Storage Time and Temperature on the Sensory Properties Broccoli

Typically, broccoli arrives at the store within 7–14 days of harvest and is kept refrigerated until purchased or considered waste. To date, information has been limited on how this time on the shelf or storage temperature affects the sensory attributes that contribute to broccoli purchase or repurchase. In this study, 100 consumers performed acceptance tests and a check-all-that-apply (CATA) section to characterize sensory changes in two cultivars of broccoli (‘Diplomat’ and ‘Emerald Crown’) stored at two temperatures (0 °C and 4 °C) over five time points: 0, 14, 21, 28, and 42 days. Due to quality degradation during storage, the overall liking of broccoli decreased regardless of holding temperature and variety. This was in accordance with a decrease in sweetness and an increase in bitterness intensity. However, there were differences between varieties in which Diplomat had more sensory changes at higher temperatures and only Emerald showed negative changes to its appearance in color. Lastly, the CATA data revealed the attributes responsible for modulating the consumer acceptance of broccoli such as tastes, colors and flavors (e.g., grassy, musty, dirt-like). This information can be used to better inform shelf-life determinations of broccoli. Additionally, these changes in taste, odor, texture, and color can inform those interested in investigating the biochemical processes related to broccoli storage.


Introduction
The United States is the third largest producer of broccoli, with California producing a majority of the US production (90%). For instance, in 2010, 121,700 acres of broccoli was harvested with an estimated value of 648 billion dollars (USD) [1]. In the US, the average consumption, according to a 2017 consumer report, was 7.1 pounds per capita [2]. Commercially, broccoli is chilled to 0 • C within five hours and packed on ice for transport [3]. The broccoli arrives at the store within 7 to 14 days of harvest and is placed in a refrigerated grocery aisle until purchased or considered waste (typically around one week) [3,4]. As harvested broccoli ages, it senesces with time causing the nutritional and sensory profiles to change. Broccoli has a rich content of bioactive molecules such as vitamin C, glucosinolates, phenolic compounds, and carotenoids [5]; however, it is a highly perishable product and its nutritional, visual, and flavor profiles depend on its storage conditions. A timely harvest, cooling, controlled atmosphere, packaging and the use of less perishable cultivars are some of the tools used to extend the shelf-life of broccoli [6,7]. Many of these practices control moisture levels, reduce the rate of respiration, slow down enzymatic reactions, and decrease the production and action of ethylene. Ethylene is well known to be involved in the maturation and senescence of fruits and vegetables [8]. Several studies have extensively investigated the overall appearance and nutritional quality during

Broccoli Production
Broccoli was purchased from a local producer in East Tennessee who utilized standard production practices recommended in the Southeastern U.S. 2018 Vegetable Crop Handbook [22].

Broccoli Harvest and Storage
Two cultivars of broccoli, 'Emerald Crown' and 'Diplomat', were harvested when their head size diameter was 10 to 15 cm. Harvested broccoli was separated into two treatments: 1) iced and 2) non-iced. Iced treatments were placed in coolers in the field, immediately topped with an ice slurry made up of crushed ice that was partially melted in the field, and then transported to a cold storage room and maintained at 0 • C. Broccoli was not submerged in water during transport as water drained out the bottom of the coolers. Non-iced treatments were placed in waxed corrugated cardboard boxes for transport and then placed in a cold storage room maintained at 4 • C. Watch Dog ® data loggers (Spectrum ® Technologies, Inc., Aurora, IL, USA) were placed into one head of broccoli per storage container, and the temperature was recorded every 30 min for the duration of storage. Harvest lasted approximately two hours per cultivar with a transportation time of approximately 45 min to cold storage rooms from the field. Iced samples were maintained in iced coolers that were drained to prevent water accumulation for the duration of the study. Iced broccoli was cooled to 2.5 • C within 45 min of harvesting and placing into coolers in the field, while non-iced broccoli was cooled to 6 • C within six hours of harvesting and four and a half hours of placing in a cold storage room set at 4 • C. Subsamples of broccoli were removed from storage at 0, 14, 21, 28, and 42 days post-harvest for sensory analysis.

