University of VermontScholarWorks @ UVM

Graduate College Dissertations and Theses Dissertations and Theses

2018

Analysis of surface crystals on soft washed rindcheeses using polarized light microscopy and theireffect on the sensory perception of grittinessPatrick PolowskyUniversity of Vermont

Follow this and additional works at: https://scholarworks.uvm.edu/graddis

Part of the Food Science Commons

This Thesis is brought to you for free and open access by the Dissertations and Theses at ScholarWorks @ UVM. It has been accepted for inclusion inGraduate College Dissertations and Theses by an authorized administrator of ScholarWorks @ UVM. For more information, please contactdonna.omalley@uvm.edu.

Recommended CitationPolowsky, Patrick, "Analysis of surface crystals on soft washed rind cheeses using polarized light microscopy and their effect on thesensory perception of grittiness" (2018). Graduate College Dissertations and Theses. 864.https://scholarworks.uvm.edu/graddis/864

ANALYSIS OF SURFACE CRYSTALS ON SOFT WASHED RIND CHEESES

USING POLARIZED LIGHT MICROSCOPY AND THEIR EFFECT ON THE

SENSORY PERCEPTION OF GRITTINESS

A Thesis Presented

by

Patrick Polowsky

to

The Faculty of the Graduate College

of

The University of Vermont

In Partial Fulfillment of the Requirements

for the Degree of Master of Science

Specializing in Nutrition and Food Sciences

May, 2018

Defense Date: March 19, 2018

Thesis Examination Committee:

Paul S. Kindstedt, Ph.D., Advisor

John M. Hughes, Ph.D., Chairperson

Catherine W. Donnelly, Ph.D.

Cynthia J. Forehand, Ph.D., Dean of the Graduate College

ABSTRACT

With the rising popularity of artisanal cheese in the United States, the soft washed

rind category has emerged as a fast-growing segment of the marketplace. There is much

anecdotal evidence to suggest a common sensorial defect in soft washed rind cheese is a

gritty/sandy texture attributed to crystal growth on the rind of these cheeses

A preliminary study was undertaken to develop a set of criteria to visually identify

crystals found on soft washed rind cheeses. Single crystal identities were presumptively

determined using polarized light microscopy (PLM), and cross-checked using powder x-

ray diffractometry (PXRD). Two distinct crystal groupings were determined based on

these metrics. Group 1 crystals had high birefringent coloring, angle of extinction (AE) ≈

90°, and were smaller and less circular than Group 2 crystals (P<0.05). Group 2 crystals

had no birefringent coloring and AE ≈ 18°. Using established mineralogical data, Group 1

and Group 2 crystals were identified to be struvite and ikaite, respectively. These

crystalline bodies are situated in an amorphous cheese matrix (i.e. smear), which create

difficulties when examining via PLM and PXRD, leading to high background noise. To

remedy these issues, a novel method for harvesting crystals was developed. Smear

scrapings were immersed in NaOH (pH=10) to dissolve smear material, which resulted in

improved PLM and PXRD performance.

A subsequent observational study was conducted to understand the prevalence of

surface crystals and grittiness associated with washed rind cheeses sourced from the U.S.A.

and Europe. Crystal types were identified via PXRD and PLM. Crystal size and shape

(circularity) metrics were determined via PLM and image analysis. A descriptive sensory

panel (n=12) was used to evaluate grittiness presence and intensity. Identified crystal types

included ikaite, struvite, calcite, and brushite. Mean crystal length and area ranged from

~30μm to ~1100μm, and ~500μm2 to ~200,000μm2, respectively. Sensory perception

threshold for grittiness occurred at a mean crystal length of ~70μm and mean crystal area

of ~2900μm2. Below these threshold levels, cheeses presented with negligible grittiness.

Above these threshold levels, grittiness was highly correlated with crystal length and area

(r=0.93 and 0.96, respectively; P<0.001). These results indicate surface crystals have a

direct impact on the sensorial quality of soft washed rind cheeses.

ii

ACKNOWLEDGEMENTS

I would like to express my sincere gratitude to my advisor, Dr. Paul Kindstedt, who

pushed my intellectual boundaries and shepherded me along my academic journey. I

consider it a great privilege to have had the chance to work with such a titan of the dairy

world.

Much appreciation and gratefulness to Dr. John Hughes, whose guidance and

know-how made this work possible. His passion and expertise for crystallography inspires

me to continue to seek ways that disparate fields can be intermingled.

Many thanks to Dr. Catherine Donnelly for her encouragement and providing

invaluable feedback along my journey.

I must recognize Dr. Gil Tansman for his friendship and mentorship. His

enthusiasm, curiosity, and knowledge was the solid foundation on which this work was

built.

Many thanks to Jasper Hill Farm and The Cellars at Jasper Hill for their support. I

could not ask for a better industry partner in conducting this work.

iii

TABLE OF CONTENTS

Acknowledgements ............................................................................................................. ii

List of Tables ..................................................................................................................... iv

List of Figures ..................................................................................................................... v

Chapter 1: Comprehensive Literature Review .................................................................... 1 Mineral Components of Milk and Cheese ...................................................................... 1 Microbiology and Physico-Chemistry of Washed Rind Cheeses ................................... 4 Basic Dynamics of Crystallization and Biomineralization ........................................... 12 Potential Surface Crystal Dynamics on Washed Rind Cheeses .................................... 17 Sensory Perception of Grittiness in Relation to Dairy Products ................................... 23 Historical Aspects of Cheese Crystals .......................................................................... 25 Basics of Polarized Light Microscopy .......................................................................... 31 References ..................................................................................................................... 35

Chapter 2: Characterization and presumptive identification of surface crystals on smear ripened cheese by polarized light microscopy ................................................ 47

Abstract ......................................................................................................................... 47 Introduction ................................................................................................................... 49 Materials And Methods ................................................................................................. 52 Results And Discussion ................................................................................................ 56 Conclusions ................................................................................................................... 63 Acknowledgments ......................................................................................................... 64 References ..................................................................................................................... 65

Chapter 3: Size, Shape, and Identity of Surface Crystals and their Relationship to Sensory Perception of Grittiness in American Artisanal and European PDO Soft Smear Ripened Cheeses ............................................................................................. 77

Abstract ......................................................................................................................... 77 Introduction ................................................................................................................... 78 Materials And Methods ................................................................................................. 81 Results ........................................................................................................................... 86 Discussion ..................................................................................................................... 89 Conclusion .................................................................................................................... 95 Acknowledgments ......................................................................................................... 95 References ..................................................................................................................... 96

Comprehensive Works Cited .......................................................................................... 111

Appendix A: Supplementary Diffractorgrams ................................................................ 126

Appendix B: Supplementary PCA Figures ..................................................................... 138

iv

LIST OF TABLES

Table 1. Mineral composition in milk. Adapted from Lucey and Horne 2009. ................. 1

Table 2. Mineral contents of common cheese varieties. Adapted from Zamberlin et al. 2012. ............................................................................................................................ 3

