Food Chemistry 370 (2022) 130910

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Identification of volatile biomarkers for high-throughput sensing of soft rot

and Pythium leak diseases in stored potatoes
Worasit Sangjan a, Afef Marzougui a, D. Scott Mattinson b, Brenda K. Schroeder c,
Austin A. Bates c, Lav R. Khot a, d, Sindhuja Sankaran a, d, *
a
Department of Biological Systems Engineering, Washington State University, 1935 E. Grimes Way, PO Box 646120, Pullman, WA, USA
b
Department of Horticulture, PO Box 646414, Washington State University, Pullman, WA, USA
c
Department of Entomology, Plant Pathology and Nematology, University of Idaho, 875 Perimeter Drive MS 2329, Moscow, ID, USA
d
Center for Precision and Automated Agricultural Systems, Washington State University, 24106 N. Bunn Road, Prosser, WA, USA

A R T I C L E I N F O A B S T R A C T

Keyword: Soft rot and Pythium leak are postharvest storage diseases of potato tubers that can cause substantial crop losses
Potato rot in the US. This study focused on detecting volatile organic compounds (VOCs) associated with rot inoculated
Disease detection tubers during storage (up to 21 days) using headspace solid-phase microextraction (SPME) coupled to gas
Volatile organic compounds
chromatography (GC) with mass spectrometry (MS) and flame ionization detector (FID) analysis. Russet Burbank
Headspace sampling
Solid-phase microextraction
and Ranger Russet tubers were inoculated with the rot pathogens. Static sampling with 50 min trapping time
Gas chromatography followed by GC–MS and GC–FID analysis identified 23 and 30 common VOCs from the pathogen inoculated
tubers. Overall, n,n–dimethylmethylamine, acetone, 1–undecene, and styrene, occurred frequently and repeat­
ability in inoculated samples based on GC–MS analysis, with the latter two found using GC–FID analysis as well.
Identification of such biomarkers can be useful in developing high-throughput VOC sensing systems for early
disease detection in potato storage facilities.

1. Introduction the process industry and consumer demands. However, it relies on

effective postharvest storage management. The bulk storage losses of
Potato (Solanum tuberosum L.) is a high-value crop with remarkably tubers in the U.S. can be up to 7.5% annually, with significant losses due
high yield potential and relevant to global food security (Devaux, Kro­ to storage rots (Lui et al., 2005). These storage rot pathogens also
mann, & Ortiz, 2014; Drewnowski, Rehm, & Zhang, 2013). It is negatively affect tuber quality. The major storage rot diseases include
commonly produced in the temperate zone of the northern hemisphere pink rot (caused by Phytophthora erythroseptica), dry rot (caused by
(Food and Agriculture Organization of the United Nations (FAO), 2019). Fusarium sambucinum), late blight (caused by Phytophthora infestans),
About 65% of the produce is used for human consumption, 13% as an­ and Pythium leak (caused by Pythium ultimum) (Olsen, Miller, & Nolte,
imal feed and about 10% are reserved as seed. The remaining 12% is 2006). In addition, Pectobacterium carotovorum subsp. carotovorum
applied in other sectors, such as in ethanol production, paper factories, [formerly Erwinia carotovora subsp. carotovora] that causes soft rot is one
and the chemical industry (Devaux et al., 2020; Izmirlioglu & Demirci, of the top ten most devastating storage diseases (Mansfield et al., 2012).
2015). In the US, potato is one of the leading horticultural crops with a This bacterium often exhibits opportunistic nature, can spread rapidly,
total annual production yield of over 22 million tons (~49.70 t/ha) and serve as a secondary invader after the initial infection by other
(United States Department of Agriculture (USDA) — National Agricul­ pathogens (Olsen & Kleinkopf, 2020).
tural Statistics Service (NASS), 2019). Potato has a nutrient density, Effective in-field disease control, storage facility management, and
which can contribute to a balanced diet. Also, it has massive demand in early rot detection—as potato tubers can be stored up to 12 months
the processing industry to produce frozen potato products (41%) and depending on location and the demand–supply cycle—are key factors to
chips (21%) (Bohl & Johnson, 2010; National Potato Council, 2020). reducing these losses. Potato tubers can be infected in many ways, and
The year-round availability of potato tubers is thus crucial to meet many of the pathogen infections do not typically originate in the storage

* Corresponding author at: Department of Biological Systems Engineering, Washington State University, 1935 E. Grimes Way, PO Box 646120, Pullman, WA, USA.
E-mail address: sindhuja.sankaran@wsu.edu (S. Sankaran).

https://doi.org/10.1016/j.foodchem.2021.130910
Received 5 January 2021; Received in revised form 16 August 2021; Accepted 17 August 2021
Available online 4 September 2021
0308-8146/© 2021 Elsevier Ltd. All rights reserved.
W. Sangjan et al. Food Chemistry 370 (2022) 130910

