linking gas exchange to quality change - ku leuvenu0040603/thesis/ch3_linking gas exc… · ddtk °...

Quality Change Modelling in Postharvest Biology and Technology

Maarten L.A.T.M. Hertog - 2004

3

Linking gas exchange to quality

change

3.1 Introduction

Quality of horticultural product is largely based on subjective consumer

evaluation of a complex of quality attributes (like taste, texture, colour,

appearance), which are based on specific product properties (like sugar content,

volatile production, cell wall structure; Sloof et al., 1996, Shewfelt, 1999). These

product properties are generally changing during time, as part of the normal

metabolism of the product. To understand the mode of action of MA on quality

change for a specific product, a good understanding of how relevant product

properties depend on storage conditions is required. This chapter will deal in

more detail with modelling the link between the effects of storage conditions on

metabolic rate on one hand and the effects of storage conditions on external

quality aspects on the other hand.

Temperature is the main factor affecting all biochemical processes through its

effects on activation enthalpy and entropy of the underlying reactions. Energy

52 Quality Change Modelling in Postharvest Biology and Technology

demanding processes are also indirectly affected by temperature through the

effect of temperature on respiration and fermentation, the main energy producing

processes. The levels of O2 and CO2 inhibiting respiration also affect the amount

of energy produced by respiration and fermentation. Those quality changes that

are either directly influenced by O2 or CO2 or driven by the energy supplied by

respiration or fermentation will all be affected by MA conditions. Some quality

degrading processes are affected more than others due to the way they depend on

atmospheric conditions. In spite of the large volume of research on modified

atmospheres (MA) showing the general effects of MA packaging and controlled

atmosphere storage on the respiration rate as such, data on the quantification of

the rates of quality loss in relation to the applied MA conditions is still limited.

The objective of this chapter is to show how the effects of MA on product

quality can be explained by the effect MA has on gas exchange. Four case studies

are discussed on four different product ³ quality attribute combinations; aroma

production of apple (section 3.2), spoilage of strawberry (section 3.3), stem

growth of Belgian endive (section 3.4) and softening of kiwifruit (section 3.5).

Together they cover a wide range of postharvest issues.

The case study on aroma production of apple (section 3.2) shows how the

effects of MA can be understood through an inhibiting effect on an early reaction

step in a linear reaction chain, without making a priori assumptions on how this

rate constant would relate to the applied gas conditions. The remaining three case

studies explicitly define the link between MA conditions and the rate of

metabolic processes by assuming that MA is affecting the rate of gas exchange

which on its turn is generating ATP to drive all energy demanding metabolic

process.

The case study on the spoilage of strawberry (section 3.3) is linking the rate

of spoilage to the overall gas exchange rate as expressed by the total CO2

production, not discriminating between respiration and fermentation. The case

study on stem growth of Belgian endive (section 3.4) is the only non-ripening

related example showing that the concept of linking rates of gas exchange to rates

of quality change also works for such a primary process as growth. As stem

growth was completely inhibited at anaerobic conditions, the rate of stem growth

was only linked to respiration. The case study on the softening of kiwifruit

(section 3.5) is the one with the most detail as the effect of MA on the rate of

softening was differentiated with respect to fermentation and respiration.

Together, the four case studies support the hypothesis that quality attributes

changes are often driven by the energy status of the tissue as indicated by the

3. Linking gas exchange to quality change 53

rates of gas exchange. This is the first time in the field of postharvest research

that such a relationship has been quantitatively characterised.

Parts of the material presented in this chapter were published in Hertog et al.

(1999a and 2004a), Hertog and Nicholson (2003) and Hertog (2003b).

3.2 Aroma production of apple

The quality of pome fruit is not only determined by appearance, firmness and

texture, but to a large extent also by flavour. Flavour is the result of both taste

and volatile aroma compounds. In order to maintain quality, pome fruit is often

stored under MA. However, though most of the quality attributes are well

preserved under MA, the aroma production capacity decreases (Thompson, 1998;

Fellman et al. 2003; Saquet et al., 2003). Modified atmospheres can thus be

detrimental to post storage volatile production when compared to regular air

storage. Also time to regenerate aroma profiles after removal from MA

conditions can be affected. When comparing post storage production of the

different volatile compounds, the time lapse for the different compounds can vary

largely (Fellman et al., 2003; Saevels et al., 2004).

Over 200 volatile components have been identified in the aroma of various

apple cultivars, most of them being esters (Dimick and Hoskin, 1983). However,

the pathway of ester synthesis in apples is not completely understood. The major

esters produced by ripening fruit are thought to arise primarily from lipid and

amino acid degradation, pathways that are active in ripening apples (Bartley et

al., 1985).

Using experimental data from Saevels et al. (2004) the possibility was studied

to explain differences in post storage volatile production in relation to the

preceding storage gas conditions applied.

3.2.1 Materials and methods

The experimental setup was described in full detail by Saevels et al., 2004, with

the relevant information summarised below.


3.2.1.1 Fruit

‘Jonagold’ apples (Malus sylvestris subsp. Mitis (Wallr.) Mansf.) were harvested

in 2002 at the experimental station ‘Proeftuin voor Pit-en Steenfruit’ (Velm,

Belgium) at their optimal harvest date for long-term storage.

3.2.1.2 Storage conditions

The apples were stored for eight months in small storage containers under three

different storage atmospheres: ultra low oxygen (ULO: 1 °C, 1 kPa O2, 2.5 kPa

CO2), controlled atmosphere (CA; 1 °C, 3 kPa O2, 2.5 kPa CO2) and regular air

(RA; 1 °C, 20.8 kPa O2, 0.03 kPa CO2). After eight months storage eight apples

were taken from each storage environment and transferred to shelf life conditions

(20 °C, 20.8 kPa O2 and 0.03 kPa CO2) to equilibrate for 1 d before the first

aroma production measurement was taken.

Table 3.1 The 22 most abundant volatile compounds in ripening ‘Jonagold’

apple fruit identified by SPME–GC/MS (after Saevels et al., 2004)

ID Volatile compound ID Volatile compound

Æ1

Æ2

Æ3

Æ4

Æ5

Æ6

Æ7

Æ8

Æ9

Æ10

Æ11

propyl acetate

2-methyl butanol

2-methylpropyl acetate

propyl propanoate +

butyl acetate a

2-methylbutyl acetate

propyl butanoate

butyl propanoate

pentyl acetate

2-methylpropyl butanoate

6-methyl-5-hepten-2-one

butyl butanoate

Æ12

Æ13

Æ14

Æ15

Æ16

Æ17

Æ18

Æ19

Æ20

Æ21

Æ22

hexyl acetate

butyl 2-methylbutanoate +

methyl 2-ethylhexanoate a

butyl pentanoate + propyl hexanoate a

2-methylbutyl 2-methylbutanoate +

hexylpropanoate a

butyl hexanoate + hexyl butanoate a

hexyl 2-methylbutanoate

isopentyl hexanoate

pentyl hexanoate + butyl heptanoate a

propyl caproate

hexyl hexanoate + butyl caproate a

Ŭ-Farnesene a Peaks in the chromatograms were not separated.

3.2.1.3 Fruit measurements

Volatile production of the apples was measured after, respectively 1 d, 5 d, 8 d,

12 d and 15 d of shelf life following long-term storage. For the collection of the

volatiles, each fruit was placed in a flushed, airtight jar at 23 °C. The headspace


was allowed to equilibrate for 1 h before sampling the accumulating volatiles by

means of solid-phase micro extraction (SPME). The volatiles were separated and

identified using GC/MS. The 22 most abundant volatiles were selected (Table

3.1) and the volatile production rate was expressed as the abundance of the peak

area under the chromatogram relative to the maximum production rate observed

for that particular volatile compound.

Saevels et al. (2004) discriminated between both propyl and butyl caproate

(Æ20 and Æ21) on one hand and propyl and butyl hexanoate (Æ14 and Æ16) on the

other hand while as a matter of fact these are identical molecules. Reassessment

of the original GC/MS-spectra showed that Æ20 (identified as propyl caproate)

should have been propyl octanoate while the butyl caproate identified in Æ21

should have been butyl caprylate.

3.2.2 Modelling approach

The basic assumption for the modelling approach is that the aroma volatiles (Æ)

result from certain lipid and amino acid degradation pathways in which the

maximum aroma production is limited by the limited size of the available pool of

substrates. A simple model was proposed consisting of three consecutive

reactions describing how the substrate (S), via one intermediate compound (I) is

degraded into an aroma compound (Æprod) that subsequently evaporates from the

fruit into the surrounding air (Æair).

airprod ÆÆ21 ½½½½½½½ vbb kkkIS (3.1)

All three rate constants (kb1, kb2, kv) were assumed to depend on temperature

according Arrhenius (Eq. 2.12). This simplified scheme can be interpreted as a

simplification of any linear chain of consecutive reactions breaking down some

substrate into an end product. The concentrations measured during the headspace

analysis are a direct measure of the flow of volatiles leaving the fruit (kvÖÆprod).

The model was used in its ODE form (Eq. 3.2) to describe the overall fruit

history of storage and shelf life to explain differences in shelf life behaviour of

fruit stored at ULO, CA or RA by assuming that only the first rate constant (kb1)

was depending on the MA conditions. In this first case study no a priori

assumptions were made on how this rate constant would relate to the gas

conditions. Instead, separate values were estimated for kb1 for the three gas

conditions applied while all other parameters were estimated in common per

aroma compound. Parameters were estimated using the optimisation routines


from MatLab (MatLab v. 6.5, 2002, The MathWorks, Inc., Natick, MA, USA;

see also chapter 5).