Table 3. Microbes isolated from rinds of smear ripened cheeses. Adapted from Wolfe et al. 2014 and Hohenegger et al. 2014. ...................................................................... 4

Table 4. Solubility, hardness, and crystal system of crystal phases found on washed rind cheese. Larger pK values indicate less soluble minerals. .................................. 17

Table 5. Crystal combination summary and explanations. ............................................... 21

Table 6. Select optical properties for isotropic and anisotropic (uniaxial and biaxial) crystals. Adapted from Carlton 2011. ........................................................................ 33

Table 7. Cheese sample compositional information. Results are from triplicate measurements from duplicate wheels of cheese. ....................................................... 68

Table 8. Crystal grouping results from image analysis of polarized light micrographs. N refers to the number of crystals analyzed in each grouping. ................................. 69

Table 9. Comparison of crystal size and shape between the two harvesting techniques. . 70

Table 10. Summary of crystal identification results from powder X-ray diffractometry (PXRD) and tentative identification from polarized light microscopy (PLM). ......... 71

Table 11. Attribute lexicon used for descriptive sensory analysis. . ............................... 100

Table 12. Production location and compositional information of cheeses surveyed in this study. Means and standard deviations (SD) are results from triplicate measurements from duplicate wheels of cheese. ..................................................... 101

Table 13. Summary of PLM data (crystal length, crystal area, circularity) and PXRD observations (crystals types). SD = standard deviation. .......................................... 102

Table 14. Summary of descriptive analysis data. Data are the means and standard deviations (SD) of panel duplicates. ........................................................................ 103

Table 15. Pearson correlation coefficients and correlation probabilities (in parentheses below) of crystal size/shape data and grittiness sensory intensity for cheeses above grittiness threshold. .......................................................................... 104

Table 16. Chi-square test results: phi-coefficients and (significance probabilities). ...... 105

v

LIST OF FIGURES

Figure 1. Schematic diagram of mineral partitioning in milk. ............................................ 2

Figure 2. Schematic view of reactions occurring in washed rind cheese during ripening. Adapted from McSweeney 2004. ................................................................. 9

Figure 3. Schematic representation of polarized light microscopy technique. ................. 31

Figure 4. Polarized light micrographs captured via the smear scraping method from cheeses A, B, C, and D. Examples of presumptive ikaite crystals labeled with solid arrows. Examples of presumptive struvite crystals labeled with open arrows. Scale bars represent ...................................................................................... 72

Figure 5. PXRD patterns of crystals isolated from cheeses A, B, C, and D via the smear scraping method. Diffractogram peaks are labeled as being characteristic of ikaite (square) or struvite (triangle). Y-axis is using arbitrary units (AU) of counts per second. ...................................................................................................... 73

Figure 6. PXRD patterns of crystals isolated from cheeses A, B, C, and D via the alkaline dispersion method. Diffractogram peaks are labeled as being characteristic of ikaite (square), struvite (triangle), or brushite (circle). Y-axis is using arbitrary units (A .............................................................................................. 74

Figure 7. Polarized light micrographs comparing image quality when the smear scraping technique is employed (A), and when the alkaline dispersion method is used (B). Scale bars represent 250µm. Viewed under quartz filter. .......................... 75

Figure 8. Polarized light micrograph captured using the alkaline dispersion method. Example of presumptive ikaite showing multi-generational growth. First generation (solid arrow) can be seen encompassed by second generation (striped arrow). Presumptive struvite showing possibility of resorption/dissolving into aqueous phase (open arrow). Scale bar represents 75µm. ......................................... 76

Figure 9. An example PXRD pattern captured from crystals isolated from cheese E. Characteristic peaks of ikaite and struvite are marked with squares and triangles, respectively. AU=arbitrary units. See Appendix A for diffractograms from each cheese. ...................................................................................................................... 106

Figure 10. Polarized light micrographs of crystals isolated from cheese smear material. Presumptive crystals of ikaite (solid arrows) and struvite (hollow arrows) are shown. Panel A contains crystals high in circularity (rounder crystals) while panel B contains crystals lower in circularity (more boxy or needle-like crystals). Scale bars represent 250µm. .................................................. 107

vi

Figure 11. Principal component analysis (PCA) bipot of cheese compositional data, sensory data, and crystal size/shape metrics. Letters represent cheese samples. PC 1 and PC 2 are shown. ....................................................................................... 108

Figure 12. Mean grittiness scores for cheeses used in this study. Error bars show one standard deviation from mean. Cheeses with grittiness value below sensory threshold had at least 75% of panelists indicate ”0” for grittiness. ......................... 109

Figure 13. Mean crystal area and mean crystal length measurements for cheeses used in this study. Error bars show one standard deviation from mean. .......................... 110

Figure 14. Contribution of variables to PC 1. Red line indicates 10% contribution. ..... 138

Figure 15. Contribution of variables to PC 2. Red line indicates 10% contribution. ..... 138

Figure 16. Contribution of variables to PC 3. Red line indicates 10% contribution. ..... 139

Figure 17. Contribution of variables to PC 4. Red line indicates 10% contribution. ..... 139

1

CHAPTER 1: COMPREHENSIVE LITERATURE REVIEW

Mineral Components of Milk and Cheese

MINERAL CONTENT OF MILK

The approximate mineral content of milk is 8-9g/L (Gaucheron 2005); however,

exact mineral content is dependent on species, breed, point in lactation cycle, and animal

nutrition/health (Lucey and Horne 2009). Milk contains many different minerals in various

forms. Major cationic minerals include potassium, calcium, sodium, and magnesium.

Major anionic minerals include phosphorus/phosphate (organic and inorganic), citrate,

chloride, and carbonate (Lucey and Horne 2009). Of present interest in this discussion are

the minerals calcium, phosphate, and magnesium. These have structural implications in

cheese texture development (Lucey and Fox 1993; Cooke and McSweeney 2017), and have

been identified in crystals occurring on the surface of soft washed rind cheeses (Tansman

et al. 2017b, c). Table 1 shows the general mineral composition of milk.

Table 1. Mineral composition in milk. Adapted from Lucey and Horne 2009.

Cationic

Mineral

Concentration

(mg/L) Anionic mineral

Concentration

(mg/L)

Potassium 1210-1680 Inorganic phosphorus (as PO4) 1800-2180

Calcium 1040-1280 Citrate 1320-2080

Sodium 350-600 Chloride 780-1200

Magnesium 100-150 Total phosphorus (PO4) (all forms) 930-1000

Carbonate (including CO2) ~200

MINERAL EQUILIBRIUM IN MILK AND CHEESE

Calcium, phosphate, and magnesium exist in soluble and insoluble forms in milk

and cheese (McSweeney 2004). Calcium and phosphate are primarily partitioned between

casein-associated insoluble colloidal calcium phosphate (CCP), and a soluble aqueous

phase (Figure 1). Roughly two-thirds of these minerals exist as CCP, while one-third is

2

soluble in milk (Lucey and Horne 2009). Magnesium follows a similar pattern, but occurs

at much lower concentrations than calcium and phosphate (Zamberlin et al. 2012). The

soluble forms of calcium and magnesium are often complexed with other minerals such as

citrates and phosphates (Cooke and McSweeney 2017).

