Postharvest Biology and Technology 154 (2019) 148–158

Contents lists available at ScienceDirect

Postharvest Biology and Technology

journal homepage: www.elsevier.com/locate/postharvbio

Modelling the biochemical and sensory changes of strawberries during T

storage under diverse relative humidity conditions
⁎
Anastasia Ktenioudakia, , Colm P. O’Donnella, M. Cecilia do Nascimento Nunesb
a
UCD School of Biosystems and Food Engineering, University College Dublin, Belfield, Dublin 4, Ireland
b
Food Quality Laboratory, Department of Cell Biology, Microbiology and Molecular Biology, University of South Florida, 4202 E. Fowler Ave., Tampa, FL, 33620, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Many studies stress the importance of keeping strawberries at high relative humidity conditions during post-
Strawberry harvest storage. However, the effect of deviations occurring across the supply chain on the appearance, ac-
Shelf life ceptability and biochemical properties of strawberries has not been adequately explored or quantified to date
Kinetics using kinetic modelling applications. This study investigated the effect of relative humidity (RH) on degradation
Weibull model
kinetics of quality and biochemical properties of ‘Strawberry Festival’, during 7 days of storage at 2 °C, using
Postharvest storage
Relative humidity
zero, first-order and Weibull models. The strawberries were stored at 40, 60, 70, 80 or 90% RH and were
evaluated using subjective quality evaluation, weight loss monitoring and biochemical analysis. The shelf life
was established based on current industry practices using subjective quality evaluation, namely shrivelling and
colour scores. The Weibull model was found to better fit the experimental chemical analysis data compared to
zero and first order kinetics models. The analysis of the rate constants quantified the significant effect of RH
conditions on the weight loss and degradation rate of chemical components. Storage at low RH conditions
accelerated the loss of ascorbic acid, and anthocyanins and negatively affect the in vitro antioxidant activity. The
overall appearance of strawberries was modelled with zero-order kinetic model and the results revealed that
lower RH conditions can limit the remaining shelf life of fresh strawberries by increasing the rate of appearance
deterioration. Using RH and time as predictors in a logistic regression model, the waste occurring due to un-
acceptable strawberry quality, was predicted; highlighting the importance of using RH in predictive modelling
when designing supply chains with the view to minimise losses.

1. Introduction Strawberry is a popular fruit around the world due its physical
characteristics and its sweet flavour; but is also considered an important
Fresh fruits and vegetables are among the most perishable food source of bioactive compounds such as vitamin C and diverse phenolic
products. Due to their high perishability, they are also among the most compounds such as phenolic acids, flavonoids and tannins (Cayo et al.,
frequently wasted foods (Kelly et al., 2018). Food waste has a devasting 2016; Kelly et al., 2016a, 2016b; Flores-Felix et al., 2018) for human
impact on food security affecting all people regardless of geography or nutrition. At the same time, strawberries are one of the fruits most often
social groupings. With a predicted increase of 1.7 billion in world po- discarded throughout the supply chain, due to their short shelf life and
pulation between now and 2050, mankind is placing more and more high perishability and due to poor management during distribution
pressure on the shrinking finite resources used to produce our food (Mc (Kelly et al., 2018; Nunes et al., 2009; Pan et al., 2014). As the straw-
Carthy et al., 2018). Yet it is estimated that globally 30–50% of all food berry fruit is metabolically active after harvest, the rate at which its
produced never reaches its intended destination, i.e. the consumer. quality will deteriorate depends on both genetic factors and on pre- and
Under-addressed supply chain issues are a key cause of commercial post-harvest conditions (Castro et al., 2002). The recommended storage
food waste (White, 2007). Emphasis has been placed over the years on conditions for strawberries are 0 °C and 90–95 % Relative Humidity
temperature management that often deviates from optimum during (RH). However, Nunes et al. (2009) found that during retail storage
storage, handling and distribution, as one of the major causes of fresh strawberries are kept at temperatures ranging from 5.8 °C to 7.1 °C and
fruit and vegetable overall quality deterioration (van Boekel, 2008, RH conditions ranging from 56% to 93%. This range does not consider
Taoukis et al., 1997; Nunes et al., 2003; Lai et al., 2011 the common practice of storing and displaying strawberries outside

⁎
Corresponding author.
E-mail address: anastasia.ktenioudaki@ucd.ie (A. Ktenioudaki).

https://doi.org/10.1016/j.postharvbio.2019.04.023
Received 29 January 2019; Received in revised form 26 April 2019; Accepted 26 April 2019
Available online 13 May 2019
0925-5214/ © 2019 Elsevier B.V. All rights reserved.
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

refrigerated compartments or displays at open markets where tem- One the day of harvest (day 0) three replicates of 10 fruit each, were
peratures and RH conditions can deviate at a larger scale from the used for sensory quality evaluation (30 fruits in total) and weight loss (3
optimum conditions. Furthermore, Lai et al. (2011) studied real clamshells of 10 fruits each). Three replicates of 10 fruit each were
strawberry supply chain conditions and their effect on quality and immediately frozen to preserve their biochemical properties and used
found that RH varies from 33.8% to 87.2% along the different supply later for chemical analysis.
chain stages (harvest to retail). Many studies stress the importance of
keeping strawberries at 90–95 % RH but the effect of the deviations 2.2. Sensory quality evaluation
occurring during the supply chain on the appearance, acceptability and
nutritional value of strawberries has not been adequately explored to Sensory quality evaluation was performed every day during a 7-day
date. Shin et al. (2007) studied the effect of three RH conditions (75%, storage period, always by the same trained personnel. Colour, shrivel-
85%, and 95%) and the effect of temperature (0.5 °C, 10 °C, and 20 °C) ling, decay severity, firmness, taste, and aroma were determined sub-
on the quality and nutritional properties of strawberries and found no jectively using 1–5 rating scales (where 1 = very poor, 3 = acceptable
effect of RH on these properties. In a later study, Shin et al. (2008) and 5 = excellent) (Nunes, 2015; Nunes and Delgado, 2014). A score of
looked at a lower RH condition (65%) and found that weight loss in- 3 was considered as the lower limit of acceptability for sale.
creased, whereas total flavonoids and total phenols decreased in fruit
stored at 65% compared to samples stored at 95% RH. Both studies 2.3. Weight loss
evaluated their nutritional analysis results in fresh weight basis, which
could fail to consider the concentration effect due to water loss and Weight loss was calculated from the initial weight of each of the
hence obscure the effect of RH on these properties. three clamshells containing 15 fruits each, and every day during a
To better define the effect of storage conditions on deterioration of seven-day storage period. The following formula was used for water
quality, the kinetics of sensory and chemical properties should be ex- loss corrections: [chemical components (fresh weight) × 100 g/(dry
amined. Traditionally, degradation of nutrients and quality character- weight) + weight loss during storage (g)]. The dry weight was de-
istics are modelled by 0, 1st or 2nd order kinetic models (Amodio et al., termined by drying three weighed aliquots of homogenized fruit tissue
2013; Oms-Oliu et al., 2009). Alsostudies have emerged that used the at 80 °C, until weight stabilized. All chemical components were ex-
Weibull model, the cumulative form of the Weibull distribution func- pressed on a dry weight basis to compensate for water loss during
tion, to study several degradation reactions (Ong et al., 2011; storage.
Odriozola-Serrano et al., 2009; Oms-Oliu et al., 2009; Amodio et al.,
2013). The Weibull model is flexible due to the inclusion of a shape 2.4. Biochemical analysis
constant in addition to the rate constant (Oms-Oliu et al., 2009).
Odriozola-Serrano et al. (2009) studied the degradation kinetics an- Three replicated samples of fifteen individual fruit per treatment
thocyanins, total phenols and ascorbic acid in fresh cut strawberries were homogenized in a laboratory blender at high speed for 120 s and
using the Weibull model and used it to explain the changes occurring in the resulting puree immediately frozen and kept at -30 °C until used.
anthocyanins and antioxidant capacity of the fruits under storage. The samples were evaluated on days 0, 2, 3, 4, 5, 6, and 7. The fol-
The present study focuses on a broad range of RH conditions (40% - lowing analyses were performed:
90%) that can be encountered along the supply chain of strawberries.
The main objective is to examine the effect of the different RH condi- 2.4.1. Total ascorbic acid
tions during storage on the shelf life, sensory and nutritional quality of Total ascorbic acid (AA) analysis was conducted using a Hitachi
strawberries; emphasising on properly describing the changes occurring LaChromUltra UHPLC system with a diode array detector and a
in both physical and chemical properties of the fruit. This paper aims: i) LaChromUltra C18 2 μm column (2 × 50 mm) (Hitachi, Ltd., Tokyo,
to examine the effect of RH conditions on the shelf life of strawberries Japan), as described in Kelly et al. (2016b). The retention time of AA
including both the sensory and nutritional quality, ii) to study the ki- peak was 0.35 min. After comparison of retention time with the AA
netics of sensory and chemical changes occurring in strawberries during standard, the peak was identified. The amount of total AA in strawberry
storage by comparing three kinetic models; and iii) to examine and was quantified using calibration curves obtained from different con-
predict through modelling the waste that would occur depending on the centrations of AA standards. The total ascorbic acid content was ex-
storage conditions. pressed in g kg−1 on a dry weight basis.