1

1 2

prod prod2

air prod

Æ Æ

Æ Æ

b

b b

b v

v

dS dt k S

dI dt k S k I

d dt k I k

d dt k

= - Öëîî = Ö - Öîì

= Ö - Öîî

= Öîí

(3.2)

3.2.3 Results and discussion

All, but three, of the aroma compounds identified were esters varying in length

from 5 to 12 C-atoms. The odd ones out were the alcohol 2-methyl butanol (Æ2),

the acyclic sesquiterpene Ŭ-farnesene (Æ22) and its autoxidation product, 6-

methyl-5-hepten-2-one (Æ10) which is a ketone.

Æ1

ULO

CA

RA

Æ2

Æ3

Æ4

Æ5

Æ6

Æ7

Æ8

Æ9

Æ10

Æ11

Æ12

Æ13

Æ14

Æ15

Æ16

0 5 10 15

Æ17

0 5 10 15

Æ18

0 5 10 15

Æ19

0 5 10 15

Æ20

0 5 10 15

Æ21

0 5 10 15

Æ22

rela

tive

vo

latile

pro

du

ctio

n

time (d)

Fig. 3.1 Volatile production of ‘Jonagold’ apple during 20 °C shelf life after 8

months storage at either RA, CA or ULO. The volatile production was expressed

as the abundance of the peak area under the chromatogram relative to the

maximum production rate observed for each volatile compound. The labels (Æ1-

Æ22) refer to the aroma compounds as identified in Table 3.1. The error bars

indicate the standard deviation of the measurements. The lines represent the

model from Eq. 3.2.


The experimental results (Fig. 3.1) show a wide range of different time lapses

during shelf life with large standard deviations as indicated by the error bars.

Depending on the volatile compound the maximum production is reached earlier

or later during shelf life. Also depending on the compound, either ULO or RA

stored fruit resulted in the highest peak value during shelf life, with CA stored

fruit showing an intermediate volatile production. Only Æ17, was an exception on

this with the CA stored fruit showing the highest peak value. The alcohol

compound (Æ2) was the only one showing more or less constant levels during

shelf life.

Based on the data (Fig. 3.1) a grouping can be made of compounds showing

comparable release patterns. However, these groupings could not be related to a

correspondence in molecular structure, functional groups or length of the C-

skeleton.

The simple linear reaction chain approach was able to capture the behaviour

of all measured volatile compounds (Fig. 3.1) by attributing the effect of the

applied MA conditions solely to an effect on the first reaction step. The effect of

the O2 level applied during storage on the biosynthesis of the different aroma

compounds during shelf life was captured in the estimate values of kb1 and varied

per volatile compound (Fig. 3.2). The effect of O2 on kb1 varied from a linear to a

sigmoidal effect while for Æ17 a minimum value of kb1 was observed during CA

storage at 3 kPa O2 (Fig. 3.2).

Fig. 3.2 The effect of O2 on the

estimated value of kb1 (in d-1). Each of

the lines represent the values of kb1 for

a different aroma compound.

0 5 10 15 200.0

0.2

0.4

0.6

0.8

1.0

Æ17

kb1

pO

2

(kPa)

This case study shows how the effects of MA on volatile production can be

understood through an inhibiting effect on an early reaction step in a linear

reaction chain. In this case study no a priori assumption was made on how this

rate constant would relate to the applied gas conditions. The different ways of

how volatile production depends on O2 levels should be related to their different

biochemical pathways, their dependence on ATP availability and their

dependence on O2 as a possible reactant directly or indirectly involved in their


biosynthesis. Besides interpreting the data from the point of view of the aroma

production itself one might also interpret the result from the point of view of de

novo synthesis of the substrate (fatty acids) which in the current approach was

ignored. Reduced respiration during MA can lead to depletion of energy storing

metabolites such as ATP and NADPH which are required for fatty acid

biosynthesis and desaturation (Saquet et al., 2003). Also the absolute requirement

for molecular oxygen in the desaturation process of fatty acids may further

modulate the suppressive MA effect on aroma production.

3.2.4 Conclusions

The aroma production during shelf life can thus be interpreted in terms of how

well substrate was conserved during the preceding storage period. By suppressing

the production of volatile during storage, more substrate will be available during

the subsequent shelf life. If the aroma production is not inhibited enough during

storage, the substrates are exhausted by the time the fruit were taken out for shelf

life. Of course this should be seen in connection with the parallel delay in

ripening as imposed by MA. As the storage temperatures were the same during

RA, CA and ULO the observed differences were the result from the different gas

conditions.

Even though the underlying biochemistry is more complex than the simplified

model might suggests, the model can add value to interpreting experimental data

on aroma production of MA stored fruit versus air stored fruit. The model can be

used to test different working hypotheses and thus create a better understanding

of the underlying processes.

In the following case studies the link between MA conditions and the rate of

metabolic processes will be made explicit by assuming that MA is affecting the

rate of gas exchange which on its turn is generating ATP to drive all energy

demanding metabolic process.

3.3 Spoilage of strawberry

Botrytis infection is a major factor limiting keeping quality of strawberries

(Browne et al. 1984; Ghaouth et al. 1991; Chambroy et al. 1993; Vaughn et al.

1993; Saks et al. 1996). Spoilage is one of the first visible attributes the

consumer is confronted with in assigning quality to strawberries. The main

criterion is whether strawberries are visibly affected or not, rather than the degree


of decay. Strawberry tissue may deteriorate because of Botrytis infection, and

Botrytis may develop consequent to tissue softening due to ripening. However,

late-harvested and ripened strawberries are more susceptible to fungal spoilage

than early-picked strawberries (Browne et al., 1984). It was therefore assumed

that fungal growth occurred as soon as the tissue structure gives the opportunity

to do so.

Since the growth of Botrytis under favourable conditions (on potato dextrose

agar; Agar et al., 1990) is much faster than the rate of ripening of strawberries

(Woodward, 1972), the latter process will be rate-limiting for quality loss of

strawberries. As a consequence, inhibition of spoilage by MA or CA can be

explained by the inhibitory effect of the gas composition on ripening.

The aim of this study was to develop an integrated model describing the

effect of changed gas conditions on keeping quality of ‘Elsanta’ strawberries as

limited by spoilage. Such a model will considerably enhance the understanding

of the underlying processes and their mutual interference. Once calibrated, the

model can be used to optimise package and transport conditions to specific

demands. A number of experimental data sets were used to examine the

assumptions made.

3.3.1 Material and methods

The experimental data originates from different experiments executed over

several years at the former ATO-DLO (Wageningen, NL; Hertog et al., 1999a).

3.3.1.1 Fruit

Fruits of ‘Elsanta’ strawberries (Fragaria x ananassa Duchesne) were obtained

from commercial growers directly after harvest at a fully ripe, red-coloured stage.

Batches of both glasshouse and field cultured fruits were obtained. Only

strawberries without visible Botrytis infection were used. Before packing, fruits

were stored for approximately 18 h at 1 °C.


Each of the experiments had its own experimental conditions with regard to

package (pallets, crates or consumer size packs), applied gas conditions (air:

21 kPa O2 and 0 kPa CO2; CA: 5 kPa O2 and 15 kPa CO2; different MA

conditions with steady state concentrations varying from 2 kPa O2 and 20 kPa


CO2 to 17 kPa O2 and 3 kPa CO2) and applied temperature (varying from 1 °C to

16 °C). Seventeen different batches of strawberries were used to prepare 370

packages which were tested at the different conditions. Each of the packages is

unambiguously defined by its temperature, O2 and CO2 concentration as a

function of experimental time.


Gas exchange rates were measured for a separate batch of ‘Elsanta’ strawberries

at a range of O2 (0 kPa, 1 kPa, 3 kPa, 5 kPa, 10 kPa, 15 kPa, 21 kPa) and CO2

(0 kPa, 5 kPa, 10 kPa, 15 kPa, 20 kPa) levels at two temperatures (4 °C and

16 °C) applying a full factorial design. Measurements of O2 consumption and

CO2 production were performed as described by Peppelenbos and Van ‘t Leven

(1996).

The quality of strawberries was visually assessed by counting the number of

strawberries visibly affected by Botrytis, expressed as a percentage of the number

of strawberries present. No discrimination was made on the basis of level of

decay. The number of strawberries used depended on the size of the packages

(pallets and crates, 200-300 strawberries; consumer size packs, 30-50

strawberries). Assessments were conducted during the storage period of the

packed product (varying from 5 d to 9 d) and usually also during a subsequent

shelf life period at ambient conditions after unpacking (varying from 2 d to 5 d).


3.3.2.1 Spoilage

When looking at spoilage of strawberries by Botrytis in terms of percentage

strawberries affected (N), a sigmoidal behaviour from 0 % to 100 % (Nmax)

spoilage can be observed. After an initial exponential increase in the percentage

of strawberries affected, the rate of increase diminishes to zero until all

strawberries are affected. This general population dynamic behaviour can be

described by the differential equation:

max

max

N

NNNkdtdN d

-ÖÖ= (3.3)

with N0 the percentage of affected strawberries at time 0. The progress of

spoilage is solely determined by the spoilage rate constant kd (d-1) which is


considered equivalent to the rate of tissue deterioration. Spoilage will only

develop when an initial infection is present. The level of initial spoilage (N0) will

be batch dependent. As tissue deterioration is a biological process, the rate

constant kd is assumed to depend on temperature according to Arrhenius’ law

(Eq. 2.12).