but are carried into storage as latent infections (Olsen, 2014; Rutolo, 2. Materials and methods
Clarkson, Harper, & Covington, 2018). The amount of inoculum, i.e.,
spores or bacteria present on the tubers, and factors such as improper 2.1. Sample preparation
ventilation, temperature, humidity, and carbon dioxide level in a stor­
age facility can influence disease development (Alamar, Tosetti, Land­ Potato tubers of Russet Burbank and Ranger Russet varieties, rep­
ahl, Bermejo, & Terry, 2017; Potato Council, 2018). Pathogens gain resenting major certified seed potato and commercial variety acreage in
entry into the tuber through wounds that occur during harvest (Wang, the US (National Potato Council, 2019), were obtained from grower/
Naber, & Crosby, 2020). The storage pathogens cause potato tubers to seed cooperators in Washington and Montana (2018 harvest season).
rot and decay. This process results in a release of moisture that needs to Both varieties are often bulk stored for up to 12 months after harvest.
be removed from the storage facilities, as high humidity will enhance Potato tubers appearing healthy and uniform in size (150–200 g) were
soft rot development (Olsen & Kleinkopf, 2020). Identification of the selected, washed to remove dirt and excess soil, and surface-sterilized in
initiation of rots and spread in a pile is done by visual monitoring by 10% sodium hypochlorite solution for five minutes. The tubers were
observing sinkage in the piles and associated rotting smell in the facility cleaned and rinsed once again with sterile water for five minutes, dried
by the managers. Such assessments of large potato piles for early rot under a laminar flow, placed in a plastic bin, and kept overnight in the
detection are often subjective and possibly lately for remedial measures. dark at room temperature before inoculation.
Hence, the industry needs sensing derived from early rot detection Soft rot pathogen, Pectobacterium carotovorum subsp. carotovorum
approaches. Using volatile organic compounds (VOCs) released by the (isolate Ec101, Dung, Johnson, & Schroeder, 2014) was grown on
rotting tubers could be a viable sensing approach. The VOCs can be nutrient broth yeast extract in agar plates overnight at 28 ◦ C (Vidaver,
utilized as a sign/an indicator or biomarker to examine the stored crop’s 1967). Five mL of sterile water was added per plate, and the plates were
morphological and physiological status without destruction (Sankaran, scraped with a sterile hockey stick. The bacterial slurry liquid was then
Mishra, Ehsani, & Davis, 2010; Toivonen, 1997). Researchers have pipetted into a sterile 50 mL centrifuge tube, and sterile water was added
developed several techniques to trap and adsorb VOCs and apply iden­ to 50 mL in total, resulting in an inoculum concentration of about 6 × 108
tification methods such as gas chromatography–mass spectrometry colony forming units (CFU) per mL. This concentration of inoculum
(GC–MS), GC–flame ionization detector (GC–FID), gas sensors, and field standardized in this fashion was applied to ensure consistent rot devel­
asymmetric ion mobility spectrometry (FAIMS) for the early disease opment in the tubers. Ten mL aliquots were then transferred into five
detection in commercial potato storage conditions. Supplementary Data, sterile 15 mL centrifuge tubes (Dung, Johnson, & Schroeder, 2014), and a
Table S1, reports the information on these analytical techniques used for new tube was used to inoculate each replicate for reducing the chance of
rot detections in potato tubers. cross-contamination between the replicates. The injection sites on the
In this study, two analytical methods, GC–MS and GC–FID, were tubers’ surface were sterilized using 70% ethanol. A sterile #11 scalpel
evaluated to identify and quantify the VOCs released from Russet Bur­ (Bard-Parker 371611) was immersed in the 15 mL centrifuge tube con­
bank and Ranger Russet potato tubers inoculated with P. carotovorum taining P. carotovorum subsp. carotovorum and used to puncture the
subsp. carotovorum or P. ultimum causing soft rot and Pythium leak under periderm of the tuber. The scalpel was inserted into the tuber to a depth
bulk storage conditions. These two techniques were utilized as a paral­ of 3 cm, and ten insertions were made about 1 cm apart on a single potato
lel/alternative approach to sample the volatile profiles. The hypothesis tuber. Each insertion introduced about 3 × 107 to 3 × 108 CFU per tuber.
was that tubers would release some key VOCs at variable conditions The scalpel blade was cleaned with a cotton swab soaked in 70% ethanol
(multiple potato varieties, pathogens, storage conditions) that can be between stab inoculations and changed with every replicate of five tu­
used as biomarkers for early rot detection. The GC–MS and GC–FID re­ bers. Potato tubers were inoculated with sterile water using the same
sults were validated using the standard chemical compounds to obtain protocol, which served as a control treatment (non-inoculated control).
consistent, reliable, and accurate measurement data. This study also Pythium leak pathogen, Pythium ultimum (isolate Pu17, Kothawade,
reports headspace sampling protocol optimization to capture VOCs as Sankaran, Bates, Schroeder, & Khot, 2020) was grown on potato dextrose
well as the sensitivity of solid-phase microextraction (SPME) fibers with agar modified with 4 g/L casamino acids vitamin assay (Difco Sigma
either of carboxen–polydimethylsiloxane (CAR/PDMS) or poly­ Aldrich, USA) plates at room temperature under black light on a 12 h on
dimethylsiloxane–divinylbenzene (PDMS/DVB) coatings. Headspace and off-cycle. A sterilized size three-core borer (7 mm) was applied to
sampling using the SPME technique is a simple yet highly sensitive make mycelial plugs that was used to inoculate potato tubers. A similar
method for VOC extraction. Nevertheless, it is vital to optimize the sterilization method of the tuber’s surface before inoculation was per­
extraction process before sampling, especially on a complex matrix. The formed. The scalpel was cleaned between inoculations and changed be­
two modes of headspace sampling, static or dynamic, were optimized. tween each replication as described above. The only difference was the
The static mode involves the headspace air not being disturbed between technique of applying inoculum; a sterile scalpel (#11, refers to the long
the SPME’s fiber and the sample during the VOC trapping. Contrarily, triangular blade used for shallow stabbing incisions, Sigma Aldrich, USA)
the headspace air is circulated in the dynamic sampling mode (Razote, was used to cut a small circle of tuber tissue 2 cm in diameter and 1 cm
Jeon, Maghirang, & Chobpattana, 2002). deep into the tissue. The scalpel was then utilized to transfer the mycelial
In our previous studies (Sinha, Khot, Schroeder, & Si, 2017; Sinha, plug to the tuber wound, and the potato plug was placed on the mycelial
Khot, Schroeder, & Sankaran, 2018), a portable FAIMS was used to plug and pushed back in place. Tubers were inoculated with agar plugs
detect soft rot disease in Russet Burbank potato tubers stored at room with no mycelium to serve as a control treatment.
temperature. The FAIMS was able to reveal the development of storage Following the inoculum preparation and inoculation processes,
rot infections in tubers (P. carotovorum subsp. carotovorum) with the inoculated or control tubers were then transferred to replicates of glass
ability to detect rot symptoms within a day after inoculation. To further sampling jars. All 3.79 L glass jars/chambers (Specialty Bottle, WA, USA)
develop the FAIMS technique for early rot detection, we need to were sterilized in an autoclave before experimentation. A customized
customize the instrument to identify specific biomarkers associated with Petri dish containing 50 mL of sterile water was placed at the chamber’s
storage rot conditions. Therefore, the rationale behind this study was bottom to generate a saturated atmosphere to enhance disease develop­
that understanding the release and identification of specific biomarkers ment. Sterile stainless-steel supports were used to hold the tubers above
will allow the development of a high-throughput volatile sensing system the Petri dish to avoid contact with the water (Supplementary Data,
targeting these biomarker compounds for early disease detection in Fig. S1). Five potato tubers from the inoculated treatment (P. carotovorum
potato storage facilities. subsp. carotovorum or P. ultimum) were placed in the glass chamber and
considered as one replicate sample of inoculated treatment. Similarly,
another five tubers inoculated with sterile water/agar plugs with no

