution, up to two months before harvest. This is valuable because it allows the best market segmentation 
and storage management, to minimise losses and to increase efficiency of storage. The effort required by the 
grower for this type of monitoring, and the costs associated are by far smaller than the associated benefits. In 
terms of time, the 5/6 measurements performed did not take more than 3 hours per ha. To date, this type of 
information on apple during the season has not been recorded and used as a driver of labor, resource and 
other management inputs. This investigation indicates that options are available to growers who would like to 
increase their knowledge on their crop status and potentially better respond to market demands of higher 
quality fruit at a lower cost. 

Acknowledgements  

The authors would like to thank Apofruit-Cesena for providing access to the orchard and the machinery for 
collecting the production data. 

References 

De Silva, H.N., Tustin, D.S., Cashmore, W.M., Stanley, C.J., Lupton, G., McArtney, S.J., 1997, The fruit fresh 
weight diameter relationship for ‘Royal Gala’ apple across seasons and among fruit production regions of 
New Zealand. HortScience, 32, 1169–1173. 

Ebel, R.C., Proebsting, E.L., Evans, R.G., 1995, Deficit irrigation to control vegetative growth in apple and 
monitoring fruit growth to schedule irrigation. HortScience, 30, 1229–1232. 

Ellis, K., Baugher, T. A., Lewis, K., 2010, Results from survey instruments used to assess technology adoption 
for tree fruit production. HortTechnology, 20(6), 1043–1048. 

Goffinet, M.C., Robinson, T.L., Lasko, A.N., 1995, A comparison of ‘Empire’ apple fruit size and anatomy in 
unthinned and hand-thinned trees. Journal of Horticultural Sciences, 70, 375–387. 

Calibit, Digital calliper for monitoring fruit growth, 2013, <calibit.it/en> accessed 7.3.2015 
HK-Horticultural Knowledge srl, Precise Fruit Growing, 2011, <hkconsulting.it/en> accessed 7.3.2015 
Lakso A.N., Corelli Grappadelli, L., Barnard, J., Goffinet, M.C., 1995, An expolinear model of the growth 

pattern of the apple fruit. Journal of Horticultural Sciences, 70, 389–394. 
Lakso, A.N., Robinson, T.L., Greene, D.W., 2006, Integration of environment, physiology and fruit abscission 

via carbon balance modeling – Implications for understanding growth regulator responses. Acta 
Horticulturae (ISHS), 727, 321–326 

Link, H., 2000, Significance of flower and fruit thinning on fruit quality. Special issue: Abscission and thinning 
of young fruit - from molecular to applied aspects. Plant Growth Regulators, 31, 17–26 

Manfrini, L., Taylor, J.A., Corelli Grappadelli L., 2009, Spatial Analysis of the Effect of Fruit Thinning on Apple 
Crop Load. European Journal Horticultural Science, 74 (2), 54–60. 

Manfrini, L., Pierpaoli, E., Taylor, J.A., Morandi, B., Losciale, P., Zibordi, M., Corelli Grappadelli, L., Bastías, 
R.M., 2012, Precision fruit growing: How to collect and interpret data on seasonal variation in apple 
orchards. Acta Horticulturae, 932, 461-470.  

Marini, R.P., 2003, Fruitlet size and position within the cluster influence apple fruitlet susceptibility to chemical 
thinners. Journal of Horticultural Sciences & Biotechnology, 78, 813–820 

Morandi B., Zibordi M., Losciale P., Manfrini L., Pierpaoli E., Corelli Grappadelli L. 2011. Shading decreases 
the growth rate of young apple fruit by reducing their phloem import. Scientia Horticulturae, 127, 347-352  

Morandi, B., Manfrini, L., Zibordi, M., Noferini, M., Fiori, G., Corelli Grappadelli, L., 2007, A low-cost device for 
accurate and continuous measurements of fruit diameter. HortScience, 42, 1380–1382 

305



 

 

Rossi, V., Salinari, F., Poni, S., Caffi, T., Bettati, T., 2014, Addressing the implementation problem in 
agricultural decision support systems: the example of vite.net. Computers and Electronics in Agriculture, 
100, 88–99 (12) 

Schueller, J.K., Whitney, J.D., Wheaton, T.A., Miller W.M., Turner, A.E., 1999, Low-cost automatic yield 
mapping in hand-harvested citrus. Computers and Electronics in Agriculture, 23, 145–154. 

Zadravec, P., Veberic, R., Stampar, F., Eler, K., Schmitzer, V., 2013, Fruit size prediction of four apple 
cultivars: Accuracy and timing. Scientia Horticulturae, 160, 177–181 

306