3.3.2.2 Gas exchange

The Michaelis Menten type gas exchange model used to describe gas exchange

of ‘Elsanta’ strawberries is described in detail in chapter 2. The O2 consumption

rate (2Or ) was modelled according Eq. 2.1. The CO2 production is the

simultaneous result of oxidative and fermentative processes and was modelled

according Eq. 2.10. Both maxO2

r and maxf)(CO2

r depend on temperature according to

Arrhenius (Eq. 2.12).

3.3.2.3 Spoilage at modified atmospheres

The inhibition of spoilage by MA is assumed to result from the inhibitory effect

of the gas composition on gas exchange of the strawberry fruit. When gas

exchange is inhibited, the overall metabolic rate, and thus the rate of ripening,

will be inhibited, resulting in an inhibition of spoilage. So, the rate of quality

loss, in this case kd, is assumed to depend on the rate of gas exchange. As

outlined in chapter 2, relative respiration rate, defined as the ratio between the

actual respiration rate under any gas condition and the respiration rate in normal

air at the same temperature, can be used as a rate index for the overall

metabolism. To also take into account possible fermentative activities, a relative

metabolic rate (RR) was expressed using the total gas exchange, expressed in

terms of CO2 production (Eq. 2.10).

Assuming a one-to-one relationship between the rate of gas exchange and the

rate of quality loss, Eq. 3.3 can be extended to account for the effect of modified

atmospheres on spoilage of ‘Elsanta’ strawberries by Botrytis resulting in:

max

max

N

NNNkRRdtdN d

-ÖÖÖ= (3.4)

The spoilage model was used in its ODE form and implemented in PROSIM

(Prosim bv, Zoetermeer, The Netherlands) to estimate the parameter values.


3.3.2.4 Keeping Quality

Quality losses, such as caused by spoilage, can be expressed in terms of keeping

quality (Tijskens, 1995; Tijskens and Polderdijk, 1996; Hertog and Tijskens,

1998a). With changing temperatures acting on a product going through a logistic

chain, quality in terms of spoilage changes following Eq. 3.4. Once the MA pack

is opened for final shelf life at a specified temperature, KQ remaining for the end

user can be calculated by integrating Eq. 3.3 and solving for time resulting in:

( )( )

d

s

s

k

NNN

NNN

KQöö÷

õææç

å-Ö

-Ö

-= maxlim

limmaxln

(3.5)

with Ns the level of spoilage at the start of shelf life, Nlim the critical level of

spoilage allowed and kd the rate of spoilage (at shelf life temperature).



At O2 levels below 5 kPa fermentation occurs (Fig. 3.3), evident from an increase

in respiration quotient (2Or as compared to

2COr ). The measurements of gas

exchange at about 21 kPa O2 determined at different CO2 conditions indicated

that the effect of CO2 was negligible (data not shown). Li and Kader (1989) and

Talasila et al. (1992) reported only slight effects of CO2 on the gas exchange of

strawberries. However, such small effects could not be determined from the

current data. Colelli and Martelli (1995) reported a clear CO2 effect on the

respiration of ‘Pajaro’ strawberries, indicating that not all strawberry cultivars

respond similarly to increased CO2 levels.

The results of the non linear regression analysis of strawberry gas exchange

simultaneously using O2, CO2 and temperature as independent variables, and CO2

production and O2 consumption as dependent variables, are given in Table 3.2.

During the iterative process of non linear regression analysis, the parameters

2COKmc ,2COKmu and )f(CO2

Kmc tended towards large values, indicating that

CO2 inhibition is not involved. Therefore,2COKmc ,

2COKmu and )f(CO2Kmc were

fixed at +¤. In accordance with chapter 2, )f(CO2Km was fixed at 1 kPa. The gas

exchange model accounted for 85 % of the observed variance as is expressed by

R2adj. The simulated data generated by the model, applying the estimated

parameters from Table 3.2, is shown as solid lines in Fig. 3.3.


0 5 10 15 20

pO2(kPa)

0.0

0.2

0.4

0.6

0.8

1.0

r(C

)O2

(µ

mo

l·kg

-1·s

-1)

0 5 10 15 20

pO2(kPa)

A. B.

0.0 2.0

0.0

0.1

0.0 2.0

0.0

0.2

Fig. 3.3 O2 consumption (2Or in µmolÖkg-1Ös-1; + and solid lines) and CO2

production (2COr in µmolÖkg-1Ös-1;Ï and dotted lines) of ‘Elsanta’ strawberries

measured at 4 °C (A) and 16 °C (B) as a function of O2. The symbols are

measured values and the lines represent simulated values. The inserts show

enlargements of the behaviour around 0 kPa O2.

Table 3.2 Parameter estimates and their standard errors (s.e.) resulting from non

linear regression analysis of gas exchange data for ‘Elsanta’ strawberries withrefT = 10 °C.

Parameter (unit) Value (s.e.) Parameter (unit) Value (s.e) refmax,

O2r (µmolÖkg-1Ös-1)

refmax,

2Or

Ea (kJÖmol-1)

2OKm (kPa)

2COKmc (kPa)

2COKmu (kPa)

RQox (-)

0.27 (0.010)

74.8 (3.4)

2.63 (0.274)

+¤ a)

+¤ a)

0.91 (0.030)

refmax,

f)(CO2r (µmolÖkg-1Ös-1)

refmax,)(2CO f

rEa (kJÖmol-1)

2O (f)Kmc (kPa)

)f(CO2Kmc (kPa)

)f(CO2Km (kPa)

0.50 (0.22)

57.4 (14.4)

0.056 (0.041)

+¤ a)

1 a)

R2adj 85 % n 294

a) fixed value, s.e. is not applicable

Strawberry fruits have a high maximum O2 consumption rate ( refmax,

O2r ) in

combination with a low2OKm . This means that the high respiration level of

strawberry is not easily suppressed by lowering the O2 level. In contrast,

considering the low value of2O (f)Kmc of 0.056 kPa, fermentation is suppressed at

already low O2 levels. The temperature dependence of maxO2

r and the value of RQox

from Table 3.2 correspond with the results reported on ‘Pajaro’ strawberries by

Chambroy et al. (1993). Their results could be described by a2

max,refO

r of


0.25 µmolÖkg-1Ös-1 and an max

2OrEa of 65 kJÖmol-1 with an RQox of 0.9. These are

close to the values presented here.

3.3.3.2 Spoilage

Because of the multitude of data available it is impossible to describe all the

results in detail. Therefore, only the general behaviour and main trends are

presented. Fig. 3.4 shows the general sigmoidal pattern of spoilage with time.

The applied MA conditions resulted in a slight inhibition of spoilage.

Considering the point where spoilage reached a level of 5 % (one strawberry

infected in a consumer pack of 20 strawberries), MA resulted in an improvement

of 1.5 d over air storage. Commercially, this is a significant improvement. The

two batches presented in Fig. 3.4 are quantitatively different. After 9 d, 80 % of

the air-packed strawberries from batch 1 are affected as compared to only 35 %

from batch 2. Apparently, spoilage is strongly batch dependent. This might be

explained by various factors such as differences in production systems or

fungicide residues.

0 3 6 9

time (d)

0

20

40

60

80

sp

oila

ge

(%

)

batch 1

batch 2

Fig. 3.4 Spoilage of ‘Elsanta’ strawberries at 10 °C as a function of time. Two

batches of strawberries, air (Ã, Â) and MA (¹, ¸) packed were sampled. The

MA packages resulted in steady state values of about 10 kPa O2 and 12 kPa CO2.

The lines represent the model results based on Eq. 3.4 and the parameter values

from Table 3.2 and Table 3.3)


In most of the experiments, spoilage was determined after a certain period of

packaging and after a subsequent period of shelf life under ambient conditions.

Fig. 3.5 illustrates the effects of low temperature (air-packed at 1 °C) and CA

conditions (5 kPa O2 and 15 kPa CO2) on suppressing spoilage. The inhibitory

effect of both treatments, as compared to air-packed at 8 °C, is only expressed

during shelf life. Spoilage increases with temperature (Fig. 3.6). Although the

inhibitive effect of MA conditions is evident, the differences established after 5 d

of packaging (Fig. 3.6A) were largely cancelled out during the subsequent shelf

life period at ambient conditions (Fig. 3.6B).

0

10

20

30

40

50A. B.

0

1

2

3

4

5

sp

oil

ag

e (

%)

storage conditions

air8°C

CA8°C

air1°C

air8°C

air1°C

CA8°C

Fig. 3.5 Spoilage of air-packed ‘Elsanta’ strawberries at 8 °C versus CA-packed

strawberries (15 kPa CO2, 5 kPa O2) at 8 °C and air-packed strawberries at

1 °C. Spoilage was determined after 6 d packaging (A) and after 5 d of

subsequent shelf life in air (2 d at 8 °C followed by 3 d at 12 °C; B).

Pathological breakdown is supposed to be inhibited by CO2 (Kader, 1986).

Smith (1992) reported a linear effect of CO2 on spoilage of ‘Redcoat’

strawberries from 1.72 % decay at 0 kPa CO2 to 0.87 % decay at 18 kPa CO2.

Such a difference could not be detected in our experiments with mainly consumer

size packages, as one affected strawberry already accounted for 2-4 %. Couey

and Wells (1970) only found a 2.5 % reduction of decay (3.5 % in normal air as

compared to 1 % decay at 30 kPa CO2, both after 36 h at 15 °C). Browne et al.