2
W. Sangjan et al. Food Chemistry 370 (2022) 130910

mycelium were placed in the chamber, considered as one replicate 2.3. Experimental protocol for VOC data collection
sample of the control treatment. The chambers were then sealed using a
food-grade cling film to create aerobic conditions during the storage. The summary of the experiments and data collection process was as
reported in Table 1. Each experiment included both treatments with
inoculated and control samples. The potato tubers were inoculated with
2.2. Sampling protocol optimization P. carotovorum subsp. carotovorum or P. ultimum as presented in the
previous section. The samples were stored at 24 ◦ C condition. Pertinent
Soft rot pathogen and Russet Burbank potato variety were utilized for to each sample, VOC profiles were monitored non-invasively at different
sampling protocol optimization, as shown in Fig. 1. In the optimization sampling time points on 0, 7, 14, and 21 DAI. The collection of VOCs
experiment, the potato tubers were prepared and inoculated with released by sample replicates was conducted following static headspace
pathogen to induce soft rot on Russet Burbank stored at 24 ◦ C, as sampling mode at Ht of 50 min. The samples were characterized by both
described in the previous section. There were eight glass chambers, GC–MS and GC–FID analysis. In addition, the experiment from Russet
consisting of four replicates of inoculated and four replicates of sterile Burbank and Ranger Russet potato tubers inoculated with
water control treatment samples. The pathogen development was P. carotovorum subsp. carotovorum as inoculated treatment and sterile
monitored on 7 and 21 days after inoculation (DAI). The two modes, water as control treatment stored at 10 ◦ C was also conducted with the
static and dynamic headspace sampling, coupled with GC–FID analysis, same protocol as described above. The related sampling time points
were used to find a reasonable method for trapping VOCs. were 0, 14, and 28 DAI for this reduced temperature condition.
During headspace sampling, the cover to seal the chamber was
changed from the cling film to a sterilized customized metal cap con­
taining one septum port for injection of the SPME with a coating of 0.65 2.4. GC–MS and GC–FID analysis
µm PDMS/DVB (Supelco Co., PA, USA). After one hour of headspace
accumulation at room temperature, the SPME was injected in the At the beginning of every sampling time point, the SPME fiber was
chamber as presented in Supplementary Data, Fig. S1, and the fiber was pre-conditioned for approximately five minutes in the GC injector at
exposed to the headspace for 30 or 50 min (referred to as Ht hereafter) 100 ◦ C prior to starting the headspace sampling. Additionally, static
and immediately desorbed in the GC–FID injector. The static (Supple­ headspace sampling using SPME with an empty sealed chamber (storage
mentary Data, Fig. S1a) and dynamic (Supplementary Data, Fig. S1b) setup unit without potatoes) was completed and analyzed under the
headspace sampling modes were tested in the volatiles’ extraction pro­ same conditions for both GC analyses to ascertain the possibility of other
cess. The difference between the two modes was a filtered airflow (25 volatile compounds being emitted from the experiment chamber (blank
mL/min) circulated inside the chamber during the dynamic headspace sample).
sampling, allowing volatile molecules to be adsorbed to the SPME fiber. GC–MS was utilized to identify VOCs using an Agilent 6890 GC
The different headspace trapping times (Ht: 30 and 50 min) were also equipped with an Agilent 5973 quadrupole mass spectrometer (Agilent
evaluated because the time influences the partition of solutes between Technologies, Santa Clara, CA, USA) and a ZB–1 MS capillary column
the headspace and the fiber coating, directly influencing the SPME (60 m × 0.32 mm; film thickness 1.00 µm) (Phenomenex, Torrance, CA,
performance (Reto, Figueira, Filipe, & Almeida, 2007). USA) by sampling two replicates for each treatment and at each sam­
In another optimization experiment, two specific SPME coatings, i.e., pling time point. The splitless injector’s temperature and the transfer
CAR/PDMS (24-gauge needle size with 0.566 mm outer diameter and line were kept constant at 250 ◦ C and 150 ◦ C, respectively. The column
1.93 µL/25.40 mm dead volume, and 75/85 µm sorbent film thickness) temperature was held at 33 ◦ C, and the fiber was retrieved after 4 min,
and PDMS/DVB (24-gauge needle size and 65/70 µm sorbent film thick­ followed by an increase to 50 ◦ C at a rate of 2 ◦ C/min, followed by 5 ◦ C/
ness) were assessed for their sensitivity analysis using GC–MS technique, min to a final temperature of 225 ◦ C/min and held for 5 min. Electron
as described in Fig. 1. The Russet Burbank tubers were inoculated with impact mode was set at 70 eV, and the molecular weight values were set
P. carotovorum subsp. carotovorum following a similar protocol described in the range of 30 to 250 atomic mass units. Compounds were identified
above using static sampling mode with Ht of 50 min. The tubers were by matching their mass spectra with the Wiley/NIST library of standard
stored at 24 ◦ C for up to 7 DAI, and two replicates of inoculated and two compounds (John Wiley & Sons Inc., Hoboken, NJ) and qualified using
replicates of control treatment samples were used to extract VOCs with Agilent ChemStation software (Agilent Technologies, Santa Clara, CA,
two SPME coatings. The samples were analyzed with the GC–MS system to USA). Over 85% match threshold was utilized during compound
evaluate these two SPME coatings’ performance. GC–MS was applied identification.
because it can identify the VOCs used to analyze the SPME coatings’ During GC–FID analysis, the SPME was injected directly into a
performance and cross-check with the VOCs from previous studies (Table Hewlett Packard 5890 Series II GC (Agilent Technologies, Santa Clara,
S1). The VOC trapping protocol was developed from the above experi­ CA, USA) coupled with a flame ionization detector. Both GC–MS and
mental methods and used in data collection as described below. GC–FID systems were set in the splitless injection mode. The separation

Fig 1. Summary of data acquisition and analysis protocol utilized in this study.