2. Materials & methods 2.4.2. Total anthocyanin content

Total anthocyanins (ANC) were extracted in 0.5% (v/v) HCl me-
2.1. Plant material and experimental setup thanol and measured using the procedure described by Nunes et al.
(2005). Total anthocyanin content (ANC) is expressed as g kg−1 pe-
‘Strawberry Festival’ strawberries were harvested twice during 2012 largonidin 3-glucoside equivalents on dry weight basis.
and2013 (March 2012 and February 2013) from a commercial field in
Florida, USA. The samples were brought to the USF-Food Quality 2.4.3. Total phenolic content
Laboratory in Tampa, Florida, USA, with minimal delay after harvest In vitro antioxidant activity of the samples was determined using
(Max delay of 30 min). The fruit were then sorted by colour, size, and the total phenols assay by Folin–Ciocalteau reagent as described by
absence of defects and packedinto clamshells (capacity ≈ 0.453 kg). In Nunes et al. (2005). The total phenolic content (TPC) of the samples
total, 210 clamshells (containing 10 fruit each) were prepared and was measured and expressed as g kg−1 gallic acid equivalents (GAE) on
distributed equally to five RH-controlled chambers (Forma dry weight basis.
Environmental Chambers Model 3940 Series, Thermo Electron
Corporation, OH, USA) at 2 °C and 40, 60, 70, 80, and 90% RH. Three 2.4.4. Sugar analysis
clamshells were removed daily from each chamber and used for sensory Quantification of sucrose, fructose and glucose was conducted using
quality evaluation and weigh loss monitoring. Three additional clam- a Hitachi HPLC with a refractive index detector and a 300 mm × 8 mm
shells were used for chemical analysis. Temperature and RH were Shodex SP0810 column (Shodex, Colorado Springs, CO) with an SP-G
monitored throughout storage using battery-powered data loggers guard column (2 mm x 4 mm) as described by Kelly et al. (2016b). Total
(Hobo® U10 Temp/RH data logger, Onset Computer Corporation, sugars were calculated as the sum of the individual sugars (sucrose and
Pocasset, MA, USA). glucose) values and expressed as g kg−1 on a dry weight basis.

149
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

2.4.5. Titratable acidity and soluble solids content 2.6. Statistical and data analysis
Titratable acidity (TA) and soluble solids content (SSC) were de-
termined according to (Nunes et al., 2006a) and expressed in g kg−1 on One-way ANOVA analysis was carried out using Minitab (Minitab
a dry weight basis. version 16, Minitab Inc., USA). Where ANOVA indicated significant
differences were present, the Tukey pairwise comparison of the means
2.5. Mathematical modelling (and 95% confidence intervals) was used to identify where the sample
differences occurred. Significant differences were identified between
2.5.1. Kinetics modelling the two harvests and the results were treated separately. The kinetic
Three kinetic models were evaluated to relate and explain the models were fitted to the data using non-linear regression procedures in
changes in appearance (colour and shrivelling) and the changes in R version 3.4.4 (R Core Team, 2017). The following packages were used
chemical properties that occurred to the strawberries during storage to perform the data analysis in R: Caret (Kuhn et al., 2018), e1071
under different RH. Zero (Eq. (1)) and first order (Eq. (2)) kinetics (Meyer et al., 2017), caTools (Tuszynski, 2014), ElemStatLearn
models have been applied to describe degradation reactions in foods (Halvorsen, 2015), ROCR (Sing et al., 2005), Metrics (Hamner and
and the Weibull model (Eq. (3)) has been shown to fit the degradation Frasco, 2017), writexl (Ooms, 2018), class (Venables and Ripley, 2002),
of chemical compounds in strawberries (Odriozola-Serrano et al., and rpart (Therneau and Atkinson, 2018).
2009), sensory and ascorbic acid changes in melon (Amodio et al.,
2013), and also colour changes in dried salak fruit (Ong et al., 2011). 3. Results and discussion