(1984) could not establish a clear effect of CO2 at levels up to 10 kPa. Ke et al.

(1991) studied the effect of higher CO2 levels (20 kPa to 80 kPa CO2) and found

a clear inhibition of the spoilage of ‘Selva’ strawberries. At these extreme high

levels of CO2, fungal growth was inhibited. This agrees with results of Agar et al.


(1990) who showed that growth of Botrytis on potato dextrose agar (expressed as

infected area) was very resistant to CO2 treatments up to 20 kPa.

4 10 16

0

20

40

60

80

sp

oil

ag

e (

%)

4 10 16

T (°C)

0

20

40

60

80

A.

B.

Fig. 3.6 Spoilage of air-packed (open bars) versus MA-packed (filled bars)

‘Elsanta’ strawberries as a function of temperature. Spoilage was determined 5 d

after packaging (A) and after 2 d of subsequent shelf life at 10 °C in air (B). The

bars represent averaged values ± the standard error of the mean based on four

packages per treatment each containing about 50 strawberries.

Therefore, under the current experimental conditions with CO2 levels below

20 kPa, the assumption seems valid that fungal growth is not drastically inhibited

and remains much faster than the ripening process. Furthermore, when spoilage is

expressed as the number of infected strawberries instead of the area covered by

the mycelium, the dynamics of growth is replaced by only ‘growth’ or ‘no

growth’ of the fungus. According to our concept, this largely depends on the

opportunities created by the fruit. The CO2 effect on both respiration and spoilage

of ‘Pajaro’ strawberries, as observed by Colelli and Martelli (1995) is consistent

with this concept, as the CO2 effect on spoilage can probably be explained by the

CO2 effect on respiration.

Studies using inoculated fruits showed an inhibitory effect of CO2 on spoilage

at levels below 20 kPa (El-Kazzaz et al., 1983; Chambroy et al., 1993). However,

inoculation of fruits involves mechanical injury, so while growth of fungi was

studied, the biology of the hyphal penetration of intact strawberry tissue was

completely ignored.


3.3.3.3 Model analysis

Analysis of the data from all packaging experiments (370 packages coming from

a total of 17 different batches of strawberries all exposed to different MA

conditions) regarding spoilage, using the relative metabolic rate based on the gas

exchange model, is shown in Table 3.3. One example of how the model fits the

data is given in Fig. 3.4.

Temperature is affecting both the rate of spoilage as the rates of gas

exchange. One could therefore argue that gas exchange is mainly serving as an

indicator of the temperature effect. However, as the experimental data covered

different MA conditions for each given temperature, the model was able to

discriminate between the main temperature effect directly on the rate of spoilage

and the rate gas exchange and the additional effect of MA on gas exchange.

To account for differences between batches, the initial spoilage (N0) was

estimated for each of the seventeen batches involved. The initial spoilage N0 is a

value representing the initial ripening stage or sensitivity of strawberries for

Botrytis infection. However, the differences in N0 can not be assessed visually as

even the highest value of 3.25 % is less then one affected strawberry in a

consumer package of 25 fruits. Eighty percent of the batches had an initial

spoilage less then 1 % (Table 3.3).

Table 3.3 Parameter estimates and their standard error (s.e.) resulting from non

linear regression analysis of the spoilage of ‘Elsanta’ strawberries

( refT = 10 °C).

Parameter (unit) Value s.e. Parameter (unit) Value s.e. refdk (d-1)

dkEa (kJÖmol-1)

N0,1 (%)

N0,2 (%)

N0,3 (%)

N0,4 (%)

N0,5 (%)

N0,6 (%)

N0,7 (%)

N0,8 (%)

0.60

70.1

2.58

0.29

0.43

0.83

0.11

0.41

0.96

3.22

0.045

7.1

1.09

0.03

0.05

0.10

0.03

0.06

0.11

0.29

N0,9 (%)

N0,10 (%)

N0,11 (%)

N0,12 (%)

N0,13 (%)

N0,14 (%)

N0,15 (%)

N0,16 (%)

N0,17 (%)

1.15

0.43

0.08

0.09

0.19

0.16

0.21

0.59

0.50

0.11

0.09

0.03

0.03

0.04

0.04

0.04

0.19

0.19

R2adj 83 % n 630


3.3.3.4 Simulated keeping quality

Using the estimated parameters, spoilage and the related keeping quality can be

studied on a more theoretical basis. Fig. 3.7 shows the effect of both temperature

and initial spoilage on the course of spoilage, for air-packed strawberries.

Keeping quality can be defined as the time spoilage stays below a certain limit of

acceptance. In this case a limit of 5 % is applied (about 1 affected strawberry per

consumer size pack).

0.0 5.0 10.0 15.0

time (d)

0

20

40

60

80

100

sp

oil

ag

e (

%)

limit ofacceptance

4°C

16°C12°C

8°C

Fig. 3.7 Spoilage of ‘Elsanta’ strawberries as a function of time and temperature

as predicted for two different batches of ‘Elsanta’ strawberries (solid line:

N0 = 0.72 %; dotted line: N0 = 2 %). The limit of acceptance was chosen at 5 %

(about 1 affected strawberry per consumer size pack).

As temperature increases the spoilage rate kd increases and the limit of

acceptance is exceeded earlier. As a consequence, keeping quality is reduced. For

poor quality strawberries (N0 = 2 % as compared to N0 = 0.72 %) keeping quality

is even shorter.

The temperature dependence of keeping quality resulting from the simulated

data in Fig. 3.7 is presented in Fig. 3.8. The average quality of ‘Elsanta’

strawberries (N0 = 0.72 %) is substantially better as compared to data from

literature on older cultivars (Sprenger Instituut, 1980). The data from Sprenger

agrees more with a batch of poor quality ‘Elsanta’ strawberries (N0 = 2 %).


0 5 10 15 20 25

T (°C)

0

5

10

KQ

(d

)N

0= 0.72

N0= 2.00

Fig. 3.8 Keeping quality as a function of temperature. Data for two different

batches of ‘Elsanta’ strawberries (solid lines; N0 = 0.72 % and N0 = 2 %) was

compared to data from the literature (dotted line; Sprenger Instituut, 1980).

3.3.3.5 Chain simulation

The models presented here were all incorporated in an extended simulation

model of MA packages as described in chapter 2. This integrated implementation

can be used to simulate and predict product behaviour throughout a logistic

chain. Subsequently, the chain can be optimised based on the predicted product

behaviour. A typical chain for ‘Elsanta’ strawberries is described as chain A in

Table 3.4 and is compared to a closed cooling chain (chain B). It was assumed

that the fruit was picked warm. In Fig. 3.9, simulation results are presented for

both chains, for both MA and RA packed strawberries.

Table 3.4 Temperature conditions used for the simulations presented in Fig. 3.9.

Temperature (°C)

Location Duration (h) Chain A Chain B

grower

auction

wholesale

transport

retail

24

16

4

20

until end of shelf life

12

4

25

10

16

4

4

4

4

4


Keeping quality remaining for the consumer was calculated assuming that the

consumer stores the opened packages at a standard temperature of 10°C. In chain

A, the application of MA only added one and a half day of keeping quality

compared to air-packed strawberries. By optimizing temperature control

throughout the logistic chain a more substantial increase in keeping quality could

be achieved regardless of the type of package. Given the high variation between

batches of strawberries, one can not make accurate predictions about strawberries

in general. In the ideal situation, the conditions of each transport should be

adapted to optimise keeping quality of the specific batch of strawberries

concerned.

0 1 2 3 4 5 6 7 8

0

1

2

3 chain A, MA packed

chain A, RA packed

chain B, MA packed

chain B, RA packed

KQ

(d)

time (d)

0

5

10

15 chain A, p

O2

chain A, pCO

2

chain B, pO

2

chain B, pCO

2

pO

2

,p

CO

2

(kP

a) 0

5

10

15

20

25

chain A, air temperature

chain A, product temperature

chain B, product temperature

T(°

C)

Fig. 3.9 The predicted loss of keeping quality of ‘Elsanta’ strawberries over time

in two different logistic chains (Table 3.4) for both MA and RA packed

strawberries (N0 = 0.72 %). Keeping quality available for the end user was

predicted assuming standard conditions (10 °C, in air) and a critical level of

Nlim = 5 %.


3.3.4 Conclusions

A set of dynamic models was used to analyse a large set of data on spoilage of

‘Elsanta’ strawberries by Botrytis. The effect of gas conditions on spoilage was

interpreted using the concept that Botrytis acts in an opportunistic way, meaning

that fruit softening is the primary event enabling Botrytis infection. At the same

time a one-to-one relationship was assumed between the rate of gas exchange as

affected by MA conditions and the rate of spoilage. Although the pathogen-host

interaction is likely to be more complex, the applied approach led to a useful

working model. The integrated model was able to discriminate between the

effects of O2, CO2, temperature and time. The differences between batches of

strawberries were traced back to different initial qualities as expressed by N0.

When N0 can be defined in terms of physiological characteristics of strawberries,

batches of strawberries can be classified at an early stage, based on their expected

keeping quality. Subsequently, package and transport conditions can be

optimised by adapting them to the specific product demands.