Figure 1. Schematic diagram of mineral partitioning in milk.

This “pseudo-equilibrium” of insoluble and soluble minerals can be affected by

several factors, including acid development and the addition of chelation agents

(Gaucheron 2005). For example, during cheese manufacture, lower pH at whey drainage

will lead to lower overall calcium content due to this solubilization effect (McSweeney

2007). This can be seen in Table 2, where cheeses that develop low pHs during

manufacture, such as brie and cottage cheese, have lower calcium and magnesium contents

than cheeses which employ a sweeter make, such as edam and gouda. Calcium and

phosphate content can also vary widely within a single cheese variety (Kindstedt and

Kosikowski 1988), probably due to differences in the make schedule and acidification

profile.

3

Table 2. Mineral contents of common cheese varieties. Adapted from Zamberlin et al. 2012.

Mineral

(mg/100g)

Cheese

Brie Cheddar Cream Cottage Edam Feta Gouda

Sodium 700 670 300 380 1020 1440 910

Potassium 100 77 160 89 97 95 91

Chloride 1060 1030 480 550 1570 2350 1440

Calcium 540 720 98 73 770 360 740

Phosphorus 390 490 100 160 530 280 490

Magnesium 27 25 10 9 39 20 38

CCP can also be dissolved from the para-casein matrix during cheese ripening due

to acid development from starter culture activity, but this process is less efficient than

decalcification during manufacture (Hassan et al. 2004). This relationship between

soluble/insoluble minerals and pH is crucial when understanding how these minerals could

potentially form crystals during the cheese aging process. Minerals play an important role

in overall cheese structure and functionality. CCP cross-links the casein backbone of

cheese and provides rigidity to the overall protein assembly via electrostatic interactions

(Lucey and Horne 2009). As pH decreases, and more acidic environments develop, the

solubilization of calcium and phosphate results in structural changes to the casein

framework, often softening the cheese body. This solubilization and transfer of minerals

from the para-casein matrix to the water phase of the cheese results in a somewhat

demineralized curd body, depending on drain pH (Lucey and Fox 1993). Understanding

how minerals are partitioned in cheese is critical when trying to determine potential

downstream effects such as crystallization and textural softening. In washed rind cheese,

these processes are linked together and have tremendous impact on overall cheese quality.

4

Microbiology and Physico-Chemistry of Washed Rind Cheeses

SURFACE ECOLOGY OF WASHED RIND CHEESES

Soft washed rind cheeses are high in moisture, are aged in humid environments for

several weeks, and can become quite low in acidity. During this aging time, a microbe-

laden brine-based washing fluid is scrubbed into the surface. This process leads to the

formation of the characteristic red-orange smear at the cheese surface. The rinds of smear

ripened (i.e. washed rind) cheeses represent a complex ecology of microorganisms that

span multiple genera and species (Brennan et al. 2004). The growth patterns and

interactions of these microbes are now being better elucidated due to improved molecular

genetic techniques. These new insights have identified numerous genera of molds, yeasts,

and bacteria that colonize cheese rinds—many never previously associated with cheese

rinds (Quigley et al. 2012; Wolfe et al. 2014). Table 3 lists everal of the main microbial

genera found within the rinds of washed rind cheeses.

Table 3. Microbes isolated from rinds of smear ripened cheeses.

Adapted from Wolfe et al. 2014 and Hohenegger et al. 2014.

Bacteria Genera1 Fungi Genera1

Arthrobacter†

Brevibacterium†

Brachybacterium

Corynebacterium†

Halomonas

Microbacterium

Pseudomonas

Psychrobacter

Serratia

Sphingobacterium

Staphylococcus†

Vibrio

Aspergillus

Acremonium

Candida†

Chyrsosporium

Debaryomyces†

Fusarium

Galactomyces

Geotrichum†

Kluyveromyces

Penicillium†

Saccharomyces

Scopulariopsis *Not an exhaustive list; alphabetical order.†Major genera

5

While cheesemakers often use a prescribed blend of microbes in the production of

washed rind cheeses, many of those genera don’t colonize the surface to a major degree

(Feurer et al. 2004). This could be an indication of an even more complex ecological

environment than once thought. It has been established that many of the genera listed in

Table 1 can be found in various areas of a cheese manufacturing facility, as well as in

different manufacturing facilities entirely (Bokulich and Mills 2013; Wolfe et al. 2014).

This suggests that much of the microbial growth on washed rind cheeses may be a result

of indigenous environmental flora, and not necessarily due to direct inoculation.

These complex microbial networks can have many downstream effects on cheese

chemistry, quality, and flavor (Bockelmann 2011). The metabolisms of these microbes,

and their various extracellular lipases and proteases, results in the formation of various

chemical products. When considering the presence of surface crystals in the rinds of this

cheese variety, surface microbes almost certainly play a critical role. Fairly unusual crystal

types have been found almost universally on washed rind cheeses. Ikaite (CaCO3•6H2O)

and struvite (NH4MgPO4•6H2O) were just recently identified as major crystal phases

present on washed rind cheeses, with calcite (CaCO3) and brushite (CaHPO4•2H2O) also

being identified (Tansman et al. 2017a, c). These crystal types are not only a novel finding

in the dairy technology community, but are also novel to the food community at-large.

(Struvite has been previously identified in canned fish products (Amano 1950)) The surface

microbiota environment of washed rind cheeses is very unique in the food world, but many

of the same genera can be found on a wide number of cheeses (Brennan et al. 2004).

6

Therefore, it stands to reason these surface flora influence commonalities across washed

rind cheeses, perhaps crystallization.

BIOCHEMICAL EFFECTS OF SURFACE MICROBES

At the onset of ripening, washed rind cheeses are usually acidic (pH ≈ 5.0–5.6 )

(Brennan et al. 2004). This, combined with the high salt content of these cheeses, usually

prompt the growth of yeast species such as Debaryomyces hansenii (Bonaïti et al. 2004).

This genus, and other related microbes, consume lactic acid and deacidify the surface of

the cheese. A gaseous by-product of this reaction can include carbon dioxide (Bockelmann

2011). CO2 can also be formed due to various other microbial metabolisms, such as

heterofermentative Lactobacilli (Mullan 2000). Many of these surface microbes are also

strongly proteolytic and can create large amounts of ammonia (Karahadian et al. 1987).

This creates a less acidic environment, and pH-sensitive microbes like Corynebacterium

can grow and proliferate. Eventually the rinds of smear ripened cheese become fully

colonized with a menagerie of microbes, often reaching a pH near or above 7 (Irlinger et

al. 2015).

The production of these gaseous components leads to increase concentrations of

carbon dioxide and ammonia in the ripening headspace (Dumont et al. 1974; Gori et al.

2007). These compounds, and their effects on cheese ripening, have been studied at length

in white mold cheeses (Ribadeau-Dumas 1984; Picque et al. 2006), but research of their

effects on soft washed rind cheeses is limited (Tansman et al. 2017a, c). In camembert

cheese it has been observed that carbon dioxide concentration in the ripening headspace

can reach levels of over 20% if no air renewal occurs (Picque et al. 2006). These gaseous

7

components can also be absorbed into the water phase of the cheese body. Do Ngoc et al.

