Tag Archives: microbiomes

Minimizing Hazards and Fraud in Milk, IBM Research Partners with Cornell University

By Food Safety Tech Staff
1 Comment

Americans consume an estimated 600 pounds of milk and milk-based products annually, according to the USDA. In an effort to minimize the hazards in the milk supply and prevent food fraud, IBM Research and Cornell University are joining forces. Combining next-generation sequencing with bioinformatics, the research project will collect genetic data from the microbiome of raw milk samples in a real-world situation at the Cornell University dairy plant and farm in Ithaca, New York.

Specifically, IBM and Cornell will sequence and analyze the DNA and RNA of food microbiomes, which will serve as a raw milk baseline, to develop tools that monitor raw milk and detect abnormalities that could indicate safety hazards and potential fraud. The data collected may also be used to expand existing bioinformatics analytical tools used by the Consortium for Sequencing the Food Supply Chain, a project that was launched by IBM Research and Mars, Inc. at the beginning of 2015.

“As nature’s most perfect food, milk is an excellent model for studying the genetics of food. As a leader in genomics research, the Department of Food Science expects this research collaboration with IBM will lead to exciting opportunities to apply findings to multiple food products in locations worldwide.” – Martin Wiedmann, Gellert Family Professor in Food Safety, Cornell University.

“Characterizing what is ‘normal’ for a food ingredient can better allow the observation of when something goes awry,” said Geraud Dubois, director of the Consortium for Sequencing the Food Supply Chain, IBM Research – Almaden, in a press release. “Detecting unknown anomalies is a challenge in food safety and serious repercussions may arise due to contaminants that may never have been seen in the food supply chain before.”

Cornell University is the first academic institution to join the Consortium for Sequencing the Food Supply Chain.

Douglas Marshall, Ph.D., Eurofins
Food Genomics

To Be or Not to Be: Choosing the Best Indicator using Microbiomes

By Douglas Marshall, Ph.D., Gregory Siragusa, Ph.D.
1 Comment
Douglas Marshall, Ph.D., Eurofins

Whenever an order is placed for an aerobic plate count, lactic acid bacteria count,  Enterobacteriaceae count, coliform count, fecal coliform count, Escherichia coli count, or yeast and mold count, it involves ordering an indicator test. So obviously, in food and water quality safety analyses, indicator microbiology is a highly routine and frequent activity. In fact, many business-to-business transactions are partially dictated by the outcome of indicator tests in the form of purchase specifications. Raw material producers and ingredient manufacturers are required to deliver products that meet the expectations of the buyer. Should such product exceed the predefined specifications, the expected transaction becomes nullified. Finished food product manufacturers also must meet specifications set by retailers and food service buyers. Some regulatory jurisdictions and public health agencies also are in this game, offering regulatory specifications for targeted groups of indicator microbes, such as EPA water quality specifications or FDA zero tolerance of pathogens in ready-to-eat foods. Here we argue that microbiomes can be a valuable tool to help choose and validate the best indicator(s) that may be used under a variety of circumstances.

The microbial indicator premise is that the presence or population size of a single indicator microbe or groups of microbes has some respective correlation with the presence of or population of either undesirable microbes (spoilers or pathogens) or desirable microbes (starter cultures or probiotics). On the public health side, indicator presence or populations can be used to define risk of an adverse public health outcome. Indicators also have utility in assessing process effectiveness, such as presence or populations of spore formers after a heat process. Sanitation efficacy can be judged by the amount of an appropriate indicator, such as residual ATP on a surface or presence of Listeria spp. in a floor drain. Culture houses (i.e., starter cultures, probiotics) and companies that manufacture fermented foods can do routine QC testing for the amount of metabolic byproducts (CO2 or acid development) as an indicator of microbial activity and also measure culture population levels.

Our customers frequently ask us, “Which is the best indicator for my ingredients, process, and products?”  Of course they are looking for a very simple answer but the reality is, we must know many details about the ingredients, product, process, and intended use before we can offer a best guess. Clearly best guesses, even by esteemed experts, can lead to inappropriate indicator choices. At worse, standard industry practice informed by years of use may not offer appropriate scientific validation of the use of chosen indicators.