3
W. Sangjan et al. Food Chemistry 370 (2022) 130910

was achieved on a DB–1 MS capillary column (60 m × 0.32 mm; film between samples/experiments. Therefore, the dataset was pre-processed
thickness 0.50 µm) (Agilent Technologies, Santa Clara, CA, USA). The using a custom-developed MATLAB algorithm (version 2017a, The
same temperature program as the GC–MS analysis was utilized during MathWorks, Natick, MA, USA). Initially, the blank VOC profiles (SPME
the GC–FID analysis. The temperature increased at the rate of 2 ◦ C/min fiber and empty chamber sample) were subtracted from all the sample
to 50 ◦ C/min, then at a rate of 5 ◦ C/min until 225 ◦ C was reached and VOC profiles. A data matrix was then generated by aligning the peaks
maintained for 5 min. The total time per SPME run was 50 min with both between samples and merging for convenient analysis. Finally, the data
the inlet and detector temperatures held constant at 200 ◦ C. were analyzed using a custom-developed algorithm in R Studio (version
Moreover, the identification using GC–MS and GC–FID analysis were 3.2.5, R core Team, Vienna, Austria) to identify peaks and their simi­
validated by comparing the GC retention times (RTs) with the standard larities as well as differences, as described in Fig. 1.
chemical compounds, which also were identified by mass spectra, as In regard to the identification of common peaks from GC–FID anal­
shown in Fig. 1. The chemicals were purchased from Sigma-Aldrich ysis between inoculated and control treatment samples based on the four
(Sigma-Aldrich, St. Louis, MO, USA). Each compound was further experiments (Table 1), the peaks appearing in two or more replicate
identified by GC–MS, therefore matching the authentic compound to the samples in both treatments (inoculated and control) at each sampling
Wiley/NIST library and comparing it to the given compounds per their time point (e.g., 7, 14, and 21 DAI) were identified and considered for
elution order. The compounds were individually diluted in distilled statistical analysis. Moreover, these peaks were also present in two or
water, mixed, and then further delivered into a 4 mL SPME vial with more experiments (of 4 experiments, as described in Table 1). Similarly,
30% NaCl. The compounds were analyzed at levels from 4.20 ug/ml for the unique peaks (inoculated samples only) present in at least two
acetone; 0.86 ug/ml for dimethyl disulfide; 0.49 ug/ml in 50% methanol replicate samples and two experiments were also identified and
for 3–hydroxyl–2–butanone; 0.45 ug/ml for indole; 17 ug/ml for n, considered for statistical analysis.
n–dimethylmethylamine and 2–propanol; 0.45 ug/ml for styrene; and at
0.85 ug/ml for 1–undecene and 2–undecanone. In addition, each com­
2.6. Statistical analysis
pound was analyzed a second time by SPME desorption into the GC–FID
system to absolutely relate the retention times of the unknowns in
Statistical analysis was performed to compare the area of common
relation to the GC–MS. Finally, the Kovats retention index (RI) was
peaks between inoculated and control samples or unique peaks between
calculated for each compound listed here as a standard and reported.
two pathogens and varieties. Four replicate samples were prepared for
The RI was performed by running a series of straight-chain alkanes
each condition (varieties, pathogens, and treatment). Initially, the
(C4–C14) on each GC system.
Shapiro-Wilk’s significance test, the density plot analysis, and the
quantile–quantile plot analysis were performed to test for normality.
2.5. Data analysis Finally, Wilcoxon Rank Test was performed to compare the datasets.
Results were inferred at 5% level of significance. Furthermore, the
In the optimization experiment, two headspace sampling modes and principal components analysis (PCA) was applied to detect the coherent
two trapping times were evaluated. Thus, evaluated were 64 GC–FID pattern in the GC–FID based response as different clusters between
profiles (2 sampling modes: static/dynamic × 4 replicate samples × 2 inoculated and control treatment samples.
treatments: inoculated/control × 2 sampling time points: 7/21 DAI × 2
Ht: 30/50 min) from this experiment to identify a suitable headspace 3. Results and discussion
sampling mode based on the number of peaks and associated peak in­
tensities. Similarly, during the optimization of fiber coatings, 32 GC–MS 3.1. Optimization of sampling protocol
profiles (2 sampling modes: static/dynamic × 2 replicate samples × 2
treatments: inoculated/control × 1 sampling time points: 7 DAI × 2 Ht: The number of peaks or VOCs detected using GC–FID analysis for
30/50 min × 2 SPME coatings: CAR/PDMS and PDMS/DVB) were Russet Burbank tubers inoculated with P. carotovorum subsp. carotovorum
analyzed. varied based on the headspace sampling mode and trapping time, as
The total number of VOC sample profiles from GC–MS analysis presented in Fig. 2a. The GC–FID analyzer run with static sampling mode
during tuber experiments (Table 1) was approximately 64 (4 experi­ for Ht of 50 min detected more than 550 peaks per treatment. On the
ments with two pathogens and two tuber varieties × 2 treatments × 2 other hand, roughly 330 peaks were detected per treatment with dy­
replicate samples × 4 sampling time points), and the percentage namic sampling mode at Ht of 50 min. Moreover, either of the two
occurrence of biomarkers was ranked after the dataset was accumulated headspace sampling modes with Ht of 50 min generated more peaks than
and pre-processed. The percent occurrence was computed by dividing Ht of 30 min (Fig. 2a). In addition, the peak intensity was affected by the
the number of known events (samples with peak presence) with the total trapping time (Fig. 2b). The Ht of 50 min resulted in detecting a higher
number of events (sample size) and presenting the data as a percentage. VOC concentration with over 104 GC area units. The static sampling
Regarding GC–FID analysis, the total number of VOC sample profiles mode detected up to 40 peaks in the range of 104 to 105 GC area units,
was 96 (4 experiments with two pathogens and two tuber varieties × 2 while the dynamic mode could detect only one peak in a similar range in
treatments × 4 replicate samples × 3 sampling time points). Each VOC the control treatment samples (Fig. 2b). Static headspace sampling mode
profile from GC–FID analysis was represented by the RT and the relative could also trap the compounds having the intensity in the range of 1 to 5
area under each peak curve. However, the VOC profiles varied slightly × 105, and some were over 5 × 105 GC area units.

Table 1
Summary of experimental details during GC–MS and GC–FID analysis performed to identify potato rot related biomarkers over 21 days storage period.
Experiment Diseasea Potato Varietyb Storage Temperature,◦ C No. of Replicates per Treatment No. of Replicates Analyzed using GC Sampling Day, DAI

Inoculated Control MSc FID 0 7 14 21

1 Pcc RB 24 4 4 4 8 Md M/F M/F M/F

2 Pcc RR 24 4 4 4 8 M M/F M/F M/F
3 PU RB 24 4 4 4 8 M M/F M/F M/F
4 PU RR 24 4 4 4 8 M M/F M/F M/F
a
Pcc: P. carotovorum subsp. carotovorum, PU: P. ultimum; b RB: Russet Burbank, RR: Ranger Russet; c About two inoculated and two control replicate samples were
analyzed per experiment; and d M: Analysis with MS, M/F: Analysis with both MS and FID.

4
W. Sangjan et al. Food Chemistry 370 (2022) 130910

Fig 2. Optimization results: (a) number of peaks from tubers inoculated with P. carotovorum subsp. carotovorum using different sampling modes and headspace
trapping time; and (b) the number of peaks with observed peak area range. The optimization experiment had control (C) and inoculated (I) replicate samples where
headspace was sampled using static (S) and dynamic (D) modes with 30 min (30) and 50 min (50) trapping times.