Data Analysis
A three-way analysis of variance (ANOVA) was used to evaluate consumer overall liking and attribute scores (hedonic and intensity) with cultivar, storage temperature, storage time, and their interactions. All ANOVA factors were considered fixed and Student t-tests were used where appropriate. Differences between conditions with CATA data were estimated using the frequency that each CATA attribute was chosen and were assessed with Cochran's Q Test and Cochran's Armitage Trend Test [23,24]. Additionally, McNemar's tests were done for CATA data between varietals at Day 0, collapsing temperature conditions. Bivariate correlations were analyzed with Pearson's r. To assess how overall liking was modulated by attributes present in the samples, the CATA descriptor counts were used to predict overall liking. A partial least squares regression model was constructed using the average overall liking scores and CATA descriptor frequencies. All data were analyzed using JMP Pro 13.0.0 (SAS Institute, Cary, NC, USA).

Consumer Acceptance
As shown in Figure 1, there was a decrease in overall liking with increasing storage time (F 4, 1980 = 7.06, p < 0.001) regardless of broccoli cultivar (F 1, 1980 = 2.35, p = 0.13) or storage temperature (F 1, 1980 = 0.06, p = 0.80). The first significant drop in overall liking, compared to fresh, was at 28 or more days of storage. However, the broccoli cultivars responded differently to the temperature at which they were stored (F 1, 1980 = 9.73, p = 0.002). 'Diplomat' was liked more when stored at lower temperatures (mean (M) = 6.45, standard deviation (SD) = 1.73) than at higher temperatures (M = 5.96, SD = 1.96; F 1, 998 = 17.59, p < 0.001), while 'Emerald Crown' was unaffected by storage temperatures (M = 6.22, SD = 1.76; F 1, 998 = 0.01, p = 0.93). The same effects and their interactions were seen for flavor liking, which was highly correlated with overall liking (r = 0.91). Texture liking showed a significant decrease with only the higher storage temperature (F 4, 995 = 4.18, p = 0.002). Aftertaste and aroma liking were also positively correlated with overall liking (r = 0.71 and r = 0.53, respectively). Liking scores for other attributes are shown in Figure 2. appropriate. Differences between conditions with CATA data were estimated using the frequency that each CATA attribute was chosen and were assessed with Cochran's Q Test and Cochran's Armitage Trend Test [23,24]. Additionally, McNemar's tests were done for CATA data between varietals at Day 0, collapsing temperature conditions. Bivariate correlations were analyzed with Pearson's r. To assess how overall liking was modulated by attributes present in the samples, the CATA descriptor counts were used to predict overall liking. A partial least squares regression model was constructed using the average overall liking scores and CATA descriptor frequencies. All data were analyzed using JMP Pro 13.0.0 (SAS Institute, Cary, NC, USA).

Consumer Acceptance
As shown in Figure 1, there was a decrease in overall liking with increasing storage time (F4, 1980 = 7.06, p < 0.001) regardless of broccoli cultivar (F1, 1980 = 2.35, p = 0.13) or storage temperature (F1, 1980 = 0.06, p = 0.80). The first significant drop in overall liking, compared to fresh, was at 28 or more days of storage. However, the broccoli cultivars responded differently to the temperature at which they were stored (F1, 1980      Concerning intensity attributes, the intensity of aroma (F 4, 1980 = 6.03, p < 0.001), and sweetness (F 4, 1980 = 21.08, p < 0.001) decreased with time; while bitterness increased (F 4, 1980 = 5.98, p < 0.001), and aftertaste showed no changes (F 4, 1980 = 0.82, p = 0.51. Aroma intensity followed a similar drop-off to overall liking (28+ days of storage) and an immediate decline in sweetness was observed while bitter changes were more stochastic (see Figure 3). However, it is important to note that bitterness had an observable (yet not statistically significant) interaction with higher storage temperatures affecting 'Diplomat' (mean difference (MD = 0.86)) more than 'Emerald Crown' (MD = 0.19; p = 0.053). Similarly, sweet intensity showed an interaction with cultivar and storage temperature (  Due to several differences observed in attribute acceptances, CATA data from varietals were analyzed separately (see Tables 2 and 3); however, no CATA differences on harvest day were found between varietals. Bitter and sweetness terms increased and decreased, respectively, during at 4 °C storage temperature for both cultivars; however, more significant trends were shown in 'Diplomat'. Aftertaste intensity significantly increased over time for each cultivar. 'Emerald Crown' had a significant increase in yellow and decrease in green color term count at both storage temperatures, indicating color distortion with age, while no color changes were significant with 'Diplomat'. Due to several differences observed in attribute acceptances, CATA data from varietals were analyzed separately (see Tables 2 and 3); however, no CATA differences on harvest day were found between varietals. Bitter and sweetness terms increased and decreased, respectively, during at 4 • C storage temperature for both cultivars; however, more significant trends were shown in 'Diplomat'. Aftertaste intensity significantly increased over time for each cultivar. 'Emerald Crown' had a significant increase in yellow and decrease in green color term count at both storage temperatures, indicating color distortion with age, while no color changes were significant with 'Diplomat'.

