Washington Tree Fruit Research Commission

Research Reports

Prediction of apple quality characteristics by tracking apple mass loss in storage. (2003)

FINAL PROJECT REPORT
WTFRC Project #
YEAR 0/0
Organization Project #
Title:Prediction of apple quality characteristics by tracking apple mass loss in storage.
PI:Dr. Steven O. Link
Organization:Washington State University Tri-Cities 2710 University Dr., WSU TC, Richland, WA 99352, 509-372-7526, slink@tricity.wsu.edu
 PDF version of report

Co-PIs

Dr. Steven Drake, USDA-TFRL, Wenatchee, WA

Objectives

The goal of the research was to determine if it is possible to electronically track mass loss for the prediction of apple quality [firmness (F), soluble solids concentration (SSC), titratable acidity (TA)] while in RA and controlled atmosphere (CA) storage conditions.  Real-time assessment of apple quality while in RA and CA storage would improve a manager’s ability to determine when to pack individual lots of fruit to maintain optimum quality.

 

Explain any deviations from original objectives or schedule:  Density analysis not yet done.

Significant findings

Introduction

Apples (Malus x domestica Borkh.) lose quality (F, SSC, TA) during storage (Hohn 1990).  The standard manner of measuring F is destructive (Hohn 1990) and requires entry into CA storage rooms for infrequent F tests.  Technologies such as acoustic impulse devices (DeBelie et al. 2000) are complicated and have only been correlated with the Magness-Taylor (Magness and Taylor 1925) pressure tester results.  Link et al. (2003) under RA storage conditions and (Link and Drake (2002) under CA storage found mass (measured by hand) loss to be linearly correlated with Magness-Taylor pressure tester results.  We are not aware of real-time non-destructive sensors that can detect or are correlated with soluble solids concentration or titratable acidity.
         In this study, we tested the hypothesis that automated measurement of mass loss is correlated with loss of F, SSC, and TA during storage.  If there is a relationship, then automated measurement of mass loss may be used to provide F, SSC, and TA values of apples while in regular atmosphere (RA) and controlled atmosphere (CA) storage.  The technology used to measure mass loss was an electronic hanging scale.            We examined large and small 'Red and Golden Delicious' apples.  Apples were weighed about every two weeks by hand as in Link et al. (2003).  Bags of apples were hung from a scale for electronic acquisition of mass.  Approximately every two weeks apples were destructively tested for F, SSC, and TA.  An error analysis of the hanging scales was performed to determine sensor stability.

Methods

The experiment was conducted in the USDA-TFRL controlled atmosphere storage laboratory in Wenatchee, Washington.  Apples were tested under RA storage conditions (95 % relative humidity and about 1-2 °C) and controlled atmosphere (CA) storage (about 95% relative humidity, 1-2 °C).  Apples were brought into storage conditions on November 7, 2002 and remained in storage for 119 days for the RA treatment and for 181 days in CA treatment. 
Large (80 apples box-1) and small (125 apples box-1) 'Golden Delicious' and 'Delicious' apples were used.  Apples were randomly chosen from each population for all experiments.   
         In the RA experiment, apples for repeated mass measurements and destructive testing were stored in covered cardboard fruit boxes with apples layered in ‘egg crate’ cardboard holders.  Golden Delicious apples were stored in plastic lined cardboard boxes and ‘Delicious’ apples were stored in unlined boxes.  In the CA experiment, apples were stored in layered egg crate cardboard holders with the top layer uncovered.
         Eight apples of each population in both RA and CA storage were repeatedly weighed about every two weeks on a manual scale as in Link et al. (2003).  Percent mass loss was computed at each observation time starting at 0% on the first observation date.  The relation between percent mass loss m) and storage time was determined using linear regression.
         Firmness was measured using a TA-XT2 Texture Analyzer (Texture Technologies, Scarsdale, NY) with an 11.1 mm probe.  Eight apples of each population both RA and CA were sampled about every two weeks after Link et al. (2003).  Firmness was measured in two locations on each apple avoiding the side exposed to the sun.  The mean of these two values was used as the F value of each apple.  The relation between storage time and F was determined using a linear regression.
         Eight apples of each population in both RA and CA storage were sampled about every two weeks for SSC and TA after Drake et al. (1981) and Drake (1993).  The eight apples were composited for a single SSC and TA analysis.  The relation between storage time and SSC or TA was determined using linear regression.
         Bags of apples were hung (Fig. 1) from Omega Inc. (Stamford, CT) Minibeam Load Cells (LCEB-5).  Bags were plastic mesh.  An additional was sensor used to determine if the sensor itself is stable over the course of the experiment.  This was done using a known and static mass.