3.4 Stem growth of Belgian endive

Witloof chicory, also called Belgian endive (Cichorium intybus L.), is a small,

cylindrical head of pale, tightly packed leaves. They are forced from roots that

have been kept in darkness and warmth so that no chlorophyll develops

(Coppenolle et al, 2001). Consumers are generally looking for tight chicory

heads. Postharvest growth of the central stem loosens the heads enhancing

evaporation from the now exposed leaves. The European market has decided that

stem length should not exceed 75 % of the crop length which has been

incorporated in the agricultural standards formulated by the UNECE (United

Nations Economic Commission for Europe) and ratified under the scheme for the

application of international standards for fruit and vegetables of the OECD

(Organisation for Economic Co-operation and Development). Within the

framework of the Belgian ‘Flandria’ label a critical stem length of 50 % is used.

Recently, Vanstreels et al. (2002) presented data on the effect of MA on the

change in quality of chicory heads. Vanstreels et al. (2002) focused on red

discoloration but they also collected an extensive set of destructive data on stem

growth. Red discoloration and stem growth are believed to be closely related.

Gillis et al. (2001) postulated that the growing stem induces mechanical stresses


in the intermediate leaves, causing cell rupture. This might result in the formation

of red coloured polyphenolic compounds.

For now, focus will be on characterising the growth of the central stem of

chicory heads as a function of the MA conditions applied, by linking the rate of

growth to the rate of gas exchange. The gas exchange of chicory has already been

extensively characterised as part of a larger EU project (Hertog et al., 1998). For

this study it was assumed that gas exchange of the current chicory batches

behaved comparably.


The experimental setup was described in full detail by Vanstreels et al. (2002)

with the relevant information summarised below.

3.4.1.1 Produce

Three experiments (August 1999, August 2000, January 2001) were conducted

on chicory heads stored under MA. Chicory (C. intybus L., cv. ‘Tabor’) was

grown hydroponically by a single grower. After harvest they were transported to

the experimental controlled atmosphere facility of the VCBT (K.U. Leuven,

Belgium).


Twelve different gas conditions were generated with O2 ranging from 2 kPa to

21 kPa and CO2 ranging from 0 kPa to 19 kPa. In addition to the different gas

conditions, a number of different storage temperatures were evaluated ranging

from 0 °C to 20 °C.


Stem length and mass of the chicory heads were evaluated six times over a period

of 3 weeks taking samples of 20 to 80 chicory heads per storage condition. Each

chicory head was weighed, then cut in halves and the length of the central stem

was measured with a ruler with an accuracy of 0.5 cm. Extensively rotten chicory

heads were excluded from the evaluations. Stem length was thus measured

destructively.



A simple model was applied assuming postharvest stem growth to be the result of

cell division followed by cell elongation where cell material is reallocated from

the leaves to the stem. As the chicory heads are isolated systems, detached from

their natural resources of carbohydrates, minerals and water, and stored in the

dark, overall mass of the heads will not be able to increase anymore.

Under the experimental conditions applied of up to 14 d storage at

temperatures between 0 °C and 20 °C, based on literature values for the rate of

gas exchange of chicory (Hertog et al., 1998), respiration losses were less than

0.5 % of the initial mass. It was assumed that because of the high RH levels in

MA, moisture losses would be limited to negligible levels as well. Therefore

water and respiration losses were neglected, assuming the overall biomass of the

chicory heads (Mhead) to be constant. The mechanism of reallocation of biomass

from the leaves (Mleaves) to the stem (Mstem) was thus simplified and represented

by the following scheme:

stemleaves MM gk½½ (3.6)

Assuming stem growth is an energy demanding process, the rate constant (kg)

was linked to the metabolic rate using the relative respiration rate of chicory.

Approaching the stem as a cylinder, stem length can be calculated according:

( )

RAO

MAO

2stemstem

2stemstem0stem,headhead

stem

2

2

r

rRR

rˊɟ

rˊɟlMe-Ml

tRR-kg

=

ÖÖ

ÖÖÖ-Ö=

ÖÖ

(3.7)

The relative respiration rate of chicory was based on the modelled gas

exchange data from Hertog et al. (1998). Using the approach outlined in chapter

2, Eq. 2.1 was applied to describe O2 consumption and Eq. 2.10 to describe CO2

production, combined with a temperature dependency applied to both maxO2

r andmax

f)(CO2r according Arrhenius’ law (Eq. 2.12). The parameter values used to

describe gas exchange were taken directly from Hertog et al. (1998; Table 3.5).

Given the experimental results on stem length showing a complete inhibition

of growth at 0 kPa O2 the relative respiration rate was expressed in terms of O2

consumption rate, thus excluding energy produced by fermentation.


Table 3.5 Parameters describing gas exchange of Belgian Endive withrefT = 10 °C (after Hertog et al., 1998)

Parameter (unit) Value Parameter (unit) Value refmax,

O2r (µmolÖkg-1Ös-1)

refmax,

2Or

Ea (kJÖmol-1)

2OKm (kPa)

2COKmc (kPa)

2COKmu (kPa)

RQox (-)

0.112

67.1

2.70

+¤+¤0.90

refmax,

f)(CO2r (µmolÖkg-1Ös-1)

max,refCO (f)2

rEa (kJÖmol-1)

2O (f)Kmc (kPa)

2CO (f)Kmc (kPa)

2CO (f)Km (kPa)

0.130

71.6

0.541

+¤1


The model from Eq. 3.7 was used to analyse data from 3 seasons on storage of

chicory heads at a range of temperatures and MA conditions. The diameter of an

average chicory stem was set to 1.5 cm and the density of the stem was set to

1013 kgÖm-3 (Sprenger Institute, 1986). The only parameters estimated where the

Arrhenius parameters to define kg and the average initial stem length at harvest

for each of the measured batches (Table 3.6). Under the assumption of a constant

mass, the measured initial mass of each of the individual chicory heads was taken

as the value for Mhead when analysing the data.

Table 3.6 Parameter estimates and their standard errors (s.e.) resulting from the

non linear regression analysis of stem growth of Belgian Endive as a function of

O2, CO2 and temperature with refT =10 °C.

Parameter (unit) Value (s.e.) Parameter (unit) Value (s.e.) refgk (d-1)

gkEa (kJÖmol-1)

0.00132 (0.00004)

86.7 (2.6)

1999stem,0l (m)2000stem,0l (m)2001stem,0l (m)

0.064 (0.0004)

0.067 (0.0002)

0.059 (0.0005)

n 14095 R2adj 51 %

Because of the multitude of data analysed coming from three different

experiments and a wide variety of MA conditions it would go too far to show the

model fits for the different MA conditions. The overall model fit is represented

by Fig. 3.10 comparing the averaged measured stem length per sample of 20 to

80 chicory heads versus the modelled stem length.

Within a batch of chicory heads, variation in both head size, mass and stem

length exists resulting in the large scattering observed in Fig. 3.10. Because of

the destructive nature of the measurement the measured stem length of a single


chicory head after storage at certain MA conditions cannot be estimated based on

its own initial stem length but only based on some batch averaged initial stem

length. This contributed further to the large scattering observed in Fig. 3.10

resulting in the low R2adj of 51 %. In spite of the overall low R2

adj the model

parameters are relatively well defined given the small standard errors (Table 3.6).

Fig. 3.10 Measured versus modelled

stem length (in mm) averaged per

sample of 20 to 80 destructive

measurements.

50 60 70 80 90 100 110 12050

60

70

80

90

100

110

120

mod

elle

d

ste

m len

gth

(m

m)

measured

stem length (mm)

Fig. 3.11 The effect of O2 and CO2 on

gas exchange rate of chicory at 21 °C;

O2 consumption (2Or , ½ ) and CO2

production (2COr ,---). Redrawn after

Hertog et al., 1998.


Besides the obvious temperature effect expressed bygkEa (Table 3.6) there

was a clear effect of O2 on stem growth, almost completely inhibiting stem

growth at 2 kPa O2. Stem growth was not affected by CO2 levels up to 19 kPa.

This was in agreement with the lack of any effect of CO2 on the rates of gas

exchange previously observed by Hertog et al. (1998; Fig. 3.11).

Based on the gas exchange parameters (Table 3.5) and the estimated model

parameters describing stem growth (Table 3.6) the model from Eq. 3.7 could be

used to simulate the averaged batch behaviour of the 1999 batch of Witloof

chicory (Fig. 3.12). This simulation clearly shows that temperature is the most

important factor in minimising stem growth with MA having only a marginal

additional effect at 1 °C.

20 °C

12 °C

1 °C

0

5

10

15

200

5

10

1520

60

70

80

90

100

110

120

130

p O2

(kPa)

ste

mle

ngth

(mm

)

time

(d)

Fig. 3.12 Modelled stem length (in mm) of chicory cv ‘Tabor’ as function of

time, O2 and temperature according to Eq. 3.7. The three planes represent stem

length at respectively 1 °C, 12 °C and 20 °C.

3.4.4 Conclusions

This case study has shown how the effects of MA on the rates of quality change

of Belgian endive was incorporated, coupling independent gas exchange data to a

simple stem growth model. The main assumption was that gas exchange is

providing the basic driving force for such a primary process as stem growth. The

MA conditions in terms of the O2 level affects stem growth through energy

provided by the gas exchange while temperature affects stem growth through its


effect on both the gas exchange rate and the growth rate itself. This explains why

temperature is more efficient in inhibiting quality decay processes than O2.

This case study on Belgian endive also touches on the limitation imposed by

destructive measurement techniques in the case of large biological variation. To

improve the model and its predictive value, non-destructive data is needed to

allow monitoring of single chicory heads during time separating measurement

error from biological variation. This will enable the development of improved

mechanistic models incorporating biological variation (Chapter 4).