Q (t ) = Q0 − kt (1) 3.1. Effect of RH and time on weight loss

Q (t ) = Q0 e−kt (2) Weight loss is an important parameter in determining the shelf life
Q(t) is the value/concentration of the quality index at time t, Q0 is of fresh fruits and vegetables, because it is responsible not only for
the intercept of the curve, and k is the rate constant. visual appearance deterioration (discolouration, shrivelling etc) but
The cumulative distribution of the Weibull distribution function also for objectionable changes in texture, flavour and nutritional value
may be re-written as follows (Amodio et al., 2013): (Nunes and Emond, 2007; Nunes et al., 1998). Robinson et al. (1975)
reported previously that the maximum acceptable weight loss in
t strawberries is 6%, before they become unacceptable for sale. A more
Q (t ) = Q0 e [−( ) β]
a (3) recent study (Nunes and Emond, 2007) found that a loss of 2.5%-3% in
Where Q(t) is the value/concentration of the quality index at time t, weight of strawberries was the maximum acceptable loss when kept at
Q0 is value/concentration of the quality index at initial time, and a and 20 °C and occurred within 2–3 days. In our study, weight loss increased
β are the scale factor and the shape parameter, respectively. The inverse with storage time and it was dependant on the RH conditions during
of a (1/a) may be considered as the kinetic constant of the process storage. The total weight loss from harvest to day 7 of storage, varied
(Corradini and Peleg, 2004). The β parameter indicates concavity or from 7.2% (90% RH) to 19% (40% RH) in 2012; and from 5.7% (90%
convexity of the curve when it takes values below and above 1 re- RH) to 15.5% (40% RH) in 2013 (Fig. 1).
spectively (Odriozola-Serrano et al., 2009). The constants a and β were Both the zero-order model and the Weibull model fit the data. The
estimated and the initial value of the samples was used for Q0. Weibull model was chosen to interpret the changes in weight loss be-
Statistical parameters such as root mean square error (RMSE), cause of the highest coefficient of determination (R2) and lower RMSE
Akaike information criterion (AIC), and the regression coefficient (R2) and AIC in both harvests. Table 1 shows the rate constants and the
were the criteria for goodness of fit of the models. The models with the goodness of fit of the Weibull model for the weight loss data. Straw-
highest R2 and lowest RMSE and AIC were chosen as the best fit for the berries kept at lower RH conditions, were found to have higher rate
data. constants; proving that weight loss occurred faster when fruit was kept
at 40, 60 and 70% RH compared to 80 and 90%. Water vapor pressure
deficit (VPD) will increase at any given temperature as RH of air de-
2.5.2. Classification models
creases (Hosahalli, 2015). Exposure of fruits and vegetables to condi-
The average of the shriveling and the colour scores, was used to
tions that lead to an increase in the VPD, will increase the transpiration
evaluate the acceptability of the strawberries. A score of 3 was the
rate and can result in increased water loss (Laurin et al., 2005). In 2012,
lower limit of acceptability for sale. Samples were divided in two ca-
higher rates of water loss were observed than in 2013 and hence the
tegories (Pass/Fail) depending on that score (score > 3 = Pass,
total weight loss in 2012 was overall higher in all storage RH. The shape
score≤3 = Fail). Data from both harvests were merged and the fol-
factor γ is less than 1 in all cases indicating that water loss is more
lowing classification models were fitted to the experimental data:
intense during the first days of storage.
binary logistic regression model, decision tree, support vector machine,
If we adopt the more conservative approach of 6% being the max-
and k-nearest neighbours. The independent variables were Time (days)
imum acceptable weight loss, it can be observed that shelf life of
and RH (%). The models were evaluated using the k-fold cross valida-
strawberries harvested in 2012 would be limited to 1.5, 2, 3, 4, and 5.5
tion method. The model with the highest average accuracy and the
days for fruit kept at 40, 60, 70, 80 and 90% respectively (Fig. 1).
lowest standard deviation of the accuracies between the folds was the
Keeping strawberries nearer optimum RH conditions (i.e., 80–90 %) can
binary logistic regression model. Therefore, it was considered the best
extend the shelf life by 1–4 days compared to fruit kept to sub-optimum
fit for our data (see Results and Discussion section).
RH conditions (40–60 %).
The general formula for the binary logistic regression model can be
written as in Eq. (4):
3.2. Effect of RH and time on sensory quality
p ⎞
Logit (p) = ln ⎛⎜ ⎟ = β
0 + β1 * RH + β2 * Time (days ) Colour became less acceptable (darker and less vibrant) with time
⎝1 − p ⎠ (4)
and significant differences were observed between day 0 and day 7 at
Where p is the probability that the sample is a “Pass”, 1-p is the all RHs (results not shown). However, colour was not considered a
probability that a sample is a “Fail”, logit(p) is the odds ratio of the two limiting factor of strawberry shelf life (i.e., rating score was always
probabilities, β are logistic coefficients, RH is the relative humidity, and higher than 3).
Time is the days after harvest. Shrivelling increased with time, and it was more intense for

150
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

Fig. 1. Effect of storage under various RH conditions on the weight loss (%) of fresh strawberries harvested (a) in 2012 and (b) in 2013. Error bars represent the
standard deviation of the mean, plotted lines correspond to the values estimated by the Weibull model.

strawberries kept at 40% and 60% RH (Fig. 2). In 2012, shrivelling was models fitted the data well (Table 2). In most cases the zero-order
more pronounced on the day of harvest (average score of 4.5) than in model had the highest coefficient of determination and the smallest AIC
2013 (average score of 5). Shrivelling increased more rapidly in and it was chosen as the best fit in both harvest years. The rate constant
strawberries stored at 40% RH for both harvests, limiting the shelf life shows that the rate of appearance degradation increased as the storage
to approximately 2.5 days in 2012 and to 4 days in 2013. Strawberries RH decreased. For example, in 2013, the rate of appearance score de-
kept at 60, 70 and 80% RH had a shelf life of approximately 5.5 days in crease, for samples kept at 40% RH, is 0.44 per day, double the rate for
2012 based on shrivelling scores; whereas for fruit kept at 90% RH, samples kept at 90% RH (0.22 /day). This can be linked to the higher
shrivelling was not a limiting shelf life factor. In 2013, shrivelling was rate of weight loss at lower RHs, which affects the discolouration and
not a limiting shelf life factor for strawberries kept at 70, 80 and 90% drying rates and hence the shrivelling and colour scores. One-way
RH; whereas fruit kept at 60% RH received a score below 3 just after ANOVA analysis showed that the score of strawberries kept at 90% RH
day 6. The differences between harvests are due to the initial significant was significantly higher on most days (and on day 7) for both harvests,
differences in shrivelling scores (2012 samples received a lower score than the scores of fruit kept at the lower RHs (40–80%). As a result,
on the day of harvest) and to the higher rate of weight loss observed based on the acceptability scores, shelf life as being evaluated to date in
during storage at all RHs in 2012 (Table 1). Nunes (2015) studied the the industry based on visual appearance will be significantly shorter
correlation between subjective measurements (such as sensory quality when samples are kept at RH conditions lower than 90%.
evaluation) and physicochemical attributes, and found strong correla-
tion between shrivelling and weight loss (R2 = 0.92) for strawberries 3.3. Biochemical analysis
when kept at different RH regimes.
The average score between colour and shrivelling was used as the Significant differences were observed on the initial values of all the
overall appearance score. To better understand the changes in the chemical compounds studied between the two harvests. Most of the
overall appearance, the three kinetic models were fitted to the data. All differences can be attributed to the difference in temperature at the