 


 

Figure 1.  Golden Delicious apples hanging from an electronic load cell.

 

         Sensors were connected to two Campbell Scientific CR10X data loggers that collected data every minute to compute a 10-minute average.  Millivolt signals are converted to grams after calibration.
         Each sensor was calibrated at the beginning of the experiment by hanging bagged (bag mass = 12.5 g) calibration masses from a hook (10 g) on the underside of the load cell.  The load cells were re-calibrated at the end of the experiment.  The calibration relation is given by:

 

Mass (g) = b0 + b1 • mV,                                                                                               (1)

 

where b0 is the intercept value (g) when the millivolt signal (mV) is zero and b1 is the slope.
Sensor drift was assessed by comparing initial and final values of b0 and b1.
         Mean percent mass loss from repeatedly weighed ( r), and percent mass loss from bagged apples (mb) were used to predict F, SSC, and TA over time.  The relationship between mass loss and F, SSC, and TA is linear.
         Observations for the analyses were initiated on December 6 after it was noted that the dead weight sensor drifted significantly for the first three weeks of the storage period starting on November 7 (Fig. 2).  After December 6 (day 30 in figure 2), we assumed the sensors were nearly stable.
         Analyses were done using JMP version 5, software (SAS Institute, 2002).  Analysis of variance (ANOVA) and Student’s t-test were used to compare treatments and means.  Error terms are one standard error of the mean.  Statistical significance is set at the  = 0.05 level.

 


Figure 2.  Drift in the response of a hanging scale using a constant dead weight.  Values are daily means.

Results

Percent mass loss of individual apples was dependent on cultivar and time, but not size in RA and CA storage (Table 1).  Apples in RA storage lost mass faster than apples in CA storage (Fig. 3).  All regression relationships (Table 2) in figure 2 were significant (p < 0.0001).  Percent mass loss increased linearly with time for bagged apples of both sizes (Figs. 4 and 5).  Bagged apples of both sizes in CA storage lost water at a greater rate than in RA storage (Figs. 4 and 5).
         Of the Delicious apples, only large apples in RA storage lost firmness during the experiment (Table 3).  Small Delicious apples did not lose firmness in RA or CA storage.  Large red delicious apples in CA storage did not lose firmness.  All golden delicious apples lost firmness except small apples in CA storage.  Only large red delicious apples in CA storage lost a significant amount of soluble solids.  Titratable acids were only lost in golden delicious apples (Table 3).
         Apple firmness, SSC, and TA were influenced by cultivar, size, storage atmosphere, and % mass loss (Table 4).  Atmospheric storage conditions had no effect on SSC.
All linear regression relationships between % mass loss (Table 5) or % mass loss of bagged apples (Table 6) and F, SSC, and TA held the same significance pattern as in Table 4.  The use of electronic hanging scales to determine % mass loss of bagged apples for the prediction of F, SSC, and TA is demonstrated in Figs. 6-8.
         The sensors drifted during the experiment.  The intercept (b0) of the calibration relationship changed an average of 13.43 ± 4.61g.  The difference between initial and final calibration values of b0 range from 0.65 to 39.36 g.  Slope estimates (b1) did not change (b1 = -0.1875 ± 0.2728).  The average percent mass loss (apples weighed individually on a platform scale at the beginning and at the end of the experiment) of bagged apples in the four CA treatments was 4.77%, which is an average loss of 84.16 g from the average initial mass of 1764.45g.  The average drift of 13.43 g is an error of about 16% (10013.43/84.16) from the true value.

Fig.  3.  Relationships between mean percent mass loss and storage time.  See table 2 for the equations.  Error bars are one standard error of the mean. 

 



Fig.  4.  Percent mass loss of bagged apples, size 80, during storage in regular (RA) and controlled atmosphere (CA) conditions.

Fig.  5. Percent mass loss of bagged apples, size 125, during storage in regular (RA) and controlled atmosphere (CA) conditions
 
Fig. 6.  Relation between firmness and % mass loss of bagged apples for size 80 Golden Delicious apples in CA storage.  See table 6 for the equation.
 Fig. 7.  Relation between soluble solids concentration and % mass loss of bagged apples for size 80 Delicious apples in RA storage.  See table 6 for the equation.