Table 2
Percent occurrence of VOCs detected during the GC–MS analysis of Russet Burbank and Ranger Russet potato tubers inoculated with P. carotovorum subsp. carotovorum
or P. ultimum and stored at 24 ◦ C.
VOC No Total Occurrence, % Volatile Organic Compound RT, min Occurrence at Sampling Period, %

0 DAI 7 DAI 14 DAI 21 DAI

(na = 6) (n = 10) (n = 9) (n = 9)
b c
Common VOC 1 85 Unknown 31.80 100 100 78 67
2 44 Decanal 38.90 100 50 22 22
3 12 3,7–dimethyl–3–octanol 34.20 67 0 0 0
4 9 Dodecanol 45.40 50 0 0 0
5 6 Hexanal 21.00 33 0 0 0

VOC from inoculated treatment only 1 71 N,N–Dimethylmethylamine 3.80 50 50 89 89

2 65 1–Undecene 34.40 50 60 78 67
3 53 Styrene 25.50 0 10 89 100
4 53 Acetone 4.40 0 30 78 89
5 47 3–Hydroxy–2–butanone 13.50 17 60 44 56
6 38 3–Methyl–1–butanol 16.30 33 30 44 44
7 29 2–Undecanone 40.70 0 20 44 44
8 29 Ethanol 4.00 0 30 22 56
9 24 2–Propanol 4.70 0 0 33 56
10 24 1–Octanol 26.00 0 0 33 56
11 24 Indole 40.50 0 0 11 78
12 21 Dimethyl disulfide 16.70 33 20 33 0
13 18 1,2–Dimethoxybenzene 35.30 0 0 11 56
14 18 2–Methyl–1–butanol 16.60 0 20 11 33
15 12 Methoxymethyl–benzene 29.80 0 0 0 44
16 9 3–Octanone 29.40 0 0 22 11
17 9 Methoxybenzene 26.40 0 0 11 22
18 9 4–Ethyl–1,2–dimethoxybenzene 41.40 0 0 11 22
19 9 2–Nonanone 33.80 0 10 0 22
20 6 1–Butanol 11.90 0 10 0 11
21 6 2–Ethyl–1–hexanol 32.50 0 10 0 11
22 3 2–Butanone 8.20 0 0 0 11
23 3 2–Pentanone 12.40 0 0 0 11
a
n: Sample size of inoculated samples; bThe total occurrences (n = 10) in control samples for unknown compound at 31.80 min RT, decanal, 3,7–dimethyl–3–octanol,
dodecanol, and hexanal were 80–100% (0–21 DAI), 60–90% (0–21 DAI), 10–30% (0–7 DAI), 40% (0 DAI), and 30% (0 DAI), respectively; and cCompound not found
from the database in the MS library.

5
W. Sangjan et al. Food Chemistry 370 (2022) 130910

When GC–MS analysis was performed to assess the fiber perfor­ 3.2. GC–MS analysis
mance, results revealed that SPME coated with CAR/PDMS trapped
VOCs better, hence the high numbers of peaks compared to PDMS/DVB Static headspace sampling followed by SPME coupled with GC–MS
(Supplementary Data, Fig. S2). Nevertheless, blank normalized data analysis on the samples, detected six typical peaks in both control and
showed similar results for both fibers (data not shown). These results inoculated treatments (major ones reported in Table 2) and one peak
included three VOCs commonly found in rotting potato tubers: dimethyl (nonanol) appearing only in the control treatment samples. The five
disulfide (RT = 16.7 min), styrene (RT = 25.5 min), and 3–octanone (RT peaks could be detected early at the 0 DAI, with an unknown compound
= 29.4 min). Overall, the static headspace sampling mode using the at RT of 31.80 min and the decanal that appeared at every sampling time
PDMS/DVB SPME coating with a Ht of 50 min was used in this study. point on both control and inoculated samples. The analysis of VOCs

Fig 3. Statistical analysis of common peaks detected during the GC–FID analysis from control and inoculated samples.