(1971) reported high levels of ammonia in camembert cheeses; they measured

approximately 7-8% of the total nitrogen content to be ammonia.

RADIAL MINERAL MIGRATION AND SOFTENING

While the creation of alkaline conditions and gaseous components (e.g. CO2 and

NH3) are direct effects of microbial metabolism, the accumulation of the other necessary

components of surface crystals involves a more complex phenomenon—radial mineral

migration and subsequent softening/liquefaction. While the softening process and mineral

migration in soft washed rind cheese has only been studied briefly (Tansman et al. 2017c),

a similar process has been studied at some length in surface mold ripened cheeses, such as

brie and camembert (McSweeney 2004). During the ripening of this variety, the surface

molds initiate similar changes as with washed rind cheeses, namely surface deacidification

via lactate metabolization, and proteolysis. The high surface pH values cause minerals such

as calcium and phosphate to precipitate and crystallize at the surface (Metche and Fanni

1978; Le Graet et al. 1983). These calcium phosphate crystals (i.e. brushite) have been

identified and studied previously (Brooker 1987; Tansman et al. 2017b).

These newly formed crystals deplete minerals from the water phase of the cheese,

establishing a concentration gradient. This prompts the migration of minerals from the

cheese center towards the surface, in a radial fashion (Amrane and Prigent 2008). This

results in the observed demineralization and of the curd body (Vassal et al. 1986; Le Graet

and Brule 1988; Tansman et al. 2017b). As soluble minerals migrate towards the cheese

surface, the “pseudo-equilibrium” of soluble/insoluble minerals in the casein matrix is

8

likely affected. Minerals solubilize and enter the water phase of the cheese. Key minerals

at play in this process are calcium and phosphate (and magnesium to a smaller degree). As

mentioned above, calcium and phosphate (namely CCP nanoclusters) are crucial structural

elements for the cheese body and their removal can have significant effects on texture

(Lucey and Fox 1993). The high pH environment, coupled with this demineralization, shift

the dominant interactions within the cheese from the type casein-casein to the type casein-

water. Water is absorbed by the casein matrix, which swells and leads to characteristic

softening and liquefaction. While this crystal-induced process has not been observed

directly in these cheeses, there is strong evidence based on related studies. Boutrou et al.

(1999) noted less juice expressed from camembert as the cheese aged, as well as a

concomitant increase in proteolysis products. Also noted was a decrease of calcium and

inorganic phosphate in the juice over time. This lends credence to the possibility of

increased casein-water interactions as a result solubilization and subsequent crystallization

of key structural minerals (i.e. calcium and phosphate). Guo and Kindstedt (1995) noted a

similar observation in mozzarella cheese, with a few crucial differences. They also

observed the amount of expressible serum (ES) diminishing over time but saw an overall

increase in mineral content of the ES. Without crystallization occurring, minerals

solubilized into the water phase of the cheese and exited with the serum. They also make

note of the possibility of salt (NaCl) having a similar effect on the casein matrix. Sodium

ions perhaps displacing calcium, and causing subsequent swelling and water uptake. These

downstream effects of casein-water interactions would help explain the reduced amount of

ES produced as these cheeses aged.

9

In conclusion, the production of gaseous components (CO2 and NH3) and induced

mineral migration (Ca, Mg, and PO4) due to surface microbial metabolic activity frees up

all the necessary components of the crystals types that have been identified on washed rind

cheeses thus far. The supposition is that these crystals allow for the radial softening

phenomena. Figure 2 presents a schematic view of these reactions.

Figure 2. Schematic view of reactions occurring in washed rind cheese during ripening.

Adapted from McSweeney 2004.

It is also worth mentioning that the minerals are not the only components of the

cheese migrating during ripening. Observations of white mold cheeses have indicated that

concentration gradients of lactate also occur due to its surface metabolization and

destruction, resulting in a net movement of lactate outwards towards the surface

10

(McSweeney 2004). Ammonia is a product of surface metabolism and can be absorbed by

the cheese body and slowly diffuse inwards (Spinnler and Gripon 2004). These factors,

combined with those already discussed, contribute to the pH gradient observed—

increasing pH moving outwards towards the cheese surface. Also of note are the effects of

proteolysis on free water at the cheese surface. Surface microbes are known to be very

proteolytic (Brennan et al. 2002). Since the cleavage of peptide bonds requires water,

protein breakdown can lower the total amount of free water. This can concentrate the

minerals in the water phase (Cooke and McSweeney 2017), and perhaps contribute to

surface crystallization.

The aforementioned surface proteolysis and deacidification process is not unique

to white mold cheese. It is important to remember that the action of the surface smear

bacteria of soft washed rind cheeses also results in increased pH values at the cheese

surface. Preliminary studies have shown that a similar crystallization and mineral migration

phenomenon also occurs in soft washed rind cheeses (Tansman et al. 2017c). A major point

of differentiation, however, is the type of crystals being formed. In these cheeses, calcium

carbonate (ikaite and calcite), magnesium ammonium phosphate (struvite), and calcium

phosphate (brushite) have all been observed (Tansman et al. 2017a, c), while only brushite

has been observed on white mold cheeses (Tansman et al. 2017b).

EFFECTS OF STARTER CULTURE

The surface microbiota is only one part of the microbial ecology associated with

washed rind cheeses. Starter culture microorganisms have also been isolated form washed

rind cheeses and washed rind cheese aging spaces (Bokulich and Mills 2013). Common

11

starter culture species include both mesophilic and thermophilic types, such as Lactococcus

lactis, Streptococcus thermophilus, and Lactobacillus helveticus (Johnson 2014). Of note

is that the number of viable starter culture microbes often diminishes greatly during cheese

aging, and other NSLAB dominate (Bokulich and Mills 2013; Blaya et al. 2017). The

relatively low number of starter microbes compared to that of smear microbes probably

indicates that their active role in influencing crystal development is limited. However, the

acid production associated with their metabolism is of crucial importance to cheesemaking

and the resulting mineral content. Cheeses with greater acidification have a lower mineral

content (Zamberlin et al. 2012), and therefore fewer raw materials for crystal formation.

Tyrosine crystals are common in cheeses such as Parmigianino Reggiano (Tansman et al.

2015a). Certain starter culture microbes, such as L. helveticus, are known to be strongly

proteolytic and result in large amounts of free tyrosine (Nakajima et al. 1998; Griffiths and

Tellez 2013). Tyrosine is quite insoluble and will precipitate at a pH range found in most

cheeses (Hitchcock 1924). Whether the presence of tyrosine on washed rind cheeses can

be attributed to L. helveticus, highly proteolytic smear microbes, or a combination of both,

remains an unanswered question.

12

Basic Dynamics of Crystallization and Biomineralization

CRYSTAL FORMATION

It is generally accepted that crystallization advances through several stages:

supersaturation, nucleation, and growth (Naka and Chujo 2001; Hartel 2013).