Table 1.  ANOVA tests on effects of cultivar, size, and storage time on % mass loss in RA and CA storage.

RA storage

Source
Nparm
DF
Sum of Squares
F Ratio
Prob > F
Cultivar
1
1
0.547232
26.0333
<.0001
Size
1
1
0.074209
3.5303
0.0618
Time
1
1
39.227011
1866.134
<.0001

CA storage

Source
Nparm
DF
Sum of Squares
F Ratio
Prob > F
Cultivar
1
1
0.846102
25.1951
<.0001
Size
1
1
0.030344
0.9036
0.3426
Time
1
1
57.192516
1703.071
<.0001
 Table 2.  Regression relationships between storage time and % mass loss.  Error terms are one standard error of the parameter estimate. 
Cultivar
Atmosphere
% mass loss = B0 + B1 storage time
B0 ± se
B1 ± se
Red
RA
0.0010 ± 0.0294
0.0177 ± 0.00060
Red
CA
0.0902 ± 0.0274
0.0076 ± 0.00031
Golden
RA
0.0267 ± 0.0182
0.0144 ± 0.00037
Golden
CA
0.0349 ± 0.0227

0.0097 ± 0.00026

 

Table 3.  Firmness, soluble solids concentration, and titratable acids as a linear function of storage time where B0 is the intercept, B1 is the slope, and P is significance.
Cultivar
Size
Atmo-
sphere
Firmness (N)
Soluble solids concentration (%)
Titratable acids (%)
B0
B1
P
B0
B1
P
B0
B1
P

Red

80
RA
48.7
-0.167
<0.0001
14.9
-0.0159
NS
0.257
0.00049
NS

Red

80
CA
50.5
-0.0302
NS
14.7
-0.0060
0.038
0.254
-0.00018
NS

Red

125
RA
44.5
-0.0269
NS
14.0
-0.0114
NS
0.250
-0.00029
NS

Red

125
CA
52.1
-0.0208
NS
13.3
0.0014
NS
0.242
-0.00009
NS

Golden

80
RA
45.7
-0.0887
<0.0001
14.8
-0.0170
NS
0.402
-0.00110
NS

Golden

80
CA
46.3
-0.022
0.0143
14.7
-0.0011
NS
0.424
-0.00048
0.001

Golden

125
RA
53.1
-0.1075
<0.0002
15.6
-0.0171
NS
0.321
-0.00087
0.014

Golden

125
CA
49.2
0.0042
NS
14.1
0.00674
NS
0.345
-0.00045
0.023

 

 

 

Table 4.  ANOVA tests on effects of cultivar, size, atmosphere, and % mass loss on firmness, soluble solids concentration, and titratable acidity.

 

Firmness

Source
Nparm
DF
Sum of Squares
F Ratio
Prob > F
Cultivar
1
1
5.0282
0.0761
0.7827
Size
1
1
1786.9692
27.0531
<.0001
Atm
1
1
2255.9320
34.1527
<.0001
AtmCult
1
1
541.5257
8.1982
0.0044
% mass loss
1
1
1540.5217
23.3221
<.0001
Soluble solids concentration
Source
Nparm
DF
Sum of Squares
F Ratio
Prob > F
Cultivar
1
1
8.4182022
25.9784
<.0001
Size
1
1
1.1508528
3.5515
0.0645
Atm
1
1
0.0021866
0.0067
0.9348
SizeCult
1
1
4.7883797
14.7769
0.0003
% mass loss
1
1
1.4589258
4.5022
0.0381
Titratable acidity
Source
Nparm
DF
Sum of Squares
F Ratio
Prob > F
Cultivar
1
1
0.14486539
343.5165
<.0001
Size
1
1
0.02406514
57.0652
<.0001
Atm
1
1
0.00360057
8.5380
0.0050
AtmCult
1
1
0.00468093
11.0998
0.0015
SizeCult
1
1
0.02554141
60.5659
<.0001
% mass loss
1
1
0.01759548
41.7238
<.0001
           
Table 5.  Firmness, soluble solids concentration, and titratable acids as a linear function of % mass loss where B0 is the intercept, B1 is the slope, and P is significance.
Cultivar
Size
Atmosphere
Firmness (N)
Soluble solids concentration (%)
Titratable acids (%)
B0
B1
P
B0
B1
P
B0
B1
P