6
W. Sangjan et al. Food Chemistry 370 (2022) 130910

appearing only from the inoculated treatment samples allowed the P. carotovorum subsp. carotovorum and P. ultimum. Comparing different
identification of 23 compounds and their repeatability in terms of sampling periods (7, 14, and 21 DAI), two unique inoculated peaks at RT
occurrence frequency at each sampling time is reported in Table 2. The of 26.75 and 26.93 min were consistently found in all three periods,
two frequently occurring compounds as a result of rot were n, which could be used as biomarkers for early detection of both patho­
n–dimethylmethylamine and 1–undecene. Additional consistently gens. In addition, peaks with RT of 2.93, 3.00, 3.11, 3.17, 16.99, 27.10,
appearing compounds, at each sampling time point, included and 31.67 min were present in at least two sampling time points.
3–hydroxy–2–butanone, 3–methyl–1–butanol, and dimethyl disulfide. Finally, unique pathogen-specific (P. carotovorum subsp. carotovorum
In another scenario, GC–MS analysis from two tuber varieties (Russet or Pythium ultimum) peaks released from Ranger Russet and Russet
Burbank and Ranger Russet) inoculated with P. carotovorum subsp. Burbank tubers were evaluated from the same dataset. Three peaks from
carotovorum or sterile water stored at 10 and 24 ◦ C were analyzed. About two sampling time points (14 and 21 DAI) were detected as VOCs
21 unique compounds originating from the inoculated treatment were released from tubers inoculated with P. carotovorum subsp. carotovorum
detected (major peaks summarized in Supplementary Data, Table S2). (Supplementary Data, Table S6). These peaks were not significantly
Results also show that n,n–dimethylmethylamine and 1–undecene were different for Russet Burbank and Ranger Russet samples. No such peaks
the two major VOCs that appeared in all sampling time points, similar to were detected from tubers of both varieties inoculated with P. ultimum.
the results described in Table 2. Furthermore, as observed from Table 2
and Supplementary Data, Table S2, n,n–dimethylmethylamine, 3.4. Chemical biomarkers
1–undecene, 3–hydroxy–2–butanone, and 3–methyl–1–butanol were
consistently detected at every sampling time point. Nonetheless, there The GC–MS analysis detected 23 VOCs from inoculated tubers that
were some differences in peaks based on the GC–MS analysis of tubers were absent in control tubers, while the GC–FID analysis detected 30
inoculated with P. carotovorum subsp. carotovorum between samples VOC peaks. In general, the GC–FID detected more VOCs than GC–MS
stored at 10 and 24 ◦ C. Overall, the number of peaks and peak intensity analysis. The sensitivity and selectivity comparison of GC–MS and
were lower at 10 ◦ C, as presented in Supplementary Data, Fig. S3. GC–FID techniques have been reported in some studies. FID has been
In addition to the compounds that were commonly found between reported to be able to detect down to a level of 5 pg with a 107 dynamic
soft rot and Pythium leak inoculated samples, efforts were made to range, whereas MS can detect to a level of 10 pg at a 105 range in the
identify peaks present in inoculated samples of one pathogen, not the total ion mode (Buffington & Wilson, 1991). Aparicio-Ruiz, García-
other. GC–MS analysis confirmed 4–ethyl–1,2–dimethoxybenzene and González, Morales, Lobo-Prieto, and Romero (2018) utilized both
methoxybenzene as the compounds related to Pythium leak samples GC–MS and GC–FID techniques for analyzing volatile compounds in
stored at 24 ◦ C, absent in soft rot inoculated samples. Overall, the virgin olive oil and stated that FID showed good selectivity, linearity,
dominant VOC based on GC–MS analysis, n,n–dimethylmethylamine and applicability at a upper working range; while MS was suitable at a
detected only from inoculated treatment, has not been reported previ­ lower working ranges with better sensitivity and limits of detection and
ously. In contrast, other VOCs such as 1–undecene, acetone, ethanol, quantification. FID was reported to be applicable for categorizing virgin
dimethyl disulfide, 1,2–dimethoxybenzene, 1–butanol, 2–butanone, and olive oil categories. Van Asten (2002) reported that even if MS detectors
2–pentanone were previously reported in association with tubers are extremely sensitive, they are not as robust as the FID; thus, sensi­
infected with soft rot pathogen (de Lacy Costello et al., 1999; Lui et al., tivity drift is noticed. Nevertheless, the GC–MS offers the ability to
2005; Ouellette, Raghavan, Reeleder, & Greenhalgh, 1990; Varns & identify the compounds using the MS database (Kushalappa, Lui, Chen,
Glynn, 1979; Waterer & Pritchard, 1985). & Lee, 2002; Romero, García-González, Aparicio-Ruiz, & Morales, 2015;
Silva et al., 2018).
3.3. GC–FID analysis In this study, the standard chemical compound peaks relating to
major VOCs, as detected by the GC–MS analysis (Table 2), were used for
3.3.1. Common peaks between inoculated and control treatment samples comparison in both GC–MS and GC–FID analysis. The peak RT of these
Nine common peaks were found from the experiments at 24 ◦ C chemicals was matched with the corresponding GC–MS and GC–FID
storage temperature, as presented in Fig. 3. In addition, the peak at RT of profiles. Table 3 summarizes the results confirming the VOCs associated
36.28 min was detected in both tuber varieties inoculated with with sample potato tubers inoculated with P. carotovorum subsp. car­
P. carotovorum subsp. carotovorum stored at 10 and 24 ◦ C. Supplemen­ otovorum or P. ultimum. These chemicals matched the peaks identified
tary Data, Table S3, summarizes the results demonstrating that many using GC–MS analysis (Table 2) and GC–FID analysis (Supplementary
common peaks were neither normally distributed nor significantly Data, Fig. S5).
different. Three peaks (RT = 25.86, 26.14, and 26.59 min) showed peak Overall, the major biomarkers from Russet Burbank and Ranger
area differences between samples inoculated with P. carotovorum subsp. Russet tubers inoculated with P. carotovorum subsp. carotovorum or
carotovorum or P. ultimum and control. Supplementary Data, Fig. S4, P. ultimum stored at 24 ◦ C would be n,n–dimethylmethylamine,
presented the result of different clusters between inoculated and control 1–undecene, acetone, and styrene based on GC–MS analysis. Amongst
treatment samples at 14 and 21 DAI from the experiments at 24 ◦ C these, 1–undecene and styrene had a high occurrence percentage and
storage temperature condition. repeatability at different sampling time points as detected using both
GC–MS and GC–FID analysis. The compound 1–undecene appeared at
3.3.2. Unique peaks in inoculated treatment samples every sampling time point. These results further support our initial hy­
There were 30 unique peaks detected from tubers inoculated with pothesis and pertinent FAIMS work by Sinha, Khot, Schroeder, and Si
P. carotovorum subsp. carotovorum or P. ultimum that are common be­ (2017), Sinha, Khot, Schroeder, and Sankaran (2018) that VOC bio­
tween both pathogens (Supplementary Data, Fig. S5). Two peaks at RT markers can be used as a reliable and potentially rapid approach for
of 16.99 and 26.75 min occurred in more than 50% samples, and their potato tuber rot detection in stored potatoes. Nevertheless, the bio­
average peak area was around 28 × 103 and 41 × 103 GC units, markers need to be further validated with more variable conditions
respectively. The maximum average peak area was 345 × 103 GC units (other types of rot, potato varieties) and biological samples.
detected at RT of 3.17 min. In general, the average peak area was about It should be noted that the VOCs and their peak intensities are
57 × 103 GC units for these unique peaks (Supplementary Data, Fig. S5). influenced by the incubation period, where the rot progression and
These 30 unique peaks were filtered to evaluate the differences be­ disease severity will tend to increase with the incubation period that can
tween tubers inoculated with different pathogens at each sampling time be associated with an increase in pathogen activity. The pathogen ac­
point (Supplementary Data, Tables S4 and S5). Overall, there were no tivity will also increase with raised temperature. Nevertheless, as the
statistical differences between the peaks from tubers inoculated with goal was towards rot detection at different rot-development stages

7
W. Sangjan et al. Food Chemistry 370 (2022) 130910

Table 3
Standard chemical compounds related to major VOCs detected using the GC–MS analysis from inoculated tubers. These compounds were validated and compared to
the retention times from both GC–MS and GC–FID analysis, and percent total occurrence of inoculated VOCs/peaks are reported.
Standard Chemical Compound GC–MS Results GC–FID Results

(Rank) RTb, min RIc (Rank) RTe, min RI Occurrence of Unique

Total Occurrencea, % (n = 34) Total Occurrenced, % (n = 48) Inoculated Peakf

7 DAI 14 DAI 21 DAI

N,N–Dimethylmethylamine (1) 71 3.80 517.6 (7) 31 2.93 541.0 Xg X

1–Undecene (2) 65 34.40 1169.2 (1) 56 26.75 1208.0 X X X
Styrene (3) 53 25.50 963.4 (2) 52 16.99 991.6 X X
Acetone (4) 53 4.40 550.5 (6) 33 3.00 544.9 X X
3–Hydroxy–2–butanone (5) 47 13.50 766.1 (21) 13 5.96 786.3
2–Undecanone (7) 29 40.70 1352.6 (12) 23 32.34 1377.7 X
2–Propanol (9) 24 4.70 570.5 (9) 27 3.17 554.4 X X
Indole (11) 24 40.50 1343.6 (8) 29 31.67 1357.5 X X
Dimethyl disulfide (12) 21 16.70 814.0 (29) 8 7.40 819.6
a
Results from Table 2; b Retention time of the standard chemical compound detected from the GC–MS analysis; c RI: Kovats retention index; d Results from Sup­
plementary Materials, Fig. S5; e Retention time of the standard chemical compound detected from the GC–FID analysis; f Results from Supplementary Materials, Tables
S4 and S5, and n = 16 for each DAI; g X: Present in samples.