Another customer question we frequently hear: “My product is not reaching intended shelf life, but indicator counts show it should be fine. What is causing product performance failure?”  In this scenario, the chosen indicator(s) may not allow for cultivation of the offending microbe, resulting in an “all clear” test result. Each indicator test (see Table I) will grow only the microbes able to multiply on the selected medium, at the selected incubation temperature, for the selected incubation time, and under the selected incubation atmosphere.1 Differences in media brand, or even slight deviations in media nutrition, media selective agents, temperature, time or atmosphere will have dramatic implications on what ultimately grows. What is telling is that there are many microbes in the sample that may not be cultivable at all, yet they may contribute to product performance failures. Wouldn’t it be nice if you could run one test and get a good snapshot of the all microbiota present in the specimen?

A further issue is that some microbes, which  may be perfectly able to grow under a certain set of conditions, might be outgrown by other competitors. Therefore, they may not contribute to the countable population. If they are found as a minor population, the odds of identifying them from a plate count are remote. Such microbes may in fact contribute to product failure and yet never be detected by an indicator assay.

An example of well-publicized historical misuse of indicators is the application of fecal coliform counts as indicators of fecal contamination of some dry leafy food products, such as tea leaves. For decades, periodic popular press exposé articles about food service iced tea with high fecal coliform counts have appeared in the news. The respective author’s dramatic conclusion is that such teas are contaminated with feces and a threat to public health. However, in reality, when the bacteria associated with these high counts were actually identified, they were determined to be natural constituents of dry tea leaves and had no association with animal feces.2

An unconventional hypothetical indicator example seems worthy here. If you are a manufacturer of a dried ready-to-eat product or ingredient and your hazard analysis has identified Salmonella as a reasonably foreseeable environmental hazard, most will choose coliform, fecal coliform, E. coli, and/or Enterobacteriaceae counts as potential indicators. We’re sure this sounds familiar so you’re feeling pretty good about now—well, read on, please. What may be less obvious is the potential usefulness of a yeast and mold count for this purpose, because low-level moisture intrusion may lead to growth of these fungal groups and also may lead to enhanced survival/growth of Salmonella. Therefore, one may find the best indicator by looking for an indicator of moisture control problems rather than an indicator of potential fecal contamination.

Finally, verification screening of all raw materials, ingredients, processes, environmental locations, and products using traditional microbiology tests can quickly become expensive if you are looking at all the potential indicators shown in the table. By first running a single microbiome on a specimen, the predominant microbes and their relative proportional populations will be determined. This knowledge can be used to develop appropriate targeted verification screening for indicators that you now know are relevant to the specimen. Furthermore, the impact of changes in suppliers, processes, or product formulation can be measured using microbiomes to again gain confidence that appropriate indicators are still being used.

We hope this installment of Food Genomics triggers the reader to rethink the indicators they are using and ask the following questions:

  • Why are we using our chosen indicators?
  • Are our indicators telling us what we really need to know?
  • Are there better indicators for my supplier verification program?
  • Are there better indicators for my process verification program?
  • Are there better indicators for my environmental monitoring program?
  • Are there better indicators that more accurately predict product shelf life?
Indicator Test Uses Microbiome
Aerobic Mesophilic Plate Count Estimate population of microbes able to grow at 35°C with air. Overall food quality indicator, shelf life/spoilage predictor Names of predominant microbes and relative proportions of each constituting the aerobic mesophilic population
Anaerobic Mesophilic Plate Count Estimate populations of microbes able to grow at 35°C without oxygen. Shelf life/spoilage predictor of vacuum packaged or modified atmosphere packaged foods. Names of predominant microbes and relative populations of each constituting the anaerobic mesophilic population
Standard Plate Count Similar to APC but used for dairy products. Estimate population of microbes able to grow at 30°C with air. Overall milk quality indicator, shelf life/spoilage predictor Names of predominant microbes and relative populations of each constituting the aerobic mesophilic population
Psychrotrophic Plate Count Estimate population of microbes able to grow at refrigerated temperatures (incubation temperature can vary from 5° to 15°C) with air. Shelf life/spoilage predictor Names of predominant microbes and relative populations of each constituting the aerobic psychrotrophic population
Anaerobic Psychrotrophic Plate Count Estimate population of microbes able to grow at refrigerated temperatures without oxygen. Refrigerated shelf life/spoilage predictor of vacuum or modified atmosphere packaged foods Names of predominant microbes and relative proportions of each constituting the anaerobic psychrotrophic population
Aerobic Thermophilic Plate Count Estimate population of bacterial spores able to grow at high storage temperatures (incubation temperature can vary but usually >45°C) in air or survive a thermal process. Indicator of process failure. Spoilage indicator of improperly hot held foods Names of predominant spores and relative proportions of each constituting the aerobic thermophilic population
Aerobic Mesophilic Spore Estimate population of bacterial spores able to grow at 35°C with air. May indicate possible Bacillus cereus risk. Names of predominant spores and relative proportions of each constituting the aerobic mesophilic spore population
Anaerobic Mesophilic Spore Count Estimate population of bacterial spores able to grow at 35°C without oxygen. Potential shelf life/spoilage indicator of vacuum or modified atmosphere packaged foods. May indicate possible Clostridium botulinum risk. Names of predominant spores and relative proportions of each constituting the anaerobic spore population
Aerobic Psychrophilic Spore Count Estimate population of bacterial spores able to grow at refrigeration temperature with air. Spoilage indicator of refrigerated foods Names of predominant spores and relative proportions of each constituting the aerobic psychrotrophic spore population
Anaerobic Psychrophilic Spore Count Estimate population of bacterial spores able to grow at refrigeration temperature without oxygen. Potential shelf life/spoilage indicator of refrigerated vacuum or modified atmosphere packaged foods. May indicate possible nonproteolytic Clostridium botulinum risk Names of predominant spores and relative proportions of each constituting the anaerobic psychrotrophic spore population
Aerobic Thermophilic Spore Count Estimate population of bacterial spores able to grow at high temperature in air. Spoilage indicator of heat processed foods. Names of predominant spores and relative proportions of each constituting the aerobic thermophilic spore population
Anaerobic Thermophilic Spore Count Estimate population of bacterial spores able to grow at high temperature without oxygen. Spoilage indicator of heat processed, vacuum or modified atmosphere packaged foods Names of predominant spores and relative proportions of each constituting the anaerobic thermophilic spore population
 Thermoduric Plate Count  Estimate population of microbes able to survive a pasteurization process. Used as a shelf life/spoilage predictor Names of predominant microbes and relative proportions surviving a thermal process
 Lactic Acid Bacteria Count  Estimate population of bacteria able to produce lactic acid during growth. Indicator of fermentation success or spoilage failure Names of predominant microbes and relative proportions that produce lactic acid
 Proteolytic Plate Count  Estimate population of microorganisms that produce protease enzymes. Indicator of putrefactive spoilage potential Names of predominant microbes and relative proportions that produce proteases
 Lipolytic Plate Count  Estimate population of microorganisms that produce lipase enzymes. Indicator of lipid hydrolytic rancidity spoilage potential Names of predominant microbes and relative proportions that produce lipases
 Saccharolytic Plate Count  Estimate population of microorganisms that produce amylase enzymes. Indicator of starch hydrolysis spoilage potential Names of predominant microbes and relative proportions that produce amylases
Pectinolytic Plate Count Estimate population of microorganisms that produce pectinase enzymes. Indicator of pectin hydrolysis spoilage potential Names of predominant microbes and relative proportions that produce pectinases
Aciduric Plate Count Estimate population of microorganisms able to grow in a high acid/low pH food. Indicator of spoilage potential Names of predominant microbes and relative proportions surviving an a high acid product
Aciduric Flat Sour Sporeformer Count Estimate population of bacterial spores able to tolerate high acid foods and produce acid without gas production. Indicator of high-acid canned food spoilage potential Names of predominant bacterial spores and relative proportions that grow in a high-acid canned food
Thermophilic Flat Sour Spore Former Count Estimate population of bacterial spores able to grow at high temperature and produce acid. Indicator of canned food spoilage potential Names of predominant bacterial spores and relative proportions that grow and produce acid in a canned food
Sulfide Sporeformer Count Estimate populations of bacterial spores that produce sulfur aroma compounds. Indicator of canned food spoilage potential Names of predominant bacterial spores that produce sulfur compounds
 Halophilic Plate Count  Estimate population of microorganisms able to grow at high salt concentrations. Indicator of microbes that can spoil low water activity foods  Names of predominant microbes and relative proportions that grow in a high-salt food
 Osmophilic Plate Count  Estimate population of microorganisms able to grow at high sugar concentrations. Indicator of microbes that can spoil low water activity foods  Names of predominant microbes and relative proportions that grow in a high-sugar food
 Yeast & Mold Count  Estimate population of fungal microbes. Indicator of fermentation success (mold-ripened cheeses) or spoilage potential  Names of predominant fungi and relative proportions in a specimen
 Preservative Resistant Yeast & Mold Count  Estimate population of fungi able to grow or survive in the presence of a food preservative. Indicator of spoilage potential  Names of predominant fungi and relative proportions that grow in the presence of a food preservative
 Coliform Count  Estimate population of bacteria able to ferment lactose at 35°C within 48 hours with gas production. Indictor of sanitation failure or possible presence of fecal pathogens  Names of predominant bacteria and relative proportions constituting the coliform population of a specimen
 Fecal Coliform Count  Estimate population of bacteria able to ferment lactose at 44°C within 48 hours with gas production. Indictor of possible presence of fecal pathogens  Names of predominant bacteria and relative proportions constituting the fecal coliform population of a specimen
 E. coli count  Estimate population of Escherichia coli. Indicator of fecal contamination and possible presence of enteric pathogens  A microbiome is not a useful addition for this species-specific test
 Total Enterobacteriaceae Count  Estimate population of bacteria able to ferment glucose at 35°C within 24 hours. Indictor of sanitation failure or possible presence of fecal pathogens Names of predominant bacteria and relative proportions constituting the Enterobacteriaceae population of a specimen
 Enterococcus Plate Count  Estimate population of enterococci. Indicator of possible fecal contamination  Names of predominant bacteria and relative proportions constituting the enterococci population of a specimen
 Listeria spp.  Estimate of the presence/absence of Listeria species in a food or environmental sample. Indicator of the possible presence of the pathogen Listeria monocytogenes  Names of predominant Listeria species and relative proportions constituting the Listeria population of a specimen
 Relative ATP Concentration  Estimate the amount of adenosine triphosphate in a food or environmental sample. Indicator for the presence of living cells (food or microbial). Used as an indicator of proper sanitation  Names of predominant microorganisms and relative proportions constituting the microbial population of a specimen taken at the same site swabbed for ATP
Table I. Common microbial indicator tests and the use of microbiomes for validation of effectiveness.1