151
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

Table 1 RH has on degradation kinetics of AA. In 2012, the degradation rate

Kinetic constants and goodness of fit parameters of zero-order, 1st-order and varied between 0.15 (80% RH) to 0.66 (40% RH) per day. These results
Weibull models of weight loss (%) in fresh strawberries kept at various RH show that the retention rate decreased 4 times faster for strawberries
conditions. kept at 40% RH compared to samples kept at 80% RH. The γ parameter
Harvest Model RH Rate constants R2 RMSE AIC was below 1 indicating that AA degraded during the first days of sto-
(per day) rage at all RH conditions. Similar results were observed for strawberries
harvested in 2013. Loss of AA in fresh fruit and vegetables during
2012 M0 40 −2.66 0.98 0.83 65
M0 60 −2.54 0.99 0.59 49
storage has been attributed to various causes, such as tissue degradation
M0 70 −1.83 0.99 0.46 37 as the product becomes overripe, cell wall damage and enzymatic
M0 80 −1.40 0.97 0.58 48 oxidation caused by bruising, but also due to water loss which can in-
M0 90 −1.03 0.99 0.19 −4 tensify AA oxidation (Nunes et al., 1998; Klein, 1987). Many other
M1 40 −0.23 0.90 1.99 107
studies have also reported the loss of AA during storage of strawberries
M1 60 −0.24 0.91 1.84 103
M1 70 −0.25 0.91 1.28 86 (Kelly et al., 2018; Cayo et al., 2016; Kelly et al., 2016a, 2016b;
M1 80 −0.24 0.88 1.14 80 Odriozola-Serrano et al., 2009). Our results show that the rate of AA
M1 90 −0.26 0.93 0.67 55 degradation is linked to water loss and highlight the fast deterioration
MW 40 0.03, γ=-0.37 0.99 0.74 60
than can occur when fruit are stored in suboptimal RH conditions.
MW 60 0.03, γ=-0.38 0.99 0.49 40
MW 70 0.02, γ=-0.35 0.99 0.47 38
MW 80 0.01, γ=-0.32 0.97 0.55 45 3.3.2. Total anthocyanin content
MW 90 0.01, γ=-0.30 0.99 0.22 2 Strawberries harvested in 2012 had significantly higher ANC (4.04 g
2013 M0 40 −2.15 0.99 0.5 41 kg−1) compared to strawberries harvested in 2013 (1.69 g kg−1), pos-
M0 60 −1.80 0.99 0.4 35
sibly due to the higher temperatures in 2012 on the week before har-
M0 70 −1.34 0.99 0.3 17
M0 80 −1.29 0.99 0.2 4 vest. Differences in ANC between years were reported by Cayo et al.
M0 90 −0.87 0.94 0.5 43 (2016), also observing that higher temperatures around time of harvest
M1 40 −0.24 0.91 1.5 93 led to higher accumulation of anthocyanins. Total anthocyanins content
M1 60 −0.24 0.91 1.3 85 decreased with time with significant differences among the different RH
M1 70 −0.25 0.92 0.9 68
M1 80 −0.25 0.92 0.9 67
conditions. At the end of storage (day 7), strawberries stored at 40% RH
M1 90 −0.25 0.85 0.8 64 had the lowest ANC both in 2012 (0.72 g kg−1) and 2013 (0.48 g kg−1).
MW 40 0.022, γ=-0.35 0.99 0.5 41 On the contrary, strawberries stored at 90% RH had nearly double the
MW 60 0.015, γ=-0.33 0.99 0.4 34 concentration of ANC on day 7.
MW 70 0.009, γ=-0.31 0.99 0.4 26
To better demonstrate the effect of RH on the retention of antho-
MW 80 0.008, γ=-0.31 0.99 0.2 6
MW 90 0.004, γ=-0.28 0.94 0.5 41 cyanins, the kinetics models were used on the retention data. The de-
gradation rate constants obtained from the Weibull model, for the dif-
M0 = Zero-order (Eq. (1)), M1 = 1st-order (Eq. (2)), MW = Weibull model (Eq. ferent RH conditions for both harvests are shown in Table 5. Consistent
(3)). with the results of AA, the rate of ANC degradation was lower for
samples kept at higher RH. In 2013, the rate of decrease was lower than
time of harvest observed in 2012 and 2013 (Table 3). To better un- in 2012 for most RHs, leading to the higher retentions observed. In
derstand the effect of RH during storage on chemical properties of 2012, the retention of ANC varied from 18.3% (at 40% RH) to 46.4%
strawberries, the three kinetic models mentioned earlier were applied. (at 80% RH), compared to 28.6% (at 40% RH) to 64.6% (at 80% RH), in
The Weibull model was the most suitable for fitting the changes in the 2013. The differences in the rate constants among the RH conditions
chemical properties described below. The goodness of fit of the model and between the harvests can be explained by the highest rate of weight
was judged based on the coefficient of determination, RMSE and AIC. loss observed under the same conditions. Nunes et al. (2005) have re-
Table 4 shows the results of the goodness of fit for all the models ap- ported a decrease in ANC in strawberries kept uncovered at 1 °C and
plied to the sugar retention data. The Weibull model had the highest 90–95 % RH over storage. The authors explained how water loss in
coefficient of determination, lowest RMSE and lowest AIC. The same combination with oxidative mechanisms can contribute to degradation
was observed for all chemical properties and hence the results for the and changes in the anthocyanin pigments. Water loss is not the only
AA, ANC, TPC, SSC and TA only show the goodness of fit parameters for reason for ANC degradation during storage (other factors such as as-
the Weibull model. corbic acid oxidation, pH changes, changes in sugar content and pre-
sence of other phenolics also contribute to loss of ANC (Kader et al.,
3.3.1. Total ascorbic acid 1999; Holcroft and Kader, 1999); however as Nunes et al. (2005)
Significant differences were observed between the initial con- mentions water loss can make the degradation more prominent. Despite
centrations of AA for the two harvests. The concentration of AA varied the higher degradation rate observed in 2012 the concentration of ANC
from 8.25 g kg−1 in 2012 to 10.92 g kg−1 in 2013. Concentration of AA was still higher at the end of storage than the content measured in
is influenced by environmental factors such as temperature, light in- 2013, due the higher ANC in strawberries on the day of harvest.
tensity, and maturity at harvest (Cayo et al., 2016). Wang and Camp
(2000) found that strawberry cultivars grown at cooler temperatures 3.3.3. Total phenol content
had higher concentration of AA possibly due to decreased metabolism. Significant differences were observed in TPC between the two har-
The higher temperatures observed in 2012 (Table 3) could explain the vests. The total phenolic content varied between 26.92 g kg−1 GAE, in
lower content of ascorbic acid that year. In general, AA decreased with 2012 to 23.72 g kg−1 GAE in 2013. The higher temperatures observed
storage time for all RH conditions. Strawberries kept at higher RHs in 2012, could have assisted in the accumulation of compounds with
showed better retention of AA throughout storage (Fig. 3). For example, antioxidant activity (i.e., anthocyanins) and hence contribute to the
the concentration of AA on day 7 for strawberries kept at 40% RH was higher antioxidant activity of the samples as measured with the TPC
1.68 g kg−1 compared to 3.55 g kg−1 for samples kept at 90% RH in assay. Also the strawberry plants from the two harvests were most
2012. likely exposed to different levels of stress, leading to different levels of
The Weibull model was the most suitable model to describe the bioactive compounds (i.e., polyphenols) to be present as a response to
changes in AA retention during storage (Table 5). The degradation that stress (Kelly et al., 2016b). Total phenols content decreased with
constant decreased as the RH increased, indicating the significant effect time of storage. Strawberries stored at 90% RH had the highest TPC on

152
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

Fig. 2. Changes observed in shrivelling of fresh strawberries during storage under various RH conditions (a-2012 harvest, b-2013 harvest). Error bars represent the
standard deviation of the mean score. Dashed straight line represents the minimum acceptable quality before strawberries become unmarketable (rating 3).

all days of storage in 2013 whereas in 2012, strawberries stored at 80% contact with polyphenols and hence a reduction in TPC is observed
and 90%, both had the highest content. An increase in TPC was ob- (Nunes et al., 2005). This in combination with the degradation of an-
served on days 2 and 3 in samples harvested in 2013 and stored at 90% thocyanins will lead to the decrease in antioxidant activity measured
RH. According to Van de Velde et al. (2017), an increase in TPC at with the TPC assay.
specific periods during storage can attributed to de novo synthesis as a
result to stress. This has also been reported before (Oms-Oliu et al.,
2009). However, such an increase in TPC on those days was not con- 3.3.4. Total sugar content
sistently observed in all strawberry samples. Total sugar content was significantly higher in strawberries har-
The Weibull model was the best fit to describe the retention of total vested in 2013 (1068 g kg−1) than in 2012 (865 g kg−1), most likely
phenols data (Table 5). The degradation constants varied from 0.09 (in due to the lower temperatures at the time of harvest in 2013. Lower
2012) and 0.1 (in 2013) per day for samples stored at 90% RH, to 0.30 temperatures result in lower metabolic rate and higher accumulation of
(in 2012) and 0.26 (in 2013) per day for samples stored at 40% RH soluble carbohydrates and starch (Wang and Camp, 2000). Loss of su-
(Table 5). The constants increased as the RH decreased, indicating that gars in strawberries postharvest has been reported elsewhere (Kelly
the degradation of bioactive compounds is faster at lower RH condi- et al., 2016a, 2016b; Kelly et al., 2018) and the degradation occurs
tions, indicating a possible link to the faster rate of weight loss. Shin regardless of harvest or genotype effects (Cayo et al., 2016). Caleb et al.
et al. (2008) found a decrease in TPC in strawberries after 12 days of (2016) also observed a depletion of sucrose and glucose in strawberries
storage at 10 °C, and observed a simultaneously increase in water loss during storage; highlighting that the role of sucrose, postharvest, is to
and decrease in anthocyanins content. The loss of water leads to sustain the energy demand during respiration and support fruit main-
membrane collapse which allows for enzymes like PPO to come in tenance and metabolic processes. In our study, total sugar content de-
creased with storage time for all RH conditions, with significant