Red

80
RA
48.6
-9.23
<0.0001
14.9
-0.878
NS
0.257
-0.0274
NS

Red

80
CA
51.2
-5.12
NS
14.8
-0.961
0.026
0.256
-0.0268
NS

Red

125
RA
44.7
-1.77
NS
14.0
-0.648
NS
0.251
-0.0172
NS

Red

125
CA
52.7
-2.95
NS
13.2
0.202
NS
0.242
-0.0086
NS

Golden

80
RA
45.9
-5.80
<0.0001
14.9
-1.14
NS
0.404
-0.0717
NS

Golden

80
CA
46.4
-2.17
  0.007
14.7
- 0.152
NS
0.424
-0.0429
0.001

Golden

125
RA
53.4
-8.11
<0.0001
15.6
-1.26
NS
0.324
-0.0651
0.012

Golden

125
CA
49.3
0.360
NS
14.1
0.764
NS
0.350
-0.0553
0.017

 

Table 6.  Firmness, soluble solids concentration, and titratable acids as a linear function of % mass loss of bagged apples where B0 is the intercept, B1 is the slope, and P is significance.

 

Cultivar
Size
Atmosphere
Firmness (N)
Soluble solids concentration (%)
Titratable acids (%)
B0
B1
P
B0
B1
P
B0
B1
P

Red

80
RA
49.0
-7.829
0.0221
15.0
-0.7327
NS
0.248
0.0023
NS

Red

80
CA
50.9
-1.391
NS
15.0
-0.4187
0.0031
0.252
-0.0051
NS

Red

125
RA
44.5
-1.548
NS
13.9
-0.6081
NS
0.251
-0.0172
NS

Red

125
CA
52.2
-0.966
NS
13.3
0.0636
NS
0.242
-0.0038
NS

Golden

80
RA
46.0
-3.973
<0.0001
14.9
-0.7864
NS
0.406
-0.0488
NS

Golden

80
CA
46.3
-0.773
0.0149
14.7
-0.0403
NS
0.424
-0.0166
0.001

Golden

125
RA
53.4
-5.966
<0.0001
15.7
-0.9299
NS
0.324
-0.0480
0.011

Golden

125
CA
49.2
0.164
NS
14.1
0.255
NS
0.345
-0.0172
0.023

Fig.  8.  Relation between titratable acidity and % mass loss of bagged apples for size 80 Golden Delicious apples in CA storage.  See table 6 for the equation.