 

References

  1. Salfinger, Y, and M.L. Tortorello. (2015). Compendium of Methods for the Microbiological Examination of Foods, 5th Ed. American Public Health Association, Washington, D.C.
  2. Zhao, T., M.R.S. Clavero, M. Doyle, and L.R. Beuchat. (1997). Health relevance of the presence of fecal coliforms in iced tea and leaf tea. J. Food Prot. 60(3):215-218.
Dr. Douglass Marshall, Chief Scientific Officer – Eurofins Microbiology Laboratories
Food Genomics

Microbiomes a Versatile Tool for FSMA Validation and Verification

By Douglas Marshall, Ph.D., Gregory Siragusa
No Comments
Dr. Douglass Marshall, Chief Scientific Officer – Eurofins Microbiology Laboratories

The use of genomics tools are valuable additions to companies seeking to meet and exceed validation and verification requirements for FSMA compliance (21 CFR 117.3). In this installment of Food Genomics, we present reasons why microbiome analyses are powerful tools for FSMA requirements currently and certainly in the future.

Recall in the first installment of Food Genomics we defined a microbiome as the community of microorganisms that inhabit a particular environment or sample. For example, a food plant’s microbiome includes all the microorganisms that colonize a plant’s surfaces and internal passages. This can be a targeted (amplicon sequencing-based) or a metagenome (whole shotgun metagenome-based) microbiome. Microbiome analysis can be carried out on processing plant environmental samples, raw ingredients, during shelf life or challenge studies, and in cases of overt spoilage.

As a refresher of FSMA requirements, here is a brief overview. Validation activities include obtaining and evaluating scientific and technical evidence that a control measure, combination of control measures, or the food safety plan as a whole, when properly implemented, is capable of effectively controlling the identified microbial hazards. In other words, can the food safety plan, when implemented, actually control the identified hazards? Verification activities include the application of methods, procedures, tests and other evaluations, in addition to monitoring, to determine whether a control measure or combination of control measures is or has been operating as intended, and to establish the validity of the food safety plan. Verification ensures that the controls in the food safety plan are actually being properly implemented in a way to control the hazards.