153
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

Table 2 2012 (22.7%) than in 2013 (16.9%). These results cannot be attributed
Kinetic constants and goodness of fit parameters of zero-order, 1st-order and to the weather conditions during harvests, as it has been shown that
Weibull models of the overall appearance of fresh strawberries kept at various higher temperatures during harvest (as observed in 2012) will lead to
RH conditions. lower TA values (Wang and Camp, 2000). These differences could be
Harvest Model RH Rate constants R2 RMSE AIC due to possible differences in maturity at harvest or due to differences
(per day) in the growing conditions between the two harvests. A reduction in TA
during storage was observed for all RH conditions in both years. Con-
2012 M0 40 0.39 0.94 0.23 −17
M0 60 0.24 0.89 0.19 −109
flicting results have been reported in the literature regarding TA during
M0 70 0.23 0.80 0.27 52 storage. A reduction in TA with time was observed by Kelly et al.
M0 80 0.26 0.78 0.31 129 (2016b) and Cayo et al. (2016), whereas an increase in TA was reported
M0 90 0.21 0.88 0.18 −137 during storage by (Nunes et al. (2006a) and Caleb et al. (2016). Shin
M1 40 0.14 0.94 0.22 −30
et al. (2007) found no effect of RH in TA of strawberries over a storage
M1 60 0.07 0.91 0.17 −168
M1 70 0.06 0.80 0.27 52 period of 4 days. The RH conditions in their study varied from 75% to
M1 80 0.07 0.77 0.32 133 95%. Using the Weibull model to fit our data, it was found that higher
M1 90 0.05 0.88 0.18 −145 rates of TA degradation were observed when strawberries were kept at
MW 40 0.16, γ = 0.82 0.92 0.34 162
lower RHs and varied between 0.08 to 0.23 per day (Table 6). The
MW 60 0.06, γ = 0.58 0.90 0.23 −15
MW 70 0.06, γ = 0.67 0.76 0.34 168
reduction in TA observed in our study is possibly due to the use of
MW 80 0.07, γ = 0.80 0.75 0.35 184 organic acids as substrates in respiration (Wang and Camp, 2000); and
MW 90 0.05, γ = 0.78 0.86 0.21 −65 as mentioned earlier a relationship exists between the respiration rate
2013 M0 40 0.44 0.93 0.27 58 and the weight loss rate.
M0 60 0.31 0.85 0.30 114
M0 70 0.26 0.83 0.28 70
M0 80 0.23 0.83 0.24 −3 3.3.6. Soluble solids content
M0 90 0.22 0.78 0.26 44 Soluble solids content (SSC) is considered a good indicator of
M1 40 0.13 0.91 0.31 130 sweetness (Jouquand et al., 2008). The SSC in strawberries harvested in
M1 60 0.08 0.86 0.30 105 2013 was significantly higher than that of strawberries harvested in
M1 70 0.07 0.81 0.29 89
M1 80 0.06 0.83 0.24 3
2012 (1370 g kg−1 and 1300 g kg−1 respectively), possibly due to the
M1 90 0.05 0.78 0.26 49 variation in temperature at the time of harvest observed between the
MW 40 0.14, γ=1.26 0.93 0.28 82 two years. A reduction in SSC during storage was observed for all RH
MW 60 0.09, γ = 0.97 0.85 0.30 107 conditions and the rate of reduction was modelled using the Weibull
MW 70 0.07, γ = 0.99 0.81 0.30 102
model. The rate of decrease in SSC increased as the storage RH de-
MW 80 0.05, γ = 0.77 0.81 0.27 56
MW 90 0.05, γ = 0.91 0.77 0.28 72 creased (Table 6). The same was observed for both harvests and varied
between 0.12 and 0.51 per day depending on the harvest and the RH
M0 = Zero-order (Eq. (1)), M1 = 1st-order (Eq. (2)), MW = Weibull model (Eq. conditions. Higher rates of decrease were observed in 2012 for samples
(3)). stored in 40% and 60% RH due to the higher rate of weight loss also
taking place.
Table 3
Daily mean temperatures for the week of harvest during 2012 and 2013 periods 3.4. Classification model
in Floral City, Florida (data from weather underground https://www.
wunderground.com). The logistic regression model was the best fit to the experimental
Harvest Temperature oC data when compared to other classification models as mentioned ear-
lier. The results of the K-fold validation showed that the logistic re-
Mean Max Min
gression had the highest mean accuracy when compared to other
March 2012 20.5 28.8 13.8 classification models (0.93 ± 0.02). Data from both harvests were
February 2013 12.1 21.6 6.8 merged and the dataset was split to a training set and a test set and
these were used to train and validate the model respectively. The results
of the logistic regression on the test set showed that both RH and time
differences in sugar content on day 7 among strawberries kept at 80 and and their interaction had a significant effect on the score (Pass/Fail) of
90% RH and the lower RH conditions (Fig. 4). the strawberry samples. The confusion matrix produced by the model is
The Weibull model was the best fit for the retention of total sugars shown in Table 7.
data, considering the high coefficient of determination (R2), low RMSE Table 7 shows that the model predicted 488 samples (out of 600) in
and low AIC (Table 4). Based on the rate constants shown in Table 4, the test set would pass and 112 samples would fail based on the ac-
sugar content reduced faster for strawberries kept at 40 and 60% RH in ceptability criterion. The actual number of samples that passed or failed
both harvests. Kader and Salveit (2003) stated that water stress caused was 508 and 92 respectively, resulting in a 96.6% accuracy in predic-
by low storage RH conditions can stimulate the rate of respiration. The tion. The accuracy calculated from the confusion matrix tables can
higher rate of weight loss experienced in these RH conditions, attrib- sometimes be misleading due to what is known as the accuracy paradox
uted to water loss, could potentially have increased the rate of re- (a model with higher accuracy has less predictive power than a model
spiration, consequently increasing the rate of simple sugars depletion. with lower accuracy). To better estimate the accuracy of the model the
In both years the parameter γ was below 1 indicating that degradation Receiving Operating Characteristic (ROC) curve was constructed
takes place the first days of storage. Sweetness intensity has been linked (Fig. 5) and the area under the curve was calculated. A ROC curve is
to consumers’ acceptance of strawberries (Schwieterman et al., 2014), commonly plotted to illustrate the diagnostic ability of a binary clas-
hence a higher rate of decrease in sugar content during storage under sifier. It shows the sensitivity (the proportion of correctly classified
low RH conditions, will become a limiting shelf life factor in straw- positive observations) and specificity (the proportion of correctly clas-
berries at consumer level. sified negative observations) as the output threshold is moved over the
range of all possible values (Robin et al., 2011). A popular way of
3.3.5. Titratable acidity summarizing the discrimination ability of a model is to report the area
Titratable acidity (TA) was significantly higher in fruit harvested in under the ROC curve (AUC) (Prasad, 2016). The AUC value for the