Drivers of Liking using Partial Least Squares (PLS)
The CATA descriptor counts were able to predict overall liking scores relatively well, accounting for 78.7% of the variance in liking. In all, 10 CATA descriptors were found to be important (Variable Importance in Projection (VIP) >0.8) in predicting liking scores (Figure 4). VIP is an assessment of the importance of each variable in the creation of a PLS model. Additionally, Figure 4 displays the regression coefficients on the x-axis. Regression coefficients indicate the direction, while the VIP (y-axis) indicates the magnitude, a variable has on predicting overall liking. The most influential positive drivers of liking include sweet, green, and intense color. The most influential negative drivers of liking include off-flavor, musty, dirt-like, and bitter. Importance in Projection (VIP) >0.8) in predicting liking scores (Figure 4). VIP is an assessment of the importance of each variable in the creation of a PLS model. Additionally, Figure 4 displays the regression coefficients on the x-axis. Regression coefficients indicate the direction, while the VIP (yaxis) indicates the magnitude, a variable has on predicting overall liking. The most influential positive drivers of liking include sweet, green, and intense color. The most influential negative drivers of liking include off-flavor, musty, dirt-like, and bitter.

Discussion
The current study reports on organoleptic changes to varieties within the Brassica genus during storage. Here, we show that several sensory changes occur during the storage of broccoli with some, but not all, changes being dependent on cultivar or holding temperatures, which in turn were sometimes dependent on each other.
Overall, the liking of the broccoli samples decreased over time, independent of cultivar and holding temperatures. The largest drop in overall liking was observed at 28 days of storage. In other words, cultivar and hold temperatures did not prevent consumers from noticing a decrease in quality of the broccoli. In a retail environment, a grocer typically receives broccoli within 7-14 days postharvest and sells this broccoli within the next week [3]. Our results demonstrate that the receiving or selling time of broccoli may have implications on overall liking of broccoli when eaten by the consumer. Therefore, producers and grocers should work to minimize distribution and shelf-storage times. Furthermore, culinary use of broccoli needs to be considered when determining the sensory shelf-life of the product. More specifically, broccoli is typically deemed to be of low quality by many in the produce industry much earlier than this study shows [25][26][27]. While it was found that broccoli acceptability decreases over time, the effects of extended storage were not as severe as previously reported [25][26][27]. This finding suggests growers and shippers of broccoli should readdress their current views of storage time to reflect common cooking methods such as steaming. In the case of steamed broccoli, the produce does not show the rapid declines in quality often used to drive industry norms.
Independent of storage time, 'Diplomat' benefitted from lower storage temperature and was more susceptible to organoleptic changes at higher temperatures than the 'Emerald Crown'. Inferring from the specific attribute changes, we can deduce several reasons for changes in overall liking. Sweetness and bitterness are both important taste attributes to the acceptance of a food product with