Discussion

The main finding of this study is that it is possible to electronically track mass loss for the prediction of apple quality [firmness (F), soluble solids concentration (SSC), titratable acidity (TA)] while in RA and controlled atmosphere (CA) storage conditions.
         The rate of mass loss in our experiments, when individual apples are repeatedly weighed, was similar to that reported in Link and Drake (2002) and in Link et al. (2003).  In our previous work, 'Delicious' apples lost mass at a rate of 0.6% per month in RA storage.  In the current experiment, Delicious apples lost mass at a rate of 0.53% per month and Golden Delicious lost 0.43% per month.  In our previous work, Delicious lost about 0.4 % per month in CA storage and the loss rate was independent of size.  In the current work, the loss rate was 0.23% per month in CA storage, which was, also independent of size.  Golden Delicious apples lost mass in RA storage at a rate of 0.43% per month and 0.29% per month in CA storage.  Loss rates were independent of size in Golden Delicious apples. There was no effect of fruit size on percent mass loss of individual apples suggesting that, under the conditions of the test, the use of mass loss as a predictor of F loss can be done independent of apple size.  This needs to be confirmed using replicated populations of apples on scales.
         The rate of mass loss for individual apples was lower than for apples in bags on the hanging scales.  On the scales in RA storage, Delicious apples lost mass at a rate of 0.57% per month and Golden Delicious lost at a rate of 0.59% per month.  On the scales in CA storage, Delicious apples lost mass at a rate of 0.73% per month and Golden Delicious lost at a rate of 0.81% per month.  The mass loss rate of apples in bags in CA storage is higher than apples in RA and much greater than the individual apples stored in CA.  A possible reason for the greater rate of mass loss for bagged apples in CA storage compared with bagged apples in RA storage is potentially lower relative humidity in the CA storage chambers.  The greatest difference in mass loss rates was between bagged apples and individual apples in CA storage.  A possible reason is the amount of apples stored in the chambers.  The bagged apples (8 large and 12 small in a bag) were much fewer in number than in chambers that contained apples for destructive analyses and for repeated mass measurement.  Controlled atmosphere chambers with many apples are, thus, at higher humidity than chambers with fewer apples.  Higher humidity will reduce the rate of mass loss for the individual apples used for repeated measures.     
         The loss of F, SSC, and TA during the experiment was not observed in all treatments.  When F loss was significant, apples stored in RA lost F faster than apples in CA storage.  Golden Delicious apples lost F for all treatments except small apples in CA storage while only large Delicious apples in RA lost F.  Link and Drake (2002) observed a significant loss of F in Delicious apples in CA storage.  The differing results may be attribute to the initial F values.  Delicious apples in Link and Drake (2002) started at about 72 N while apples in the current experiment started at about 51 N.  Apples at high pressure may lose F faster than apples that are starting out at lower F values.  Loss of SSC was significant in only large Delicious apples in CA storage.  Golden Delicious apples did not lose SSC during the experiment, which confirms a similar finding for Golden Delicious (Drake et al. 1981).  There was no loss of TA in Delicious apples in RA (81 days) or CA (152 days) storage.  This finding is similar to Drake (1993) over 60 days of storage.  In contrast, Ebel et al (1993) observed loss of TA in Delicious apples in RA storage over a longer storage period of 150 days.  While there was little loss of TA in Delicious apples, Golden Delicious apples lost significant amounts of TA in RA and CA treatments.  Drake et al. (1981) also noted significant losses of TA in Golden Delicious apples over 5 months of RA storage. 
         All apples lost mass during the experiment, but not all treatments resulted in reductions in F, SSC, and TA.  When apples lost F, SSC, or TA it was possible to predict the loss by electronically tracking the loss of mass of bagged apples.  Thus, we have largely met our study objective.  The load cells did drift though, and is a cause for concern.  The average drift in b0 resulted in an error of 16% from the true value of mass loss, which results in an error of 0.6 N at the end of the experiment for large Golden Delicious apples in CA storage (Table 6).  The calibration drift of the sensors was also highly variable with one sensor drifting 53% (10039.36/84.16) from the true value.  This would result in an error of 1.7 N for large Golden Delicious apples in CA storage. The maximum error of 53% (1.7 N) out a total change of 3.7 N over the course of the experiment for large Golden Delicious apples in CA storage would lead to significant error in determining when to remove apples from storage.  The sensors also took weeks (Fig. 1) to come to steady state, which would also restrict confidence in their use in an industrial setting.           
         There is a simple solution for the drift problem and is the subject of the next proposed effort.  Given that the load cells only drift in the offset (b0) it will be possible to largely eliminate the drift errors by recalibrating the sensor before each mass reading.  This can be done using a platform scale where apples in a box are automatically lifted off the scale using air jacks.  The scale can then be zeroed.

 

References

De Belie, N., Schotte, S., Coucke, P., De Baerdemaeker, J., 2000.  Development of an automated monitoring device to quantify changes in firmness of apples during storage.  Postharvest Biol.  Technol. 18, 1-8.
Drake, S.R., Proebsting, E.L., Mahan, M.O., Thompson, J.B., 1981.  Influence of trickle and sprinkle irrigation on 'Golden Delicious' apple quality.  J. Amer. Soc. Hort. Sci. 106, 255-258.
Drake, S.R., 1993.  Short-term controlled atmosphere storage improved quality of several apple cultivars.  J. Am. Soc. Hortic.  Sci. 118, 486-489.
Ebel, R.C., Proebsting, E.I., Patterson, M.E., 1993.  Regulated deficit irrigation may alter apple maturity, quality, and storage life.  HortScience 28, 141-143.
Hohn, E., 1990.  Quality criteria of apples.  Acta Hortic.  285, 111-118.
Link, S. O. and S. R. Drake.  2002.  Final report, “A new sensor for tracking apple firmness in CA storage.”

Link, S.O., Drake, S.R. Thiede, M.E., 2003.  The prediction of firmness from mass loss and shrinkage in apples.  Journal of Food Quality (in press)

Magness, J.R., Taylor, G.F., 1925.  An improved type of pressure tester for the determination of fruit maturity.  U.S. Dept. Agric. Circular No. 350, 8.

WA Tree Fruit Research Commission , 1719 Springwater Ave, Wenatchee WA 98801, 509-665-8271, Contact Us

The WTFRC Research Reports database is hosted on the servers of WSU-Tree Fruit Research & Extension Center