154
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

Table 4 model was 0.97, showing that the logistic regression using the RH and
Kinetic constants and goodness of fit parameters of zero-order, 1st-order and time after harvest as predictors, has a strong discriminating power
Weibull models of sugar retention in fresh strawberries kept at various RH between strawberry samples that would score above or below the ac-
conditions. ceptability threshold.
Harvest Model RH Rate constants R2 RMSE AIC
(per day)
4. Conclusions
2012 M0 40 9.70 0.69 14.55 178
M0 60 9.99 0.72 13.90 176 Strawberries are metabolically active during postharvest storage,
M0 70 10.75 0.86 9.58 160 and their quality and shelf life is influenced among other things by the
M0 80 8.14 0.81 8.71 156
storage conditions they experience. Temperature and RH are environ-
M0 90 8.62 0.76 10.83 166
M1 40 0.31 0.92 7.89 152 mental factors with major impact on strawberries postharvest quality.
M1 60 0.32 0.94 6.88 147 However, the RH effect on strawberries sensory and nutritional prop-
M1 70 0.26 0.96 4.93 133 erties has not been explored sufficiently. This study emphasised on
M1 80 0.16 0.91 6.20 142
studying the kinetics of the physicochemical changes occurring during
M1 90 0.19 0.88 7.73 151
MW 40 0.50, γ = 0.37 1.00 1.04 67 storage under RH conditions experienced by strawberries under simu-
MW 60 0.44, γ = 0.45 1.00 0.97 64 lated commercial situations. The results provided a comprehensive
MW 70 0.28, γ = 0.72 0.98 3.59 119 description and explanation of these changes, information which is
MW 80 0.14, γ = 0.51 0.98 3.04 112 useful when designing supply chain aiming at minimising food losses.
MW 90 0.20, γ = 0.41 0.98 3.15 114
One of the most significant impacts of RH on strawberry quality during
2013 M0 40 9.94 0.84 9.69 161
M0 60 8.77 0.62 15.48 181 storage, was on weight loss. Weight loss significantly increased when
M0 70 8.79 0.72 12.15 171 the storage RH decreased. In turn, the weight loss was correlated with
M0 80 8.29 0.84 8.1 154 the changes that occurred in visual appearance and chemical proper-
M0 90 7.97 0.88 6.59 145
ties.
M1 40 0.23 0.96 5.10 134
M1 60 0.27 0.84 10.38 164
Commercially, strawberry quality and shelf life are determined by
M1 70 0.22 0.88 8.08 153 visual inspection. Overall appearance (i.e., average score of colour and
M1 80 0.16 0.92 5.8 140 shrivelling), was modelled with a zero-order kinetics model for the
M1 90 0.14 0.94 4.52 129 various RH conditions. It was found that the lower RH increased the
MW 40 0.25, γ = 0.6 0.99 2.38 102
rate of appearance deterioration and hence limited the remaining shelf
MW 60 0.45, γ = 0.3 0.97 4.57 129
MW 70 0.24, γ = 0.4 0.98 3.21 115 life of strawberries. The Weibull model was the best fit for the data
MW 80 0.14, γ = 0.5 0.97 3.2 114 related to the chemical properties and it was found to be an important
MW 90 0.12, γ = 0.7 0.97 3.3 115 tool in explaining the changes occurring and the effect of the various
RH conditions during storage. Our results showed that improper RH
M0 = Zero-order (Eq. (1)), M1 = 1st-order (Eq. (2)), MW = Weibull model (Eq.
storage conditions have a significant impact on the rate of degradation
(3)).
of chemical components and affects negatively nutritional and bio-
chemical properties such as retention of AA, TPC, and ANC. In addition,
sugar content, a major quality attribute in strawberries that determines
acceptability by the consumer, was negatively affected by lower RH

Fig. 3. Effect of storage under various RH conditions on the AA retention (%) of fresh strawberries harvested in 2012. Error bars represent the standard deviation of
the mean, plotted lines correspond to the values estimated by the Weibull model.

155
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

Table 5
Kinetic constants and goodness of fit parameters of the Weibull model of ascorbic acid, anthocyanins, and total phenol retention in fresh strawberries kept at various
RH conditions.
Retention % RH 2012 2013

Rate constants R2 RMSE AIC Rate constants R2 RMSE AIC

(per day) (per day)

Ascorbic acid 40 0.66, γ = 0.31 0.98 3.48 172.92 0.66, γ = 0.31 0.92 7.93 173
60 0.34, γ = 0.42 0.98 3.69 193.08 0.57, γ = 0.20 0.82 11.53 168
70 0.17, γ = 0.48 0.95 5.95 177.06 0.17, γ = 0.64 0.78 11.26 167
80 0.15, γ = 0.61 0.96 3.88 122.56 0.24, γ = 0.47 0.77 12.76 173
90 0.13, γ=1.02 0.42 17.95 112.23 0.14, γ = 0.66 0.55 17.91 187
Anthocyanins 40 0.73, γ = 0.37 0.99 3.27 115.35 0.17, γ = 0.95 0.96 4.22 126
60 0.80, γ = 0.34 0.99 3.14 113.63 0.10, γ = 0.81 0.93 4.70 131
70 0.14, γ = 0.48 0.81 9.57 160.48 0.09, γ = 0.85 0.84 6.74 146
80 0.12, γ = 0.91 0.85 7.62 150.90 0.02, γ = 0.33 0.83 5.67 138
90 0.003, γ = 0.12 0.67 10.94 166.06 0.10, γ=1.66 0.72 8.48 155
Total Phenols 40 0.30, γ = 0.56 0.97 4.34 127.29 0.26, γ = 0.52 0.99 1.79 90
60 0.33, γ = 0.40 0.99 2.84 109.44 0.20, γ = 0.56 0.99 1.64 86
70 0.26, γ = 0.51 1.00 1.61 85.69 0.10, γ=1.31 0.84 6.71 146
80 0.12, γ = 0.75 0.91 5.94 140.40 0.07, γ=1.42 0.71 6.81 146
90 0.09, γ = 0.76 0.89 5.58 137.83 0.11, γ=4.45 0.60 12.52 172

Fig. 4. Effect of storage under various RH conditions on the total sugar retention (%) of fresh strawberries harvested (a) in 2012 and (b) in 2013. Error bars represent
the standard deviation of the mean, plotted lines correspond to the values estimated by the Weibull model.

storage conditions. Overall, the Weibull model showed the rate of de- between harvests. Higher contents of ANC and TPC were found in
terioration was the highest when strawberries were kept at 40 or 60% strawberries from the 2012 harvest possibly due to the higher tem-
RH. peratures occurring around the time of harvest. On the contrary, the
Furthermore, significant differences in quality were observed higher temperatures in 2012, may have affected negatively the sugar

156
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

Table 6
Kinetic constants and goodness of fit parameters of the Weibull model of SSC and TA contents in fresh strawberries kept at various RH conditions.
Composition RH 2012 2013

2
Rate constants R RMSE AIC Rate constants R2 RMSE AIC
(per day) (per day)

SSC 40 0.51, γ = 0.39 1.00 3.01 111.92 0.21, γ = 0.42 0.98 3.97 124
60 0.46, γ = 0.45 1.00 2.89 110.23 0.18, γ = 0.64 0.99 3.37 117
70 0.26, γ = 0.54 0.99 3.50 118.18 0.15, γ = 0.64 0.98 4.24 126
80 0.18, γ = 0.47 0.96 6.11 141.61 0.13, γ = 0.63 0.98 4.03 124
90 0.14, γ = 0.54 0.99 3.89 122.60 0.12, γ = 0.95 0.93 6.73 146
TA (%) 40 0.32, γ = 0.44 0.99 2.20 98.73 0.23, γ = 0.44 0.99 0.35 21
60 0.25, γ = 0.58 0.99 2.20 98.64 0.17, γ = 0.55 0.99 0.36 23
70 0.19, γ = 0.80 0.97 2.21 98.88 0.12, γ = 0.58 0.98 0.45 32
80 0.15, γ = 0.81 0.95 2.28 100.19 0.11, γ = 0.56 0.93 0.87 60
90 0.11, γ=1.05 0.92 2.20 98.68 0.08, γ = 0.76 0.93 0.71 51

Table 7 levels of AA, TPC, ANC and sugars due to pre-harvest conditions.
Confusion matrix of the logistic regression model. Finally, using the logistic regression classification method, our re-
Harvests 2012 & 2013 sults showed that RH and time after harvest have a strong predictive
power in forecasting the number of samples that would be unacceptable
Predicted based on the visual appearance. This highlighted the importance of RH
in determining the shelf life of fresh strawberries and therefore it be
Fail Pass Total
considered when predictive modelling is used for designing the supply
Actual Fail 80 12 92 chain with the view to minimising unnecessary losses.
Pass 32 476 508
Total 112 488 600 Acknowledgments

This research was funded by USDA – NIFA Specialty Crop Research

Initiative Grant (Project CA-D-PLS-2044-OG). The authors have also
received funding from the European Union's Horizon 2020 research and
innovation programme under the Marie Sklodowska-Curie grant
agreement No 70,837.