across environments, biomarkers were identified by integrating the data review & editing. Afef Marzougui: Methodology, Formal analysis,
across time points, although the number of peaks and peak intensities Writing - review & editing. D. Scott Mattinson: Methodology, Formal
tend to increase with the incubation period. analysis, Investigation, Writing - review & editing, Supervision. Brenda
K. Schroeder: Methodology, Formal analysis, Investigation, Writing -
4. Conclusion review & editing, Supervision. Austin A. Bates: Methodology, Investi­
gation. Lav R. Khot: Conceptualization, Methodology, Formal analysis,
The qualitative analysis of VOC peaks detected by the GC–MS and Investigation, Resources, Writing - review & editing, Project adminis­
GC–FID analysis identified 23 and 30 unique peaks released from Russet tration, Funding acquisition. Sindhuja Sankaran: Conceptualization,
Burbank and Ranger Russet inoculated with either P. carotovorum subsp. Methodology, Formal analysis, Investigation, Resources, Writing - re­
carotovorum or P. ultimum stored at 24 ◦ C, respectively. These com­ view & editing, Project administration, Funding acquisition.
pounds are indicative of infection occurring in the potato tubers.
Quantitative GC–FID analysis results revealed no significant differences Declaration of Competing Interest
in the unique inoculated peaks for Russet Burbank and Ranger Russet
samples, The two key findings from this research are: (1) Validation The authors declare that they have no known competing financial
using standard chemical compounds indicated that n, interests or personal relationships that could have appeared to influence
n–dimethylmethylamine, 1–undecene, acetone, and styrene would serve the work reported in this paper.
as biomarkers for Russet Burbank and Ranger Russet tubers infected
with P. carotovorum subsp. carotovorum or P. ultimum using GC–MS Acknowledgments
analysis, and (2) Static headspace sampling for 50 min using the PDMS/
DVB coated SPME was optimal for extracting VOCs with low to medium This study was funded, in part, by the Washington State Department
molecular weights and low to medium boiling points. The proposed of Agriculture, Specialty Crop Block Grant # K2296, and supported by
optimization protocols (static/dynamic, SPME coatings) can be relevant the USDA National Institute of Food and Agriculture Hatch project #
and applicable to other researchers utilizing similar techniques (GC–MS, 1014919. We would like to thank Blaine Meek from AgriNorthwest for
GC–FID) for volatile sampling to study food chemistry of crop or pro­ providing potato tuber samples and additional support during this
cessed food products. project and thank Dr. Mark Pavek for assisting us with the acquisition of
The release of the VOCs can be associated with the potato tuber potato tuber samples.
biochemistry and the pathogenicity process of the pathogens. Studying
the changes in VOC profiles considering both variabilities in potato Appendix A. Supplementary data
varieties and pathogens, thus, is important to develop VOC-based sensor
technologies for early disease detection in storage facilities with broader Supplementary data to this article can be found online at https://doi.
implementations. This would be one of the key contributions of this org/10.1016/j.foodchem.2021.130910.
study. The biomarkers identified herein can be utilized for the detection
of potato tuber rot caused by P. carotovorum subsp. carotovorum or References
P. ultimum. Potato storages could be monitored using the portable FAIMS
system or similar types of sensors for these biomarkers’ presence and Alamar, M. C., Tosetti, R., Landahl, S., Bermejo, A., & Terry, L. A. (2017). Assuring potato
serve as an early indicator of tuber rot development. This equipment and tuber quality during storage: A future perspective. Frontiers in Plant Science, 8, 2034.
https://doi.org/10.3389/fpls.2017.02034
methods would provide critical information for storage managers to Aparicio-Ruiz, R., García-González, D. L., Morales, M. T., Lobo-Prieto, A., & Romero, I.
measure the stored potato tubers’ overall health and enable informed (2018). Comparison of two analytical methods validated for the determination of
decisions to ensure the potato stakeholders’ profitability in monitoring volatile compounds in virgin olive oil: GC-FID vs GC-MS. Talanta, 187, 133–141.
https://doi.org/10.1016/j.talanta.2018.05.008
rot conditions. Bohl, W. H., & Johnson, S. B. (2010). Commercial potato production in North America. The
Potato Association of America. https://potatoassociation.org/wp-content/uploa
CRediT authorship contribution statement ds/2014/04/A_ProductionHandbook_Final_000.pdf.
Buffington, R., & Wilson, M. K. (1991). Detectors for gas chromatography -A practical
primer. Avondale, PA: Hewlett-Packard Avondale Division.
Worasit Sangjan: Methodology, Investigation, Validation, Formal de Lacy Costello, B., Evans, P., Ewen, R. J., Gunson, H. E., Ratcliffe, N. M., & Spencer-
analysis, Data curation, Visualization, Writing - original draft, Writing - Phillips, P. T. N. (1999). Identification of volatiles generated by potato tubers
(Solanum tuberosum CV: Maris Piper) infected by Erwinia carotovora, Bacillus