3.5 Softening of kiwifruit

The softening rate of kiwifruit (Actinidia deliciosa (A Chev) Liang et Ferguson

cv ‘Hayward’) is affected by time, temperature, exogenous ethylene levels and

maturity of the fruit (MacRae et al., 1989; Ritenour et al., 1999). The effect of

endogenic ethylene on softening can be neglected as kiwifruit only starts to

produce ethylene by the time the fruit has completely softened to below 10 N

(Ritenour et al., 1999).

Kiwifruit softening follows a triphasic curve with different enzymes

responsible during the subsequent softening phases (MacRae et al., 1990;

Wegrzyn and MacRae, 1992; Redgwell and Fry, 1993; Bonghi et al., 1997). For

a mature main harvest crop this softening curve is reduced to a biphasic curve

(MacRae et al., 1989) that can be described using an exponential decay function

(Benge et al., 2000). MA is known to retard the rate of kiwifruit softening as a

result of both increased levels of CO2 and decreased levels of O2 (Harman and

McDonald, 1983; Manolopoulou et al., 1997). However, the effects of MA on

gas exchange rates of kiwifruit, have not been studied extensively.

This study focuses on characterising the gas exchange and softening of

kiwifruit as a function of the MA conditions applied and studies the quantitative

relationship between the two. Such a quantitative relationship enables the

prediction of quality changes during complex postharvest chains incorporating

the combined effects of O2, CO2 and temperature on the softening of kiwifruit.



The experimental data originate from experiments executed at Fresh

Technologies, Massey University (Palmerston North, New Zealand; Hertog et al.,

2004).

3.5.1.1 Fruit

Export quality 'Kiwistart' fruit (Actinidia deliciosa (A Chev) Liang et Ferguson

cv ‘Hayward’) were obtained from Te Puke, New Zealand. One batch of fruit

was harvested on 15 May 2000, another batch on 14 May 2001. After harvest,

fruit were graded, packed and couriered overnight to Massey University. All fruit

used were count size 33. On arrival, fruit were randomised, individually labelled,

weighed and initial fruit firmness measurements were taken. Samples of 30 fruit

were assigned to each of the storage treatments.


During the 2000 season fruit were stored at 2 °C, 5 °C or 10 °C, while during the

2001 season fruit were also stored at 0 °C. The duration of storage varied per

temperature. During the 2000 season fruit were stored for either 42 d (2 °C), 37 d

(5 °C) or 35 d (10 °C). During the 2001 season fruit were stored for either 45 d

(0 °C), 44 d (2 °C), 42 d (5 °C) or 36 d (10 °C).

For each storage temperature sixteen PVC containers (with a volume of

0.0135 m3 each) were packed with 30 fruit each (average fruit weight of 105 g)

using one container per MA condition. The different MA gas mixtures were

generated by mixing flows of dry air, O2–free N2 and food grade CO2 (BOC,

Palmerston North, NZ), to create combinations of roughly 8 different O2 levels

(0 kPa to 21 kPa) and 3 different CO2 levels (0 kPa to 5 kPa; Table 3.7). Before

entering the PVC containers, the gas mixtures were humidified by bubbling

through jars with water resulting in ca. 98 % RH. The flow rate was controlled at

0.1 LÖmin-1 to 0.3 LÖmin-1 depending on the temperature applied.

The MA conditions were held constant throughout the whole time span of the

experiment. Gas conditions inside the containers were checked regularly. The

CO2 levels remained constant over time with an average standard error of

0.13 kPa CO2, and the O2 levels stayed constant with an average standard error of

0.16 kPa O2.


At the same time gas conditions inside the containers were checked by

removing a sample using 100 µl glass syringes, respiration rates of the contained

fruit were measured by temporarily closing the tubes to allow accumulation of

CO2 and depletion of O2 by about 0.5 kPa. Depending on the MA conditions, this

took between 1 h to 5 h.

Table 3.7 Matrix of modified atmospheres targeted at the different storage

temperatures. Every O2 level was combined with two levels of CO2; 0 kPa CO2

and depending on temperature either 2 kPa or 5 kPa CO2. The same O2 ³ CO2

combinations were applied to 0 °C and 5 °C and to 2 °C and 10 °C storage.

During the 2000 season, only 2 °C, 5 °C and 10 °C storage was used while for

the 2001 season 0 °C storage was included as well.

2Op (kPa) 2COp (kPa)

0 °C and 5 °C storage 2 °C and 10 °C storage

0

0.25

0.5

1

3

5

10

21

0 and 2

0 and 5

0 and 2

0 and 5

0 and 2

0 and 5

0 and 2

0 and 5

0 and 5

0 and 2

0 and 5

0 and 2

0 and 5

0 and 2

0 and 5

0 and 2


At the start of the experiment, destructive firmness readings were taken on a

separate batch of 30 fruit. At the end of the MA treatment firmness of all fruit

was destructively measured. Destructive firmness readings were taken using the

standard penetrometer cylinder probe (7.9 mm diameter) mounted on a TA-XT2

texture analyser (Stable Micro Systems Ltd.). A piece of skin about 2 mm thick

was removed using a cutting device with a fixed blade. The test was run using a

pre-test and a test speed of both 10 mmÖs-1, a trigger force of 15 g, and allowing

the probe to travel 9 mm deep into the tissue, measuring the maximum force (in

N) encountered.

3.5.1.4 Gas analysis

All gas samples were analysed using an O2 electrode (Citicell C/S type, City

Technology Ltd., London, UK) in series with a miniature infrared CO2 transducer


(Analytical Development Company, Hoddesdon, UK), with O2-free N2 as carrier

gas (flow rate 35 mLÖmin-1). Output signals were linear over the range applied

and analysed using HP integrators (Hewlett Packard, model 3396A).

Commercially prepared standards were used for calibrating the gas analysers. All

samples were collected in duplicate through the two sampling ports of the

containers. When duplicates differed by more than 0.1 kPa, new samples were

taken and the system was checked for possible errors until consistent results

could be obtained. Standard gasses were routinely used to check for possible drift

in the signal.

3.5.1.5 Data analysis

All data collected was expressed according to the units proposed by Banks et al.

(1995). Data was analysed statistically with the iterative non linear regression

routine of Statistical Analysis System (SAS software, version 6.11, SAS institute

Inc., Cary, NC, USA). The non linear equations were applied directly, without

transformation to data or equations. During the analyses the averaged measured

gas conditions were used, not the targeted values from Table 3.7.



Gas exchange data from both seasons was analysed separately using Michaelis

Menten type gas exchange models. In a first approach, data was analysed using a

model formulation incorporating a combined inhibition of O2 consumption by

CO2 (Eqs. 2.1 and 2.10) applying a temperature dependency to maxO2

r and max

f)(CO2r according Arrhenius (Eq. 2.12). However, no significant effect of CO2 on

gas exchange rates could be determined. As a result, the model was reduced to a

simple Michaelis Menten model by fixing2COKmc ,

2COKmu and )f(CO2Kmc to

+¤. In accordance with chapter 2, )f(CO2Km was fixed at 1 kPa.

3.5.2.2 Fruit firmness

To describe firmness (F) as a function of storage time (t) the approach of Benge

et al. (2000) was adopted. The exponential decay used by Benge et al. was

rewritten as:

( ) tkeFFFtF s Ö-Ö-+= fix0fix)( (3.8)


with ks being the rate of softening; Ffix the asymptotic firmness value at plus

infinite time and F0 the initial firmness at harvest. Time (t) was set to the

different storage times used with the different storage temperatures as indicated

in materials and methods. From a kinetic point of view this expression describes

how the firmness component is broken down into debris according a simple

irreversible first order conversion reaction that can be represented as:

debrisF sk½½ . In this concept, ks is a rate constant, which at constant ambient

conditions will remain constant. The apparent rate of softening, given by

sktF Ö)( , will vary with time due to the changing firmness.

To incorporate the effect of MA on the rate of softening a comparable

approach was taken as for apple and avocado (Hertog et al., 2001; 2003) by

describing the rate constant ks as a function of O2, CO2 and temperature using a

Michaelis Menten type model including inhibition by CO2 (Peppelenbos and

Van ‘t Leven, 1996; Hertog et al., 1998). As at 0 kPa O2 softening was not

completely inhibited, it was assumed that the rate of softening was driven by

energy provided by both oxidative and fermentative processes as was observed

for apple (Hertog et al., 2001). From the initial analysis it became clear that the

uncompetitive inhibition was favoured to the competitive type of inhibition.

Therefore the following uncompetitive model was selected (based on Eqs. 2.1

and 2.10) to describe the effect of MA conditions on the rate of softening:

)/1(/22222

2

22 COCOOO

Omax

)f(OO

max

KmuppKm

pk

Kmcp

kk

Of

s +Ö+

Ö+

+= (3.9)

where maxfk and max

Ok are the maximum rates of softening (both in d-1)

unconstrained by O2 or CO2 related to respectively fermentative and oxidative

processes;2COKmu (in kPa) is the Michaelis Menten constant for the inhibition of

softening by the applied levels of CO2 while the other parameters are defined in

analogy to the gas exchange model. The rate constants maxfk and max

Ok are assumed

to depend on temperature according Arrhenius' law (Eq. 2.12).