Discussion
The current study reports on organoleptic changes to varieties within the Brassica genus during storage. Here, we show that several sensory changes occur during the storage of broccoli with some, but not all, changes being dependent on cultivar or holding temperatures, which in turn were sometimes dependent on each other.
Overall, the liking of the broccoli samples decreased over time, independent of cultivar and holding temperatures. The largest drop in overall liking was observed at 28 days of storage. In other words, cultivar and hold temperatures did not prevent consumers from noticing a decrease in quality of the broccoli. In a retail environment, a grocer typically receives broccoli within 7-14 days post-harvest and sells this broccoli within the next week [3]. Our results demonstrate that the receiving or selling time of broccoli may have implications on overall liking of broccoli when eaten by the consumer. Therefore, producers and grocers should work to minimize distribution and shelf-storage times. Furthermore, culinary use of broccoli needs to be considered when determining the sensory shelf-life of the product. More specifically, broccoli is typically deemed to be of low quality by many in the produce industry much earlier than this study shows [25][26][27]. While it was found that broccoli acceptability decreases over time, the effects of extended storage were not as severe as previously reported [25][26][27]. This finding suggests growers and shippers of broccoli should readdress their current views of storage time to reflect common cooking methods such as steaming. In the case of steamed broccoli, the produce does not show the rapid declines in quality often used to drive industry norms. Independent of storage time, 'Diplomat' benefitted from lower storage temperature and was more susceptible to organoleptic changes at higher temperatures than the 'Emerald Crown'. Inferring from the specific attribute changes, we can deduce several reasons for changes in overall liking. Sweetness and bitterness are both important taste attributes to the acceptance of a food product with an increase and decrease increasing acceptance, respectively [28,29]. An increase in storage time decreased overall liking and subsequently decreased the sweetness and increased the bitterness of the broccoli samples. Additionally, the texture liking of the broccoli dropped over time for the higher stored temperature. This may be due to a reduction in water loss with rapid cooling [30]; however, weight of the samples were not collected. Similar results were obtained from a 2004 study in which storage degradation of broccoli under different packaging conditions (at two temperatures, 4 and 10 • C) resulted in reduced texture quality. In their study, they showed, with instrumental texture measurements, the broccoli became softer across four weeks for the most common packaging material, polyvinyl chloride (PVC), at 4 • C [7]. However, 'Diplomat', unlike the 'Emerald Crown', decreased in sweetness in both intensity scores and descriptive CATA analysis. Furthermore, "Dirt-like", a typically negative term, was increased only in the higher storage temperature for 'Diplomat', while color differences were seen in 'Emerald Crown'. Differences among cultivars are not uncommon [11]. A study looking at several cultivars of broccoli (n = 5) and cauliflower (n = 4) used descriptive and consumer data (n = 100) to show taste, appearance, and overall liking differences. Consumers preferred samples with lower levels of bitter tasting glucosinolates (alkenyl and indole glucosinolates) and a higher sucrose content (mostly fructose and glucose). Most glucosinolates decline with storage, especially at higher temperatures [31], but some have been shown to increase into deterioration (4-methoxglucobrassicin) [32]. Thus, maybe the decrease of sweetness unmasked the underlying bitterness [33] or perhaps other factors were at play. For instance, there is a high variability in the sensitivity to bitterness across the population due to genetics, which may be reflected in the variability of bitterness intensity across time points observed in our study (each of which had a different set of participants) [34]. Additionally, a large variation across broccoli samples existed for flavor terms such as green/ grassy, spicy, cabbage-like, and leek-like [11].
There were increasing levels of aftertaste revealed by the CATA data collected. Similarly, aftertaste and dirt-like were shown to be important variables to predict overall liking. CATA also showed a clear effect of storage temperature with higher temperatures.

Conclusions
Similar to most fresh produce, broccoli is a nutritious but perishable product whose purchase by consumers depends on its sensory attributes. Once harvested, broccoli continues to respire and senesce; and its quality decreases, reducing willingness to purchase. In particular, storage time imparts negative changes to appearance, basic tastes, flavor attributes, and texture. However, many of these changes can be slowed or deterred by decreasing the storage temperature and choosing a more resilient cultivar. For example, taste (bitter and sweet) and texture are major determinants of broccoli liking for both cultivars. Specifically, the Diplomat variety was more susceptible to negative taste outcomes and flavor changes (e.g. increase in attributes like cabbage and dirt-like) while the color degraded more for Emerald. Therefore, practitioners should use this information to implement procedures that could minimize harvest-to-consumption time and decrease storage temperatures while testing additional broccoli cultivars to determine quality resilience to these factors.