Supersaturation of a solute phase within a solvent generates a driving force for

crystallization, which is heavily dependent on the temperature of the system (Mullin 2001).

Nucleation is generally bifurcated into two regimes: homogeneous and heterogeneous.

Homogeneous nucleation occurs without preferential nucleation sites, while heterogeneous

nucleation occurs at nucleation sites, such as foreign particles or air bubbles (Diao et al.

2011). Recent advents in crystallization research have developed a two-step model for

homogenous nucleation (Hartel 2001). Step one consists of the formation of amorphous

clusters of molecules, which possess some short-order organization, but are not situated in

a crystal lattice yet. These clusters aggregate, grow, and reach a critical size. Step two

consists of reorganization of atoms in these clusters into a crystal lattice (Erdemir et al.

2009). The mechanism(s) of heterogeneous nucleation is still poorly understood, and an

active area of research in surface chemistry (Diao et al. 2011). Crystal growth can occur

due to aggregation of smaller crystals, or by deposition of molecules/atoms into a growing

crystal lattice (Hartel 2013).

Crystallization can be influenced by many different factors: solute concentration,

temperature, pressure, pH, moisture, and the availability of nucleation sites (Agarwal et al.

2006; Tian et al. 2010). Varying any of these factors could influence the type of crystal that

is formed, and can also result in the formation of metastable crystals (Bloss 1994). Several

13

of the crystals identified on washed rind cheeses thus far have been a metastable form,

namely the crystal hydrates of calcium carbonate (i.e. ikaite), magnesium ammonium

phosphate (i.e. struvite), and calcium phosphate (i.e. brushite). Calcite, the stable form of

calcium carbonate, has also been observed (Tansman et al. 2017a, c).

FACTORS EFFECTING CRYSTALLIZATION

The formation of certain single crystal species does not necessarily happen directly

from a supersaturated solution. Amorphous precipitation can occur when solute

concentration is very high, and then crystallization can proceed through various

metastable/stable phases (Luft et al. 2011). This has been observed in calcium carbonate

solutions, where amorphous calcium carbonate was formed first, followed by various

metastable forms (Clarkson et al. 1992). In some crystal systems, the growth matrix and

solute concentration can affect how crystallization proceeds (Naka and Chujo 2001).

Numerous chemical kinetic phenomena can also cause metastable forms to preferentially

crystallize, which has been observed during brushite crystallization (Arifuzzaman and

Rohani 2004). Crystallization can also be influenced by the presence of mineral

components that provide a stabilizing effect, but do not take an active role in the crystal

lattice. This effect has been observed in calcium carbonate crystallization, where

metastable ikaite has formed due to the stabilizing effects of phosphates. Stable forms of

crystals, such as calcite, can form from metastable crystal hydrates such as ikaite due to

dehydration of the crystal lattice (Buchardt et al. 2001). It is also possible for

recrystallization to occur. Recrystallization is the formation of a crystal due to the

incomplete/complete breakdown and solubilization of solutes from a different crystal

14

(Gorski and Fantle 2017). Amano (1950) observed sodium phosphate crystals

(Na2HPO4•12H2O) in salmon, and theorized their occurrence resulted from the

recrystallization of struvite crystals, which commonly occur in fish products. The full scope

and extent to which crystallization can be influenced by reaction conditions is outside the

scope of this discussion, but the various factors most certainly play a central role in the

formation of crystals in soft washed rind cheeses.

BIOMINERALIZATION

Biomineralization is the overarching term referring to the process by which living

organisms sequester and bind minerals (Crichton 2012). These types of phenomena are

relatively common in biological systems. Mollusk shells, egg shells, bones, and teeth are

all examples of end products of biomineralization (Cusack and Freer 2008). A subset of

biomineralization is microbial or bacterial biomineralization, where bacteria take an active

role in sequestering minerals as part of their metabolic processes (Konhauser et al. 2008).

This process was first observed by Beveridge and Murray (1976), where a Bacillus subtilis

strain isolated from soil was observed chelating metal ions from growth media. The bound

minerals formed dense aggregates within the cell wall. Later work demonstrated that

negatively-charge carboxyl groups found within Gram-negative bacteria, such as B.

subtilis, were responsible for the binding of metal cations (Beveridge and Murray 1980).

On the other hand, anionic phosphate groups were shown to bind metal ions in cell

membranes by Gram-positive species, such as E. coli (Ferris and Beveridge 1986). This

pioneering work laid the foundation for many metal staining techniques that are used in

transmission electron microscopy (Konhauser et al. 2008).

15

Of interest to the present discussion are instances of pertinent crystals types being

formed via biomineralization. While a large body of research relating to biomineralization

exists within the geological and biological sciences (Konhauser et al. 2008), no research

has been conducted to study its possible relationship to crystallization in cheese. Calcium

carbonate crystals, such as the types found on washed rind cheeses (ikaite and calcite), are

known to occur due to microbial biomineralization. A strain of B. linens associated with

marine environments has been observed forming calcium carbonate precipitates (Zhu et al.

2017). Upon genomic analysis, the authors posited that the precipitation of calcium

carbonate was involved with the quorum sensing effects of free Ca2+ ions. Sanchez-Roman

et al. (2007) observed the formation of both carbonate and phosphate crystals (calcite and

struvite) by several halophilic bacterial genera: Halomonas, Salinivibrio, Marinomonas,

and Marinobacter. Many different species of Halomonas have been observed forming

crystal precipitates, such as calcite and struvite. Rivadeneyra et al. (2006) determined that

over twenty Halomonas species can form crystals. The concentration of salt (NaCl) and

the Mg2+-to-Ca2+ ratio were determining factors in what type of crystals were formed.

Brevibacterium Linens, Halomonas, and Marinomonas have been identified on the rinds

of smear ripened cheeses previously (Wolfe et al. 2014; Hohenegger et al. 2014). Bacterial

cells (Stocks-Fischer et al. 1999; Anbu et al. 2016), fungal hyphae (Luo et al. 2018), and

excreted organic matrices (Dupraz et al. 2009) can serve as nucleation sites for calcium

carbonate minerals. Microbially-induced crystallization of struvite has also been observed

by other bacterial genera (Ben Omar et al. 1998; Prywer and Torzewska 2010). Marine

strains of E. coli have been known to precipitate calcium phosphate (Cosmidis et al. 2015).

16

The association between marine microbes and these crystals types is fascinating,

considering that these crystal types have been previously identified in marine

environments: struvite in fish products (Amano 1950) and wastewater reservoirs (Wu et al.

2005), and ikaite in artic sea water environments (Rysgaard et al. 2012). The dual

occurrence of microbes and crystals in these contexts hint at a commonality between cheese

rinds and marine environments. Wolfe et al. (2014) suggested the use of sea salt might play

a role, as well as the unique chemical conditions of cheese rinds.

17

Potential Surface Crystal Dynamics on Washed Rind Cheeses

The various physico-chemical changes described in the previous sections result in

the formation of a range of solutes which can crystallize when certain conditions are met.