Validation establishes the scientific basis for food safety plan process preventive controls. Some examples include using scientific principles and data such as routine indicator microbiology, using expert opinions, conducting in-plant observations or tests, and challenging the process at the limits of its operating controls by conducting challenge studies. FSMA-required validation frequency first includes before the food safety plan is implemented (ideally), within the first 90 calendar days of production, or within a reasonable timeframe with written justification by the preventive controls qualified individual. Additional validation efforts must occur when a change in control measure(s) could impact efficacy or when reanalysis indicates the need.

FSMA requirements stipulate that validation is not required for food allergen preventive controls, sanitation preventive controls, supply-chain program, or recall plan effectiveness. Other preventive controls also may not require validation with written justification. Despite the lack of regulatory expectation, prudent processors may wish to validate these controls in the course of developing their food safety plan. For example, validating sanitation-related controls for pathogen and allergen controls of complex equipment and for how long a processing line can run between cleaning are obvious needs.

There are many routine verification activities expected of FSMA-compliant companies. For process verification, validation of effectiveness, checking equipment calibration, records review, and targeted sampling and testing are examples. Food allergen control verification includes label review and visual inspection of equipment; however, prudent manufacturers using equipment for both allergen-containing and allergen-free foods should consider targeted sampling and testing for allergens. Sanitation verification includes visual inspection of equipment, with environmental monitoring as needed for RTE foods exposed to the environment after processing and before packaging. Supply-chain verification should include second- and third-party audits and targeted sampling and testing. Additional verification activities include system verification, food safety plan reanalysis, third-party audits and internal audits.

Verification procedures should be designed to demonstrate that the food safety plan is consistently being implemented as written. Such procedures are required as appropriate to the food, facility and nature of the preventive control, and can include calibration of process monitoring and verification instruments, and targeted product and environmental monitoring testing.

Gregory Siragusa, Eurofins
Food Genomics

Microbiomes Move Standard Plate Count One Step Forward

By Gregory Siragusa, Douglas Marshall, Ph.D.
No Comments
Gregory Siragusa, Eurofins

Last month we introduced several food genomics terms including the microbiome. Recall that a microbiome is the community or population of microorganisms that inhabit a particular environment or sample. Recall that there are two broad types of microbiomes, a targeted (e.g., bacteria or fungi) or a metagenome (in which all DNA in a sample is sequenced, not just specific targets like bacteria or fungi). This month we would like to introduce the reader to uses of microbiomes and how they augment standard plate counts and move us into a new era in food microbiology. Before providing examples, it might be useful to review a diagram explaining the general flow of the process of determining a microbiome (See Figure 1).

Microbiome
Figure 1. General process for performing a targeted microbiome (bacterial or fungal)

By analogy, if one thinks of cultural microbiology and plate counts as a process of counting colonies of microbes that come from a food or environmental sample, microbiome analysis can be thought of as identifying and counting signature genes, such as the bacterial specific 16S gene, from the microbes in a food or environmental sample. Plate counts have been and remain a food microbiologist most powerful indicator tool in the tool kit; however, we know there are some limitations in their use. One limitation is that not all bacterial or fungal cells are capable of outgrowth and colony formation on specific media under a set of incubation conditions (temperature, time, media pH, storage atmosphere, etc.). Individual plate count methods cannot cover the nearly infinite number of variations of growth atmospheres and nutrients. Because of these limitations microbiologists understand that we have not cultured but many different types of bacteria on the planet (this led to the term “The Great Plate Count Anomaly” (Staley & Konopka, 1985). Think of a holiday party where guests were handed nametags on which was printed: “Hello, I grow on Standard Methods Agar” or “Hello, I grow at 15°C”, etc. We can group the partygoers by ability to grow on certain media; we can also count partygoers, but they still do not have names. As effective as our selective and differential media have become, bacterial colonies still do not come with their own “Hello, My Name Is XYZ” name tags. Therefore, in the lab, once a plate is counted it is generally tossed into the autoclave bag, along with unnamed colonies and all they represent. Microbiomes can provide a nametag of sorts as well as what proportion of people at that party share  a certain name. For instance: “Hello, My Name is Pseudomonas” or “Hello, My Name is Lactobacillus”, etc. The host can then say “Now we are going to count you; would all Pseudomonas pleased gather in this corner?” or “All Lactobacillus please meet at the punch bowl”.