8
W. Sangjan et al. Food Chemistry 370 (2022) 130910

polymyxa and Arthrobacter sp. Plant Pathology, 48(3), 345–351. https://doi.org/ Razote, E., Jeon, I., Maghirang, R., & Chobpattana, W. (2002). Dynamic air sampling of
10.1046/j.1365-3059.1999.00357.x volatile organic compounds using solid phase microextraction. Journal of
Devaux, A., Goffart, J. P., Petsakos, A., Kromann, P., Gatto, M., Okello, J., Suarez, V., & Environmental Science and Health, Part B, 37(4), 365–378. https://doi.org/10.1081/
Hareau, G. (2020). Global food security, contributions from sustainable potato agri- PFC-120004477
food systems. In H. Campos and O. Ortiz (Eds.), The potato crop. Its agricultural, Reto, M., Figueira, M. E., Filipe, H. M., & Almeida, C. M. M. (2007). Analysis of vitamin K
nutritional and social contribution to humankind (pp. 3–35). Springer, Cham. https:// in green tea leafs and infusions by SPME–GC–FID. Food Chemistry, 100(1), 405–411.
doi.org/10.1007/978-3-030-28683-5_1. https://doi.org/10.1016/j.foodchem.2005.09.016
Devaux, A., Kromann, P., & Ortiz, O. (2014). Potatoes for sustainable global food Romero, I., García-González, D. L., Aparicio-Ruiz, R., & Morales, M. T. (2015). Validation
security. Potato Research, 57(3–4), 185–199. https://doi.org/10.1007/s11540-014- of SPME–GCMS method for the analysis of virgin olive oil volatiles responsible for
9265-1 sensory defects. Talanta, 134, 394–401. https://doi.org/10.1016/j.
Drewnowski, A., Rehm, C. D., & Zhang, H. (2013). Vegetable cost metrics show that talanta.2014.11.032
potatoes and beans provide most nutrients per penny. PLOS ONE, 8(5), e63277. Rutolo, M. F., Clarkson, J. P., Harper, G., & Covington, J. A. (2018). The use of gas phase
https://doi.org/10.1371/journal.pone.0063277 detection and monitoring of potato soft rot infection in store. Postharvest Biology and
Dung, J. K. S., Johnson, D. A., & Schroeder, B. K. (2014). Role of co-infection by Technology, 145, 15–19. https://doi.org/10.1016/j.postharvbio.2018.05.016
Pectobacterium and Verticillium dahliae in the development of early dying and aerial Sankaran, S., Mishra, A., Ehsani, R., & Davis, C. (2010). A review of advanced techniques
stem rot of Russet Burbank potato. Plant Pathology, 63(2), 299–307. https://doi.org/ for detecting plant diseases. Computers and Electronics in Agriculture, 72(1), 1–13.
10.1111/ppa.2014.63.issue-210.1111/ppa.12086 https://doi.org/10.1016/j.compag.2010.02.007
Food and Agriculture Organization of the United Nations (FAO). (2019). World Food and Silva, F. A. N., da Silva, A. A., de Sousa Fernandes, N., Rodrigues, T. H. S., Canuto, K. M.,
Agriculture — Statistical Pocketbook 2019. http://www.fao.org/3/ca6463en/ca646 do Nascimento, R. F., … Zocolo, G. J. (2018). Evaluation of headspace solid-phase
3en.pdf. microextraction gas chromatography–mass spectrometry for the characterization of
Izmirlioglu, G., & Demirci, A. (2015). Enhanced bio-ethanol production from industrial volatile organic compounds from Melon (Cucumis melo L.) flowers. Chromatographia,
potato waste by statistical medium optimization. International Journal of Molecular 81(8), 1231–1239. https://doi.org/10.1007/s10337-018-3550-0
Sciences, 16(10), 24490–24505. https://doi.org/10.3390/ijms161024490 Sinha, R., Khot, L. R., Schroeder, B. K., & Sankaran, S. (2018). FAIMS based volatile
Kothawade, G. S., Sankaran, S., Bates, A. A., Schroeder, B. K., & Khot, L. R. (2020). fingerprinting for real-time postharvest storage infections detection in stored
Feasibility of volatile biomarker-based detection of Pythium leak in postharvest potatoes and onions. Postharvest Biology and Technology, 135, 83–92. https://doi.org/
stored potato tubers using field asymmetric ion mobility spectrometry. Sensors, 20 10.1016/j.postharvbio.2017.09.003
(24), 7350. https://doi.org/10.3390/s20247350 Sinha, R., Khot, L. R., Schroeder, B. K., & Si, Y. (2017). Rapid and non-destructive
Kushalappa, A. C., Lui, L. H., Chen, C. R., & Lee, B. (2002). Volatile fingerprinting detection of Pectobacterium carotovorum causing soft rot in stored potatoes through
(SPME–GC–FID) to detect and discriminate diseases of potato tubers. Plant Disease, volatile biomarkers sensing. Crop Protection, 93, 122–131. https://doi.org/10.1016/
86(2), 131–137. https://doi.org/10.1094/PDIS.2002.86.2.131 j.cropro.2016.11.028
Lui, L. H., Vikram, A., Abu-Nada, Y., Kushalappa, A. C., Raghavan, G. S. V., & Al- Toivonen, P. M. (1997). Non-ethylene, non-respiratory volatiles in harvested fruits and
Mughrabi, K. (2005). Volatile metabolic profiling for discrimination of potato tubers vegetables: Their occurrence, biological activity and control. Postharvest Biology and
inoculated with dry and soft rot pathogens. American Journal of Potato Research, 82 Technology, 12(2), 109–125. https://doi.org/10.1016/S0925-5214(97)00048-3
(1), 1–8. https://doi.org/10.1007/BF02894914 United States Department of Agriculture (USDA) — National Agricultural Statistics
Mansfield, J., Genin, S., Magori, S., Citovsky, V., Sriariyanum, M., Ronald, P., … Foster, Service (NASS). (2019). Potatoes 2018 Summary. https://downloads.usda.library.co
G. D. (2012). Top 10 plant pathogenic bacteria in molecular plant pathology. rnell.edu/usda-esmis/files/fx719m44h/7p88cv74d/b5645489z/pots0919.pdf.
Molecular Plant Pathology, 13(6), 614–629. https://doi.org/10.1111/j.1364-3703. Van Asten, A. (2002). The importance of GC and GC-MS in perfume analysis. TrAC Trends
2012.00804.x. in Analytical Chemistry, 21(9–10), 698–708. https://doi.org/10.1016/S0165-9936
National Potato Council. (2019). Potato Statistical Yearbook 2019. https://www.nation (02)00807-5
alpotatocouncil.org/files/5015/6380/8213/2019_Stat_Book__fnl.pdf. Varns, J. L., & Glynn, M. T. (1979). Detection of disease in stored potatoes by volatile
National Potato Council. (2020). Annual Potato Yearbook 2020. https://www.nationalpo monitoring. American Potato Journal, 56(4), 185–197. https://doi.org/10.1007/
tatocouncil.org/files/8415/9231/8115/Annual_Potato_Yearbook_FNL.pdf. BF02853365
Olsen, N., & Kleinkopf, G. (2020). In Potato Production Systems (pp. 523–545). Cham: Vidaver, A. M. (1967). Synthetic and complex media for the rapid detection of
Springer International Publishing. https://doi.org/10.1007/978-3-030-39157-7_17. fluorescence of phytopathogenic pseudomonads: Effect of the carbon source. Applied
Olsen, N. (2014). Potato storage management: A global perspective. Potato Research, 57 Microbiology, 15(6), 1523–1524. https://digitalcommons.unl.edu/plantpath
(3-4), 331–333. https://doi.org/10.1007/s11540-015-9283-7 papers/59.
Olsen, N., Miller, J., & Nolte, P. (2006). Diagnosis and management of potato storage Wang, Y., Naber, M. R., & Crosby, T. W. (2020). Effects of wound-healing management
diseases. University of Idaho Extension. http://www.cals.uidaho.edu/edcomm/pdf/ci on potato postharvest storability. Agronomy, 10(4), 512. https://doi.org/10.3390/
s/cis1131.pdf. agronomy10040512
Ouellette, E., Raghavan, G. S. V., Reeleder, R. D., & Greenhalgh, R. (1990). Volatile Waterer, D. R., & Pritchard, M. K. (1985). Production of volatile metabolites in potatoes
monitoring technique for disease detection in stored potatoes. Journal of Food infected by Erwinia carotovora var. carotovora and E. carotovora var. atroseptica.
Processing and Preservation, 14(4), 279–300. https://doi.org/10.1111/jfpp.1990.14. Canadian Journal of Plant Pathology, 7(1), 47–51. https://doi.org/10.1080/
issue-410.1111/j.1745-4549.1990.tb00134.x 07060668509501513
Potato Council. (2018). Store Managers’ Guide. https://ahdb.org.uk/knowledge-librar
y/potato-store-managers-guide.