Saturation of the water phase of the cheese and a suitable nucleation environment are

important factors when considering if, and what, crystals will form. Another important

factor to consider is the relative solubilities of the various crystal phases that have been

identified on the surface of washed rind cheeses. As already discussed, previous work has

identified four crystal types on soft washed rind cheeses: brushite, struvite, ikaite, and

calcite (Tansman et al. 2017c). Table 4 shows an overview of solubility constants, Mohs

hardness values, and the crystal systems. It’s important to note that the solubility constants

in Table 4 reflect neutral pH conditions, and will not necessarily correlate with how these

crystals occur in cheese.

Table 4. Solubility, hardness, and crystal system of crystal phases found on

washed rind cheese. Larger pK values indicate less soluble minerals.

Crystal Type Solubility

(pK @ 25°C)

Mohs

Hardness5

Crystal

System5

Brushite 6.591 2.5 Monoclinic

Struvite 13.152 1.5-2.0 Orthorhombic

Ikaite 6.593 3.0 Monoclinic

Calcite 8.484 3.0 Trigonal 1Jokanovic et al. 2014

2Wu et al. 2005 3Bischoff et al. 1993

4Plummer and Busenberg 1982 5Hudson Institute of Mineralogy 2018

While a single type of crystal can be found on soft washed rind cheeses, a

combination of crystals is much more common (Tansman et al. 2017c). By understanding

the crystal formation dynamic individually, we can then begin to understand how

combination of crystals can occur as soft wash rind cheeses age and ripen.

18

BRUSHITE

Brushite, and calcium phosphate in general, is perhaps the most studied crystal type

that occurs on surface ripened cheeses (Le Graet et al. 1983; Brooker 1987; Tansman et al.

2017b). Its mineral components are plentiful in milk and cheese. The high concentration

of these minerals, the aforesaid migration phenomenon, and high surface pH values surely

contribute to the formation of this crystal type. Brushite becomes less soluble at higher pH

values, leveling off around pH = 8 (S-A Johnsson and Nancollas 1992).

The reason(s) why brushite, a metastable form of calcium phosphate, has been

identified in washed rind cheese, but not a more stable form remains unanswered and

warrants further study. The occurrence of this metastable phase highlights the complexities

involved with these crystallization reactions. Perhaps a biomineralization phenomenon is

occurring which preferentially forms brushite. Tansman et al. (2017c) noted a sluggish

accumulation of phosphate on the rind compare to calcium. This was given as a possible

reason for the disappearance of brushite when other crystal phases emerged.

STRUVITE

The mineral origins of struvite result from microbial metabolism (ammonia

dissolving into the water phase forming ammonium), and the accumulation of minerals at

the cheese surface (magnesium and phosphate). As already mentioned, ammonia can be

produced in very high amounts during cheese ripening, which readily dissolves into the

cheese (Do Ngoc et al. 1971). While magnesium is found in much lower quantities in

cheese than calcium (Cooke and McSweeney 2017), struvite has a much lower solubility

than the other calcium-containing minerals found in soft washed rind cheeses (Table 4).

19

Struvite is known to crystallize in a pH range of 6.0 to 13.0 (Kim et al. 2016). Many soft

washed rind cheeses fall well within this pH range (Bockelmann 2011). Biomineralization

may also play a role (Sanchez-Roman et al. 2007).

IKAITE

The formation of ikaite, similar to that of struvite, results from the formation of a

gaseous microbial metabolite (carbon dioxide dissolving into the water phase forming

carbonate), and surface accumulation of a key mineral component—calcium. The buildup

of these components surely exceeds the solubility threshold and they crystallize out,

forming ikaite.

The relatively high moisture of these cheeses and the smear that forms atop them

(Brennan et al. 2004), provide an ample opportunity for the hydrated crystal form to grow.

Also, as already mentioned, the presence of inorganic phosphate can encourage preferential

crystallization of ikaite, as opposed to the usually more stable calcite (Hu et al. 2014).

Solubility of ikaite decreases as pH increases (Giannimaras and Koutsoukos 1987).

Therefore, the gradual alkalization of cheese rinds and accumulation of phosphate could

encourage ikaite crystal formation. Biomineralization may also play a role, as mentioned

above.

CALCITE

The formation of calcite is directly related to formation of ikaite. It’s been shown

that metastable ikaite will dehydrate and change into calcite (Marland 1975). Calcite, as

expected, is also less soluble than its crystal hydrate: ikaite (Table 4). As with ikaite, calcite

20

is also less soluble at higher pHs. Tansman et al. (2017c) showed that calcite can form

before and/or alongside ikaite. This is difficult to reconcile with the accepted ikaite to

calcite pathway, which hints at a complex crystallization system occurring, perhaps with

biomineralization or recrystallization playing a role.

CRYSTAL TYPE AND EFFECT ON GRITTINESS

Although no previous studies have been conducted to determine if a certain crystal

type influences grittiness more than another, we can speculate based on available

information (Table 4). Struvite has a lower Mohs hardness value than the other relevant

crystal types, as well as an orthorhombic structure. The other crystal types (i.e. ikaite and

calcite) have greater Mohs hardness and a monoclinic or trigonal structure. It’s possible

that the differences in size and shape among these crystal types could influence the gritty

mouthfeel associated with their presence.

CRYSTAL COMBINATIONS SUMMARY

There are five mineral components for all currently known surface-situated crystals

on soft washed rind cheeses: calcium, magnesium, ammonia, carbon dioxide, and

phosphate. Table 5 provides a brief overview of the possible crystal dynamics when

multiple types of crystals are present concurrently.

21

Table 5. Crystal combination summary and explanations.

Crystal Phase(s) Mineral Components* Possible Explanation

1 struvite only Mg, NH3, PO4 Mineral migration, microbial metabolism, struvite’s

very low solubility

2 ikaite only Ca, CO2 Mineral migration, microbial metabolism, high amount

of Ca

3 struvite + ikaite Ca, Mg, NH3, CO2, PO4 Combination of 1 and 2 above, stabilizing effect of

phosphates on ikaite

4 struvite + brushite Ca, Mg, NH3, PO4 Like 1 above, lack of CO2 production, excess phosphate

5 struvite + calcite Ca, Mg, NH3, CO2, PO4 Like 3 above, dehydration of ikaite and/or less PO4 for

stability

6 ikaite + struvite +

brushite Ca, Mg, NH3, CO2, PO4 Like 3 above, with excess Ca and PO4

7 ikaite + struvite +

calcite Ca, Mg, NH3, CO2, PO4

Like 3 above, dehydration of ikaite and/or less PO4 for

stability due to advanced struvite formation

8 ikaite + struvite +

calcite + brushite Ca, Mg, NH3, CO2, PO4

Advanced demineralization, advanced microbial

metabolites, prolonged aging, combination of what’s

discussed above *Water not shown

POTENTIAL METHODS TO CONTROL CRYSTAL GROWTH

Several possible ways to regulate crystal growth on washed rind cheese include:

chelation of key mineral components, altering biomineralization phenomena, and

modification of crystal size. The relevant crystal types are all carbonate or phosphate

minerals containing calcium or magnesium. The use of a chelation agent could bind

calcium or magnesium, preventing the formation of crystals. Sodium gluconate is used for

this purpose in cheddar cheese, to limit calcium lactate crystal formation (Phadungath and