3.5.2.3 Integrated approach

To further explore the suggested functional relationship between gas exchange

and fruit softening the rate of softening can be explicitly linked to gas exchange

according:

)/1(/22222

22

22

2

COCOOO

Omax

Omax

(f)OO

max)f(CO

max

KmuppKm

prk

Kmcp

rkk

Of

s +Ö+

ÖÖ+

+

Ö= (3.10)


While in the previous approach (Eqs. 3.8 and 3.9) a Michaelis Menten type

model was used to describe the rate of softening estimating all parameter values

solely based on firmness data, the integrated approach estimates the

corresponding parameters based on the information from the gas exchange data

using the gas exchange model from Eqs. 2.1 and 2.10 and subsequently transfers

these parameter values into the softening model (Eqs. 3.8 and 3.10) estimating

the remaining softening related parameters using the firmness data.

3.5.2.4 Keeping quality predictions

Using the quantitative relationship between rates of gas exchange and the rates of

firmness loss, predictions can be made on the keeping quality of kiwifruit as a

function of the applied MA conditions. Based on Eq. 3.8 the time needed to reach

certain critical limit firmness (Flim) level (KQ) can be calculated as:

skFF

FF-KQ öö

÷

õææç

å--

=lim0

fixlimln (3.11)

Substituting Eq. 3.10 into Eq. 3.11 results in a versatile expression to calculate

the shelf life of kiwifruit given all the other parameters are known.



Decreasing levels of O2 inhibited O2 consumption rates (2Or ) while CO2

production (2COr ) at low O2 levels started to increase again as a result of

fermentation (Fig. 3.13).

Temperature had a relatively small effect on both oxidative O2 consumption

and fermentative CO2 production. However, the observed temperature effect was

in agreement with the respiration rates generally cited for kiwifruit stored under

normal air (Harris and McDonald, 1975). The applied CO2 levels (0 kPa to

5 kPa) did not affect the rates of gas exchange (data not shown).

Analysing the gas exchange data per season showed that the model could

explain most of the observed variation (93 % for 2000 data and 86 % for 2001

data) and also showed that the model parameters were not significantly different

between the two seasons (Fig. 3.14). This implied that gas exchange was the

same for the two seasons and therefore could be combined and analysed as one


set of data. This resulted in the final parameter estimates given in Table 3.8

(column 1) and the simulated model values represented by the lines in Fig. 3.13.

Fig. 3.13 O2 consumption (2Or ,ǒ)

and CO2 production (2COr ,ƺ) rate of

‘Hayward’ kiwifruit (both in

nmolÖs-1Ökg-1) stored at constant

temperatures ranging from 0 °C to

10 °C as a function of the O2 level

applied. The duplicate data of both

seasons were averaged. The lines

represent the model results using the

parameter estimates from Table 3.8. 0 5 10 15 20

0

20

40

60

80 0 °C

pO

2

(kPa)

0

20

40

60

80 2 °C

r O2

,r C

O2

(n

mo

l.s

-1.k

g-1)

0

20

40

60

80 5 °C0

20

40

60

80

100

10 °C

Fig. 3.14 Parameter estimates for the gas exchange model and their 95 %

confidence intervals as estimated on the separate data from the two seasons. refT = 5 °C (278.15 K).


The rate of oxidative O2 consumption ( maxO2

r ) was in good agreement with

literature data (Harris and McDonald, 1975) although the energy of activation

( max

2OrEa ) was small as compared to the data from Crisosto and Kader (1999) or

Feng et al. (2003). The reason for this is that the current experiments only

covered temperatures up to 10 °C while Crisosto and Kader (1999) and Feng et

al. (2003) included data up to 25 °C. The effect of temperature largely becomes

visible at these higher temperatures affecting the estimation of max

2OrEa . The

estimated value of2OKm was comparable to values found for strawberry (Hertog

et al., 1999a), apple (Hertog et al., 1998) and blueberry (Cameron et al., 1994).

Generally the RQox of fruit stored at aerobic conditions is between 0.8 and 1.2

(Beaudry et al., 1992; Cameron et al., 1994; Peppelenbos et al., 1996; Yearsley

et al., 1997b; Hertog et al., 1998, 1999, 2001, 2003; Petracek et al., 2002) while

for the current data RQox was estimated to be 0.5. The raw data on RQ (Fig. 3.15)

showed quit some scattering although no systematic influence could be found of

temperature or CO2. At lower O2 levels (5 and 10 kPa) the RQ was closer to a

ratio of one as would be expected.

No literature data on kiwifruit exists to compare the fermentation parameters

to. However, in general one can state that, if alcoholic fermentation occurs, the

maximum rate ( refmax,

)f(CO2r ) and the energy of activation ( max

CO (f)2r

Ea ) are of the same

order as the values found for oxidative respiration. (Peppelenbos et al., 1996;

Peppelenbos and Van ‘t Leven, 1996; Hertog et al., 1998, 1999, 2001). This was

also the case for kiwifruit (Table 3.8).

Fig. 3.15 Respiratory quotient of

kiwifruit as a function of storage O2

levels.

0 5 10 15 200

2

4

6

RQ

ox

pO

2

(kPa)

The value of2O (f)Kmc , indicating the inhibition of fermentation by O2,

generally is below 1 kPa (Peppelenbos et al., 1996; Hertog et al., 1998, 1999,

2001), except for tomato where it was estimated at 1.4 kPa (Hertog et al., 1998).

For kiwifruit2O (f)Kmc showed a value of 0.85 kPa (Table 3.8). The relative high


value for tomato might be explained from the relative dense tissue of tomato

(1034 kgÖm-3; Antunes et al., 1995) in combination with the impermeable skin.

Table 3.8 Parameter estimates and their standard errors (s.e.) resulting from the

non linear regression analyses of gas exchange rates and firmness of ‘Hayward’

kiwifruit as a function of O2, CO2 and temperature. The data of 2000 and 2001

were analysed simultaneously.

Gas exchange

data only

Firmness data

only

Integrated

approach

Parameter a) (unit) estimate (s.e.) estimate (s.e.) estimate (s.e.)

Parameters related to oxidative processes

refmax,

O2r (nmolÖkg-1Ös-1) 74 (5.5) - 74 (1.5)

max

2OrEa (kJÖmol-1) 30.9 (6.0) - 31.0 (1.6)

refmax,Ok (d-1) - 0.12 (0.03) 0.0010 (0.0002)

maxOk

Ea (kJÖmol-1) - 213.8 (19.8) 155.4 (20.7)

2OKm (kPa) 3.3 (0.7) 8.2 (4.7) 3.3 (0.2)

2COKmu (kPa) - 2.4 (1.3) 1.7 (0.2)

Parameters related to fermentative processes

RQox (-) 0.5 (0.1) - 0.5 (0.1)

2

max,refCO (f)

r (nmolÖkg-1Ös-1) 104 (0.6) - 104 (1.8)

maxCO (f)2

rEa (kJÖmol-1) 38.3 (7.8) - 38.3 (2.0)

refmax,fk (d-1) - 0.014 (0.002) 0.00022 (0.00006)

max

fk

Ea (kJÖmol-1) - 135.4 (11.3) 120.0 (37.3)

2O (f)Kmc (kPa) 0.85 (0.3) 2.6 (9) 2.2 (1.0)

Batch specific parameters

F0, 2000b) (N) - 63 (14) 63 (14)

Ffix, 2000 (N) - 27.1 (0.7) 26.8 (2.1)

F0, 2001b) (N) - 54 (13) 54 (13)

Ffix, 2001 (N) - 6.3 (0.8) 5.8 (2.3)

n 225 3360 3585

R2adj 88 % 88 % 88 %

a)2COKmc ,

2COKmu and )f(CO2Kmc were fixed to +¤. )f(CO2

Km was fixed at 1 kPa;refT = 5 °C.

b) fixed values based on experimental data, not estimated.

With increasing density of the tissue and decreasing permeance of the skin,

the resistance of the pathway of gas exchange will increase, resulting in a further


modification of the internal gas composition as compared to the surrounding gas

conditions. As a consequence, the apparent2O (f)Kmc of 1.4 kPa might in fact

relate to an internal O2 level of far below 1 kPa. The denser the fruit the bigger

this discrepancy between the apparent2O (f)Kmc based on external gas conditions

and the actual2O (f)Kmc taking into account the internal gas conditions.

3.5.3.2 Fruit firmness

In spite of the lack of a clear CO2 effect on the rate of gas exchange, softening

did show a noticeable effect of CO2 (Fig. 3.16). Especially at the lower

temperatures (0 °C and 2 °C) fruit stored at 5 kPa CO2 were up to 15 N firmer

then the fruit stored at 0 kPa. At 10 °C, no difference was observed between the

different CO2 levels. This was due to the fact that after 36 d of storage at 10 °C,

for most of the MA conditions, fruit had already softened to their final firmness

level of around 6 N. However, this should not be interpreted as CO2 not having

any effect during 10 °C storage. If we had measured earlier in time differences

between the different CO2 levels would have become obvious.

0

20

40

600 °C 2 °C 5 °C 0

kP

a C

O2

10 °C

0

20

40

60

firm

ne

ss (

N)

2 k

Pa C

O2

0 5 10 15 20

0

20

40

60

0 5 10 15 20

pO

2

(kPa)

0 5 10 15 20 0 5 10 15 20

5 k

Pa

CO

2

Fig. 3.16 Final firmness of ‘Hayward’ kiwifruit (in N) stored during the 2001

season for either 45 d (0 °C) 44 d (2 °C), 42 d (5 °C) or 36 d (10 °C) as a

function of the O2 and CO2 levels applied. The bars represent the standard

deviation of the 30 fruit stored at each treatment. The lines represent the model

from the final integrated model approach.