References

Amodio, M.L., Derossi, A., Colelli, G., 2013. Modelling sensorial and nutritional changes
to better define quality and shelf life of fresh-cut melons. J. Agric. Eng. 43 (4).
https://doi.org/10.4081/jae.2013.e6. 6-e6.
Caleb, O.J., Wegner, G., Rolleczek, C., Herppich, W.B., Geyer, M., Mahajan, P.V., 2016.
Hot water dipping: impact on postharvest quality, individual sugars, and bioactive
compounds during storage of ‘Sonata’ strawberry. Sci. Hortic. 210, 150–157. https://
doi.org/10.1016/j.scienta.2016.07.021.
Castro, I., Gonçalves, O., Teixeira, J.A., Vicentesc, A.A., 2002. Comparative study of selva
and camarosa strawberries for the commercial Market. J. Food Sci. 67 (6),
2132–2137. https://doi.org/10.1111/j.1365-2621.2002.tb09515.x.
Cayo, Y.P., Sargent, S., Nunes, C.D., Whitaker, V., 2016. Composition of commercial
strawberry cultivars and advanced selections as affected by season, harvest, and
postharvest storage. Hortscience 51 (9), 1134–1143. https://doi.org/10.21273/
HORTSCI10996-16.
Corradini, M.G., Peleg, M., 2004. A model of non-isothermal degradation of nutrients,
pigments and enzymes. J. Sci. Food Agric. 84 (3), 217–226. https://doi.org/10.1002/
jsfa.1647.
Flores-Felix, J.D., Velazquez, E., Garcia-Fraile, P., Gonzalez-Andres, F., Silva, L.R., Rivas,
Fig. 5. Receiver operating characteristic (ROC) curve of logistic regression R., 2018. Rhizobium and phyllobacterium bacterial inoculants increase bioactive
model predicting “Pass-Fail” score of fresh strawberries kept under various RH compounds and quality of strawberries cultivated in field conditions. Food Res. Int.
conditions. The area under the ROC curve, is 0.97 (p = 0). 111, 416–422. https://doi.org/10.1016/j.foodres.2018.05.059.
Halvorsen, B.K., 2015. Material from the book’s webpage, R port and packaging.
ElemStatLearn: Data Sets, Functions and Examples from the Book: "The Elements of
content and the ascorbic acid contents. Arising from the Weibull model Statistical Learning, Data Mining, Inference, and Prediction" by Trevor Hastie,
Tibshirani, R. and Friedman, J. R Package Version 2015.6.26. . https://CRAN.R-
the degradation rate of many chemical properties was different between project.org/package=ElemStatLearn.
the two harvests, following the same pattern as the rate of weight loss Hamner, B., Frasco, M., 2017. Metrics: Evaluation Metrics for Machine Learning. R
for these harvests. Pre-harvest conditions can affect postharvest quality Package Version 0.1.3. https://CRAN.R-project.org/package=Metrics.
Holcroft, D.M., Kader, A.A., 1999. Controlled atmosphere-induced changes in pH and
and shelf life. This study showed the significant interaction between
organic acid metabolism may affect color of stored strawberry fruit. Postharvest Biol.
pre-harvest and postharvest conditions on the overall concentration of Technol. 17 (1), 19–32. https://doi.org/10.1016/S0925-5214(99)00023-X.
nutritional properties over storage. Higher initial concentrations due to Hosahalli, S.R., 2015. Post harvest physiology of fruits and vegetables: transpiration.
favourable conditions, resulted in higher concentrations after 7 days of Post-Harvest Technologies of Fruits & Vegetables). DEStech Publications Inc.,
Pennsylvania, USA.
storage even when the degradation rate was higher compared to a Jouquand, C., Chandler, C., Plotto, A., Goodner, K., 2008. A sensory and chemical ana-
harvest with lower initial concentrations. Similarly, the study showed lysis of fresh strawberries over harvest dates and seasons reveals factors that affect
that higher RH storage (RH ≥ 80%) can extend the shelf life and aid in eating quality. J. Am. Soc. Hortic. Sci. 133 (6), 859–867. https://doi.org/10.21273/
JASHS.133.6.859.
the retention of overall quality even in strawberries with lower initial Kader, A.A., Salveit, M.E., 2003. Respiration and gas exchange. In: Bartz, J.A., Brecht, J.K.