Metzger 2011). In a similar vein, addition of crystal inhibitors have been used in ice cream

to prevent excessive ice crystal growth (Damodaran and Wang 2017). As already

discussed, certain microbe species have been observed forming crystals such as ikaite and

struvite (Rivadeneyra et al. 2006). The same species and crystal types have been identified

on washed rind cheeses (Wolfe et al. 2014; Tansman et al. 2017c). Therefore, these

microbes and similar species might be influencing crystal growth. Perhaps targeting the

22

destruction of these microbes, or encouraging the growth of a competitive non-crystal-

forming species could control crystal formation. This may not be feasible due to many of

the pertinent microbes being indigenous to the cheesemaking environment (Bokulich and

Mills 2013). Finally, controlling crystal size may be one method to influence grittiness

perception. It is well established that the rapid growth of a large number of crystals usually

results in very small crystals. While on the other hand, slow growth of fewer crystals results

in larger entities. Moreover, numerous crystals of smaller size might be advantageous to

preventing grittiness. The use of seed crystals has been used to accomplish this in products

such as ice cream (Nickerson 1954). Advanced techniques such as sonocrystallization via

exposure to ultrasound have been used to control crystallization in food systems. The input

of vibrational energy can encourage nucleation and growth (Hartel 2013).

Sonocrystallization has been used to control crystallization of lactose crystals in dried

concentrated cheese whey (Zisu et al. 2014), ice crystals in ice cream (Zheng and Sun

2006), and fat crystallization in anhydrous milk fat powder (Martini et al. 2008). In

conclusion, the aforementioned techniques could all be used to control crystal development

on washed rind cheeses, thereby governing the resulting grittiness.

23

Sensory Perception of Grittiness in Relation to Dairy Products

ORAL PERCEPTION OF PARTICLES AND GRITTINESS

Texture stimuli that are associated with the mouth are collectively known as oral-

tactile sensory sensations (Lawless and Heymann 2010). This includes all interactions

between food and mouth, tongue, teeth, and other oral mucosa. An important part of oral-

tactile sensory response are tribological effects. Tribology refers to the study of friction

and lubrication of interfacing surfaces in motion. In oral processing, interfacing surfaces

include food-tongue, food-teeth, teeth-teeth, and others (Stokes et al. 2013). Of interest to

the present discussion is the perception of particles in the oral-tactile system and the

resulting grittiness. Physiological differences and the tribological effects of the food

matrix/saliva can influence the perception of particles (Guinard and Mazzucchelli 1996).

To further confound the discussion, the size, shape, hardness, and concentration of particles

can also influence the perception of grittiness (Hough et al. 1990; Engelen et al. 2005;

Lawless and Heymann 2010). Tyle (1993) demonstrated that flat particles of varying

hardness were imperceptible up to a threshold of about 80μm, while angular particles were

perceived as gritty in a range above 11μm to 22μm. In other studies, particles as small as

3μm could be perceived, with particles smaller than that contributing to a creamy mouthfeel

(Lawless and Heymann 2010). These large variations in threshold size highlight the

complex nature of the perception of grittiness. Not only can the particles themselves

influence grittiness, the matrix in which they’re found can also influence overall texture.

Larger samples sizes can lead to greater perceived hardness (Cardello and Segars 1989).

Dan et al. (2008) noted that biting dynamics and how samples are chewed can impact oral-

tactile sensory perception as well.

24

GRITTINESS IN DAIRY PRODUCTS AND CRYSTAL SIZE THRESHOLD

Grittiness, graininess, or sandiness have been noted in many different dairy

products: ice and lactose crystals in ice cream (Donhowe et al. 1991; Fidaleo et al. 2017),

protein particles in cream cheese (Sainani et al. 2004), enriched cheddar cheese (Torri et

al. 2016), lactose crystals in Dulce de Leche (Hough et al. 1990), lactose crystals in brown

whey cheese (Jelen 1992), and salt (NaCl) crystals in butter (Hunziker 1920). Hough et al.

(1990) noted a detection threshold range of 45μm to 105μm for sandiness in Dulce de

Leche, depending on the concentration of lactose crystals present in sample. Fewer crystals

of larger size induced a comparable grittiness sensation as smaller more numerous crystals.

A similar threshold of 45μm has been used in the ice cream industry; ice crystals larger

than this size causing grittiness (Goff & Hartel, 2013). In Norwegian brown whey cheese

(i.e Brunost), grittiness can occur when lactose crystals get larger than 100μm (Skeie and

Abrahamsen 2017). In related products such as margarine spreads and chocolate, a crystal

size threshold of approximately 20μm has been used for grittiness occurrence (Vaisey-

Genser et al. 1989; Brooker 1990). As discussed above, crystal size is only one property

that may influence the sensory perception of grittiness. The number of crystals can also

play a role. Sainani et al. (2004) concluded that the grittiness caused by protein particles in

cream cheese can be intensified by a greater number of particles, as well as particles of a

larger size. The viscosity of the matrix in which the crystals are embedded can also play a

role. Lopez et al. (2016) observed the masking of particle perception with increased

viscosity of the dispersion matrix. This could mean the firmness and amount of softening

of a specific cheese may influence perception of crystal-induced grittiness.

25

Historical Aspects of Cheese Crystals

HISTORICAL PRESENCE OF CHEESE CRYSTALS

The first mention of cheese crystals in the American dairy literature probably

occurred in Wisconsin, around the start of the 20th century. Babcock and Russell (1901),

of Wisconsin’s Agriculture Experiment Station, noted that cheddar cheese cured at low

temperatures (<50°F) developed “white specks”. Upon more thorough inspection, many

cheeses kept in cold storage were found to contain these specks. They mentioned that the

usual practice of the era was to avoid storage of cheese at cold temperatures (<50°F).

Cheesemakers believed (erroneously) such treatment would result in a bitter product.