A somewhat overly simplified analogy, but it makes the point that microbiome technology gives names and proportions. Microbiomes too have limitations. First, with current technologies microbiomes need a relatively large threshold of organisms of a specific group to appear in the microbiome pie chart— approximately 103. In theory, a colony on a plate of agar medium can be derived from a single cell or colony-forming unit (CFU). Not all amplified genes in a microbiome are necessarily from viable cells (A topic that will be covered later in this series of articles). Forming a colony on an agar surface on the other hand requires cell viability. Finally, the specificity of a microorganism name assigned to a group in a microbiome depends on the size of the sequenced amplicon (an amplicon is a segment of DNA, in this case the 16S gene DNA, resulting from amplification by PCR before sequencing) and how well our microbial databases cover different subtypes in a species. Targeted microbiomes can reliably name the genus of an organism, however resolution to the species and subspecies is not guaranteed. (Later in this series we will discuss metagenomes and how they have the potential to identify to a species or even subspecies level). Readers can find very informative reviews on microbiome specificity in the following cited references: Bokulich, Lewis, Boundy-Mills, & Mills, 2016; de Boer et al., 2015; Ercolini, 2013; Kergourlay, Taminiau, Daube, & Champomier Vergès, 2015.

When we consider the power of using cultural microbiology for quantitative functional indicators of microbial quality together with microbiomic analysis, with limitations  and all for both, microbiomes have opened a door to the vast and varied biosphere of our food’s microbiology to a depth never before observed. This all sounds great, but how will we benefit and use this information? We have constructed Table 1 with examples and links of microbiome applications to problems that would have required years to study by cultural microbiology techniques alone. Please note this is by no means an exhaustive list, but it serves to illustrate the very broad and deep potential of microbiomics to food microbiology. We encourage the reader to email the editors or authors with questions regarding any reference. Using PubMed and the search terms “Food AND microbiome” will provide abstracts and a large variety of applications of this technology.

Foodstuff Reference
Ale (Bokulich, Bamforth, & Mills, 2012)
Beef Burgers (Ferrocino et al., 2015)
Beefsteak (De Filippis, La Storia, Villani, & Ercolini, 2013)
Brewhouse and Ingredients (Bokulich et al., 2012)
Cheese (Wolfe, Button, Santarelli, & Dutton, 2014)
Cheese and Listeria growth (Callon, Retureau, Didienne, & Montel, 2014)
Cherries, Hydrostatic Pressure (del Árbol et al., n.d.)
Cocoa (Illeghems, De Vuyst, Papalexandratou, & Weckx, 2012)
Dairy Starters and Spoilage Bacteria (Stellato, De Filippis, La Storia, & Ercolini, 2015)
Drinking Water Biofilms (Chao, Mao, Wang, & Zhang, 2015)
Fermented Foods (Tamang, Watanabe, & Holzapfel, 2016)
Foodservice Surfaces (Stellato, La Storia, Cirillo, & Ercolini, 2015)
Fruit and Vegetables (Leff & Fierer, 2013)
Insect Protein (Garofalo et al., 2017)
Kitchen surfaces (Flores et al., 2013)
Lamb (Wang et al., 2016)
Lobster (Tirloni, Stella, Gennari, Colombo, & Bernardi, 2016)
Meat and storage atmosphere (Säde, Penttinen, Björkroth, & Hultman, 2017)
Meat spoilage and processing plant (Pothakos, Stellato, Ercolini, & Devlieghere, 2015)
Meat Spoilage Volatiles (Casaburi, Piombino, Nychas, Villani, & Ercolini, 2015)
Meat Stored in Different Atmospheres (Ercolini et al., 2011)
Milk (Quigley et al., 2011)
Milk and Cow Diet (Giello et al., n.d.)
 Milk and Mastitis  (Bhatt et al., 2012)
 Milk and Teat Preparation  (Doyle, Gleeson, O’Toole, & Cotter, 2016)
 Natural starter cultures  (Parente et al., 2016)
 Olives  (Abriouel, Benomar, Lucas, & Gálvez, 2011)
 Pork Sausage  (Benson et al., 2014)
Spores in complex foods (de Boer et al., 2015)
Tomato Plants (Ottesen et al., 2013)
Winemaking (Marzano et al., 2016)
Table 1. Examples of microbiome analysis of different foods and surfaces.

See page 2 for references