157
A. Ktenioudaki, et al. Postharvest Biology and Technology 154 (2019) 148–158

(Eds.), Postharvest Physiology and Pathology of Vegetables, Second Edition, Revised conditions encountered during typical consumer retail display affect fruit and vege-
and Expanded. Marcel Dekker, Inc., New York, USA, pp. 7–31. table quality and waste. Postharvest Biol. Technol. 51 (2), 232–241. https://doi.org/
Kader, F., Nicolas, J.-P., Metche, M., 1999. Degradation of pelargonidin 3-glucoside in the 10.1016/j.postharvbio.2008.07.016.
presence of chlorogenic acid and blueberry polyphenol oxidase. J. Sci. Food Agric. 79 Odriozola-Serrano, I., Soliva-Fortuny, R., Martin-Belloso, O., 2009. Influence of storage
(4), 517–522. https://doi.org/10.1002/(SICI)1097-0010(19990315)79:4<517::AID- temperature on the kinetics of the changes in anthocyanins, vitamin C, and anti-
JSFA212>3.0.CO;2-N. oxidant capacity in fresh-cut strawberries stored under high-oxygen atmospheres. J.
Kelly, K., Nunes, C., Whitaker, V.M., 2016a. A comparison of physical and chemical at- Food Sci. 74 (2), C184–191. https://doi.org/10.1111/j.1750-3841.2009.01075.x.
tributes of strawberry cultivars and advanced breeding selections from the University Oms-Oliu, G., Odriozola-Serrano, I., Soliva-Fortuny, R., Martín-Belloso, O., 2009. Use of
of Florida. Proc. Florida State Hortic. Soc. 129, 185–189. weibull distribution for describing kinetics of antioxidant potential changes in fresh-
Kelly, K., Whitaker, V.M., Nunes, M.Cd.N., 2016b. Physicochemical characterization and cut watermelon. J. Food Eng. 95 (1), 99–105. https://doi.org/10.1016/j.jfoodeng.
postharvest performance of the new sensation ® ‘Florida127’ strawberry compared to 2009.04.016.
commercial standards. Sci. Hortic. 211, 283–294. https://doi.org/10.1016/j.scienta. Ong, S.P., Law, C.L., Hii, C.L., 2011. Effect of pre-treatment and drying method on colour
2016.09.012. degradation kinetics of dried salak fruit during storage. Food Bioprocess Technol. 5
Kelly, K., Madden, R., Emond, J.P., do Nascimento Nunes, M.C., 2018. A novel approach (6), 2331–2341. https://doi.org/10.1007/s11947-011-0647-y.
to determine the impact level of each step along the supply chain on strawberry Ooms, J., 2018. Writexl: Export Data Frames to Excel’ Xlsx’ Format. R Package Version
quality. Postharvest Biol. Technol. 147, 78–88. https://doi.org/10.1016/j. 1.0. https://CRAN.R-project.org/package=writexl.
postharvbio.2018.09.012. Pan, L., Zhang, W., Zhu, N., Mao, S., Tu, K., 2014. Early detection and classification of
Klein, B.P., 1987. Nutritional consequences of minimal processing of fruits and vege- pathogenic fungal disease in post-harvest strawberry fruit by electronic nose and gas
tables. J. Food Qual. 10 (3), 179–193. https://doi.org/10.1111/j.1745-4557.1987. chromatography–mass spectrometry. Food Res. Int. 62, 162–168. https://doi.org/10.
tb00857.x. 1016/j.foodres.2014.02.020.
Kuhn, M., Contributions from Jed Wing, J, Weston, S., Williams, A., Keefer, C., Prasad, Y.L., 2016. Big Data Analytics Made Easy. Notion Press, Chennai, India.
Engelhardt, A., Cooper, T., Mayer, Z., Kenkel, B., R Core Team, Benesty, M., R Core Team, 2017. R: A Language and Environment for Statistical Computing. URL. R
Lescarbeau, R., Ziem, A., Scrucca, L., Tang, Y., Candan, C., Hunt, T., 2018. Caret: Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/.
Classification and Regression Training. R Package Version 6.0-79. https://CRAN.R- Robin, X., Turck, N., Hainard, A., Tiberti, N., Lisacek, F., Sanchez, J.C., Muller, M., 2011.
project.org/package=caret. pROC: an open-source package for R and S+ to analyze and compare ROC curves.
Lai, Y.P., Emond, J.P., Nunes, M.C.N., 2011. Environmental conditions encountered BMC Bioinf. 12, 77.
during distribution from the Field to the Store affect the quality of strawberry Robinson, J.E., Browne, K.M., Burton, W.G., 1975. Storage characteristics of some ve-
(‘Albion’). Proc. Florida State Hortic. Soc. 124, 213–220. getables and soft fruits. Ann. Appl. Biol. 81 (3), 399–408.
Laurin, E., Nunes, M.C.N., Brecht, J.K., Emond, J.P., 2005. Vapor pressure deficit and Schwieterman, M.L., Colquhoun, T.A., Jaworski, E.A., Bartoshuk, L.M., Gilbert, J.L.,
water loss patterns during simulated air shipment and storage of Beit alpha cu- Tieman, D.M., Odabasi, A.Z., Moskowitz, H.R., Folta, K.M., Klee, H.J., Sims, C.A.,
cumbers. Mencarelli, F., Tonutti, P. (Eds.), Proceedings of the 5th International Whitaker, V.M., Clark, D.G., 2014. Strawberry flavor: diverse chemical compositions,
Postharvest Symposium Vols 1-3, 1697–1703. a seasonal influence, and effects on sensory perception. PLoS One 9 (2), e88446.
Mc Carthy, U., Uysal, I., Badia-Melis, R., Mercier, S., O’Donnell, C., Ktenioudaki, A., 2018. https://doi.org/10.1371/journal.pone.0088446.
Global food security – issues, challenges and technological solutions. Trends Food Sci. Shin, Y., Liu, R.H., Nock, J.F., Holliday, D., Watkins, C.B., 2007. Temperature and relative
Technol. 77, 11–20. https://doi.org/10.1016/j.tifs.2018.05.002. humidity effects on quality, total ascorbic acid, phenolics and flavonoid concentra-
Meyer, D., Dimitriadou, E., Hornik, K., Weingessel, A., Leisch, F., 2017. e1071: Misc tions, and antioxidant activity of strawberry. Postharvest Biol. Technol. 45 (3),
Functions of the Department of Statistics, Probability Theory Group (Formerly: 349–357. https://doi.org/10.1016/j.postharvbio.2007.03.007.
E1071), TU Wien. R Package Version 1.6-8. https://CRAN.R-project.org/package= Shin, Y., Ryu, J.-A., Liu, R.H., Nock, J.F., Watkins, C.B., 2008. Harvest maturity, storage
e1071. temperature and relative humidity affect fruit quality, antioxidant contents and ac-
Nunes, M.C.N., 2015. Correlations between subjective quality and physicochemical at- tivity, and inhibition of cell proliferation of strawberry fruit. Postharvest Biol.
tributes of fresh fruits and vegetables. Postharvest Biol. Technol. 107, 43–54. https:// Technol. 49 (2), 201–209. https://doi.org/10.1016/j.postharvbio.2008.02.008.
doi.org/10.1016/j.postharvbio.2015.05.001. Sing, T., Sander, O., Beerenwinkel, N., Lengauer, T., 2005. “ROCR: visualizing classifier
Nunes, M.C.N., Delgado, A., 2014. Quality of Organic Compared to Conventionally Grown performance in R.”. Bioinformatics 21 (20), 7881. < URL. http://rocr.bioinf.mpi-sb.
Strawberries at the Retail Level, in 1049 ed. International Society for Horticultural mpg.de.
Science (ISHS), Leuven, Belgium, pp. 723–730. Taoukis, P., Labuza, T., Saguy, I., 1997. Kinetics of food deterioration and shelf-life
Nunes, M.C.N., Emond, J.P., 2007. Relationship between weight loss and visual quality of prediction. In: Valentas, K., Rotstein, E., Singh, R. (Eds.), Handbook of Food
fruits and vegetables. Proc. Florida State Hortic. Soc. 120, 235–245. Engineering Practice. CRC Press, Boca Raton, pp. 361–403.
Nunes, M.C.N., Brecht, J.K., Morais, A.M.M.B., Sargent, S., 1998. Controlling temperature Therneau, T., Atkinson, B., 2018. Rpart: Recursive Partitioning and Regression Trees. R
and water loss to maintain ascorbic acid levels in strawberries during postharvest Package Version 4.1-13. https://CRAN.R-project.org/package=rpart.
handling. J. Food Sci. 107, 1033–1036. https://doi.org/10.1111/j.1365-2621.1998. Tuszynski, J., 2014. caTools: R Package Version 1.17.1. https://CRAN.R-project.org/
tb15848.x. package=caTools.
Nunes, M.C.N., Ä–mond, J.P., Brecht, J.K., 2003. Predicting Shelf Life and Quality of Van de Velde, F., Fenoglio, C., Piagentini, A.M., Pirovani, M.E., 2017. Modeling the im-
Raspberries under Different Storage Temperatures, in 628 ed. International Society pact of the type of cutting and storage temperature on the bioactive compound
for Horticultural Science (ISHS), Leuven, Belgium, pp. 599–606. content, phenylpropanoid metabolism enzymes and quality attributes of fresh-cut
Nunes, M.C.N., Brecht, J.K., Morais, A.M.M.B., Sargent, S.A., 2005. Possible influences of strawberries. Food Bioprocess Technol. 11 (1), 96–109. https://doi.org/10.1007/
Water loss and polyphenol oxidase activity on anthocyanin content and discoloration s11947-017-1996-y.
in fresh ripe strawberry (cv. OsoGrande) during storage at 1 °C. J. Food Sci. 70 (1), Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics With S, fourth edition.
79–84. https://doi.org/10.1111/j.1365-2621.2005.tb09069.x. Springer, New York.
Nunes, M.C.N., Brecht, J.K., Morais, A.M.M.B., Sargent, S.A., 2006. Physicochemical Wang, S.Y., Camp, M.J., 2000. Temperatures after bloom affect plant growth and fruit
changes during strawberry development in the field compared with those that occur quality of strawberry. Sci. Hortic. 85, 183–199. https://doi.org/10.1016/S0304-
in harvested fruit during storage. J. Sci. Food Agric. 86 (2), 180–190. https://doi.org/ 4238(99)00143-0.
10.1002/jsfa.2314. White, J., 2007. How cold was It? Know the whole story. Frozen Food Age 56, 38–40 ISSN
Nunes, M.C.N., Emond, J.P., Rauth, M., Dea, S., Chau, K.V., 2009. Environmental 00162191.

158