Firmness also showed a clear effect of O2 with fruit retaining their firmness

better at lower O2 levels. When interpreting Fig. 3.16 one needs to realise this is

representing just one single snapshot at a certain time (which was different for

each temperature) and is not covering the whole time course.

Firmness data of both seasons was analysed together assuming that the kinetic

parameters were the same for the two seasons. The differences between the two

seasons would be caused by differences in the initial firmness levels F0 or to

differences in Ffix. The parameter F0 was set to the measured averaged initial

firmness (63 N for the 2000 season and 54 N for the 2001 season) while Ffix was

estimated separately for the two seasons. The model explained 88% of the

variation in the firmness data of the individual fruit (Table 3.8, column 2; Fig.

3.16) using the combined approach of a simple exponential decay (Eq. 3.8) with

a rate affected by the gas conditions using a Michaelis Menten type model (Eq.

3.9) and temperature dependencies according Arrhenius' law (Eq. 2.12).

Taking into account the standard errors, the estimates of2OKm and

2O (f)Kmc on the gas exchange data are not significantly different from the

estimates on the firmness data (Table 3.8) supporting the hypothesis that both

processes are affected by O2 to the same extent. The estimates of max

O2r

Ea and

max

CO (f)2r

Ea are significantly different from respectively max

Ok

Ea and max

fk

Ea indicating

that the effect of temperature on softening is not the same as its effect on gas

exchange. This is not surprising as also under the assumption that the rate of

softening is driven by gas exchange the enzymatic breakdown of cell wall

components has its own thermodynamic response imposed on the effect

temperature already has on gas exchange.

The effect of CO2 on firmness breakdown is reflected by the low value of

2COKmu (Table 3.8) indicating that the rate of softening related to oxidative

processes is already inhibited to half its value at a CO2 level of 2.4 kPa. If CO2

would have no significant effect on softening the value of2COKmu would tend to

infinity. The values of Ffix varied quite a lot between the two seasons but this

kind of variation has been observed before (Harman and McDonald, 1983).

The estimated value for the maximum rate of softening at aerobic conditions

(refmax,

Ok , 0.12 d-1) was about 17 times as large as the maximum softening rate at

anaerobic conditions (refmax,

fk /2 = 0.007 d-1). This coincides with the ratio of

ATP being produced under aerobic (36 ATP per glucose molecule) versus

anaerobic conditions (2 ATP per glucose molecule) which gives a ratio of 18.

Assuming the same glucose consumption under aerobic and anaerobic

conditions, this would be a strong indication that the rate of softening is directly


linked to ATP production only. However, based on the refmax,

O2r , kiwifruit is

consuming about 12 nmolÖkg-1Ös-1 of glucose at aerobic conditions at 5 °C. Taking

into account the RQox of 0.5 this would only be 6 nmolÖkg-1Ös-1. At anaerobic

conditions, based on the refmax,

)f(CO2r , about 26 nmolÖkg-1Ös-1 of glucose is being

consumed. Consequently, going from fully aerobic to fully anaerobic conditions,

the ATP production is reduced by a factor 4-8 while the rate of softening was

reduced by a factor 17. Apparently, the rate of softening was not only inhibited

by MA through the reduced ATP production but, in addition, also by MA

through other means.

Still, based on the similarities the way both processes depend on O2 one can

argue that the process of softening is most driven by the energy provided by gas

exchange even though there is an additional influence by O2 and CO2. This

influence could be a direct influence of O2 and CO2 on the enzymatic breakdown

of cell wall components.

3.5.3.3 Integrated approach

To further explore the suggested functional relationship between gas exchange

and fruit softening the combined set of data on gas exchange and fruit softening

was once more analysed together to explicitly define this relationship . Both gas

exchange and fruit softening were still described in the same way, but the rate of

softening was now directly linked to gas exchange following the integrated

approach, transferring the gas exchange model parameters directly into the

softening model (Eqs. 3.8 and 3.10) estimating only the remaining softening

related parameters using the firmness data.

The parameter estimates resulting from this integrated analysis (Table 3.8,

column 3) were in agreement with the earlier estimates on the separate data sets,

again explaining 88 % of the observed variation.

The value ofrefmax,

Ok from the integrated approach was also in agreement with

the previous analyses, as the value ofrefmax,

Ok found for the analysis of the

firmness data (Table 3.8, column 2) corresponds in the integrated approach

(Table 3.8, column 3) to the product of refmax,O2

r andrefmax,

Ok . The equivalent is the

case forrefmax,

fk and refmax,

(f)CO2r . From Table 3.8 it becomes clear that the value of

max

Ok

Ea found for the analysis of the firmness data (Table 3.8, column 2)

represented the lumped effect of temperature on the overall process, which in the

integrated approach (Table 3.8, column 3) is again separated into the temperature

effects on gas exchange ( max

2Or

Ea ) and softening ( max

Ok

Ea ). When one takes into


account the standard errors of the estimates, max

2Or

Ea and max

Ok

Ea from the integrated

approach sum to the value of max

Ok

Ea from the analysis of the firmness data only.

The equivalent is the case for max

fk

Ea and max

(f)2COr

Ea .

From the values ofrefmax,

Ok andrefmax,

fk it can be concluded that softening

related more closely to oxidative respiration than to anaerobic fermentation as

was also seen for apple (Hertog et al., 2001). This can partially be explained from

the less efficient character of fermentation in terms of the amount of energy

produced per mol of CO2 released through gas exchange. In spite of the high CO2

production during fermentation only a small amount of energy is being fixed as

ATP to drive the fruit’s metabolism.

3.5.3.4 Keeping quality predictions

Using the parameters from the integrated approach of the 2001 data (Table 3.8)

shelf life predictions were made, using the keeping quality model from Eq. 3.11

and Eq. 3.10 assuming a critical level of 20 N (Fig. 3.17).

43

21

0

05

1015

20

0

50

100

150

200

6 °C

3 °C

0 °C

pCO

2

(kPa)

KQ

(d

)

pO

2

(kPa)

Fig. 3.17 Predicted keeping quality (KQ in d) of ‘Hayward’ kiwifruit from the

2001 season as a function of temperature (the three planes) and the O2 and CO2

levels applied. Predictions are based on the keeping quality model from Eq. 3.11

in combination with Eq. 3.10 using the parameter values from the integrated

model approach (Table 3.8). The critical level for acceptance was set to a

firmness of 20 N.

To maximise shelf life with regard to firmness within the range of MA

conditions studied, temperature and O2 should be minimised while maximising


CO2. This is in agreement with the recommended storage conditions for kiwifruit

which are 1 to 2 kPa O2 in combination with 3 to 5 kPa CO2 at 0 °C (Gross et al.

2003). As can been seen from Fig. 3.17, temperature control is of the utmost

importance. Only once cold storage (0 °C) can be guaranteed, lowering O2 or

raising CO2 will give a substantial additional benefit. At the higher temperatures

(3 °C and 6 °C) the additional effect of MA is limited.

This study was focussing on the quality attribute firmness while at more

extreme conditions other quality attributes can become limiting; below 1 kPa O2

off-flavours might be induced and above 7 kPa CO2 internal breakdown of the

flesh can occur (Gross et al., 2003).

3.5.4 Conclusions

This case study has described the effects of MA on the rates of quality change of

kiwifruit. The effect of MA on both the rate of gas exchange and the rate of

quality loss was compared. Based on the similarities and differences in their

response to MA the effect of MA on quality loss was modelled through the effect

MA has on gas exchange. The main assumption was that gas exchange is

providing the basic driving force for processes occurring in living tissue

including quality loss related processes.

Even though generally no direct relationships can be found between the MA

conditions applied and quality or shelf life as such, a clear relationship was

shown between the rate of gas exchange and the rate of quality loss both affected

by MA. Depending on the product and the quality attribute under study the nature

of this relationship might vary.

3.6 Conclusions

The developed model approach to explain the effect of MA on quality changes by

linking the rates of gas exchange to the rates of quality loss has shown to be

generic. This approach could be applied to predict product quality taking into

account the combined effects of O2, CO2 and temperature.

One could argue that the rate of gas exchange is mainly serving to convey the

temperature effect. And of course, temperature is the most important factor in

postharvest and thus also in MA systems. However, in all examples a distinction

was made between the direct effect of temperature on the rates of gas exchange,


the additional temperature effect on the rate of quality decay and the additional

effect of O2 and/or CO2 on the rates of gas exchange. With that, the developed

modelling approach goes beyond the step of only using gas exchange as an

indicator of temperature.

To further improve the understanding of the behaviour of certain external

quality aspects a good understanding of the underlying product physiology is

needed. The examples outlined clearly illustrate the link between the effects of

storage conditions on metabolic rate on one hand and the effects of storage

conditions on external quality aspects on the other hand. Subsequent evidence

needs to be gathered to identify the exact relationships, whether gas conditions

affect quality-degrading processes directly through their involvement as a

reactant, or indirectly through their involvement in ATP production.

Mechanistic models are only the first step in interpreting experimental data to

determine the likelihood of the possible underlying mechanisms and to direct

future research to elucidate these mechanisms at a physiological and biochemical

level. Even though exact details are still to be unravelled, the simplified approach

of directly linking metabolic rate to the rate of quality breakdown has already

proven successful in describing the effects of modified atmosphere on external

quality attributes through their known effects on metabolic rate.

linking gas exchange to quality change - ku leuvenu0040603/thesis/ch3_linking gas exc… · ddtk °...

Documents