Interestingly, the work by Babcock and Russell in implementing lower storage

temperatures to improve cheese quality probably encouraged the growth of these “white

specks”. A year later, Babcock et al. (1902) performed a series of experiments in order to

determine the factors affecting formation of these specks. They concluded that storage

temperature and salt content of the finished cheese were the main factors determining speck

formation. High storage temperatures (>60°F) and high salt content seemed to limit the

formation of these white specks. Looking at these experiments through the lens of modern

science, it’s highly likely that they were observing the first wide-spread occurrence of

calcium lactate crystals. Cooler temperatures can result in lower solubility and more

pronounced crystallization. Higher salt content will lead to a larger salt-to-moisture ratio

of the final cheese. This can affect starter culture metabolism and limit the growth of these

microbes. With reduced starter culture activity, less acid is developed via fermentation, and

the resulting cheese has a higher pH. These effects were confirmed to reduce calcium

lactate formation over a century later (Agarwal et al. 2008). The first work done to

26

determine the composition of these “white specks” was likely performed at New York

State’s Agricultural Experiment Station by Van Slyke and Publow (1909). Through

chemical analysis, they determined calcium was present within these crystals. Based on the

waxy texture, fat/fatty acids were thought to be another major substituent. The combination

of these phases led Van Slyke to postulate that these crystals were “calcium soaps”. To

account for their occurrence at lower temperatures, Van Slyke made the supposition that

fat was being broken down by “bacteria acting only at lower temperatures” (Van Slyke and

Publow 1909). Up to this point, much of the knowledge of these “white specks” were

isolated in research communities. Researchers noted that the specks were innocuous, and

largely not noticed by the cheese industry. In 1910, “white specks” were described in an

article entitled Cheese Defects in the trade publication The New York Produce Review and

American Creamery (Monrad 1910). This publication was the leading trade journal at the

time, and was in large circulation among the dairy industry. Around the same time, Arthur

Dox of University of Connecticut’s Storrs Agricultural Experiment Station, claimed he

found the first occurrence of tyrosine crystals in cheese (Dox 1911); he noted white

crystalline entities in Roquefort bleu cheese. This determination was made through

exclusionary testing, and the positive chemical identification of phenol groups.

Throughout the early observations of these crystals, visual inspection and

rudimentary chemical testing were the primary analysis methods. By the 1920s,

microscopy techniques began to be employed to further study cheese crystals. Professor

Otakar Laxa, of the University of Chemical Technology in Prague, conducted extensive

microscopic studies of cheese that included many mentions and sketches of crystals (Laxa

27

1926). Laxa noted the presence of leucine, tyrosine, and lactate of lime (calcium lactate)

across many different cheese varieties; crystals appeared to be endemic in alpine-style

cheeses, grana-style cheeses, blue cheeses, and others. A year later, Laxa further studied

the morphological differences of tyrosine and leucine in cheese (Laxa 1927). He indicated

tyrosine, leucine, and their anhydrous forms all had unique structures that could be

identified via microscopy. Several years later, H. H. Sommer of the University of

Wisconsin, captured micrographs of calcium tartrate crystals in processed cheese (Sommer

1930). He concluded this was the cause of sandiness in processed cheese, a common defect

of the period.

Rather serendipitously during this same time period, the discovery and novel use

of X-rays was underway. In 1895, Wilhelm Röntgen observed the florescent effects of

“invisible” cathode rays. This, ultimately, turned out to be the discovery of X-rays (i.e.

Röntgen rays). Shortly thereafter, Sir William Lawrence Bragg and his father developed

Bragg’s Law, describing the interaction of X-rays and crystals (Bragg and Bragg 1913).

This would lay the foundation for X-ray diffraction, and create a field that has been

responsible for some the greatest scientific discoveries of the modern era.

USE OF POWDER X-RAY DIFFRACTOMETRY ON CHEESE

The earliest mentions of X-ray technology used in an agricultural setting were most

likely by Professor George L. Clark of the University of Illinois. (formerly of M.I.T.) He

mentions the use of X-ray imaging by cheesemakers to track the development of eyes (i.e.

holes) in Swiss cheese (Clark 1929). Not too long after, Clark and dairy scientist

collaborators explored the use of powder X-ray diffractometry (PXRD) in dairy

28

technology. Crystalline substances analyzed included: mineral deposits in dairy equipment

and lactose crystals in milk powder (Tuckey et al. 1934). The same group performed PXRD

studies on casein and cheddar cheese, and in the process generated reference patterns (d

values) for crystals of calcium phosphate and calcium lactate (Tuckey et al. 1938a, b). This

led to the first positive identification of the white specks in cheddar, via PXRD, to be

calcium lactate (Tuckey et al. 1938c). McDowall and McDowell (1939) confirmed this

result via chemical analysis, and further stated that the pentahydrate form of the calcium

lactate was identified in cheddar cheese. Shock et al. (1941) published results that indicated

the crystals found within cheddar, via PXRD, to be a mixture of calcium lactate and

tyrosine. The presence of tyrosine may have been due to the advanced age (>2 years) of

the samples they analyzed, which would have had large amounts of proteolytic products.

Dorn and Dahlberg (1942) analyzed canned cheddar cheese and concluded that the crystals

present throughout the body and cheese surface were tyrosine, with a possible calcium

phosphate “impurity”. They conjectured that the previous reports of calcium lactate may

have been due to experimental error, or trace amounts being an impurity contained within

the tyrosine crystals. Although no analytical techniques were used, Jacquet and Saingt

(1952) observed crystals on the surface of a camembert cheese, embedded within a

bacterial (Brevibacterium linens) smear that was present. This may be the first documented

case of finding surface crystals in washed rind (i.e. smear ripened) cheese. They assumed

the crystals to be tyrosine and leucine. Further PXRD studies confirmed the presence of

calcium lactate and tyrosine in cheddar cheese, with the amino acid cysteine also being

identified (Harper et al. 1953). Authors of the time indicated that the chief limitation of

29

PXRD was the need for a crystal concentration of at least 10% in order to be accurately

identified (Harper et al. 1953), which is in stark contrast to the standards of today. This

necessitated the need for chemical identification to be paired with PXRD studies. This

prompted the work by Swiatek and Jaworski (1959), who used histochemical techniques

to identify calcium phosphate in a wide variety of cheese types: Emmental, Tilsit, Trappist,

and Roquefort. Conochie et al. (1960) further used PXRD to identify calcium lactate and

tyrosine in well-aged cheddars. The first PXRD-confirmed presence of brushite was

reported by the same group not too much later (Conochie and Sutherland 1965). They

concluded that the presence of brushite was one of the principal causes of the seaminess

defect in cheddar cheese. Scharpf and Michnick (1967) demonstrated the first use of PXRD

to identify crystals in situ. Processed cheese was loaded into the diffractometer and a

diffractogram was generated. Crystal peaks indicative of disodium phosphate

dodecahydrate were identified, likely caused by the addition of emulsifying salts during

the manufacture of processed cheese. By the 1970s and 1980s, techniques such as PXRD

and electron microscopy were in use to identify cheese crystals (Brooker et al. 1975b;

Severn et al. 1986). The use of PXRD continues into the 21st century, with PXRD being

used to confirm identities of cheese crystals—mainly calcium lactate in cheddar (Chou et

al. 2003; Agarwal et al. 2006). Recent studies have demonstrated the utility of PXRD in

analyzing many different crystal types (e.g. calcium lactate, tyrosine, leucine, ikaite,

struvite, brushite, calcite), across a wide number of cheeses (e.g. cheddar, Parmesan,

Gouda, washed rind) (Tansman et al. 2015a, 2017b, c).

30

USE OF SINGLE CRYSTAL X-RAY DIFFRACTOMETRY ON CHEESE

The presence of single crystals in cheese is a recent discovery. Single crystals of

struvite, ikaite, and calcite have been analyzed via single crystal X-ray diffractometry

(SCXRD) (Tansman et al. 2015b, 2017a). Tansman et al. (2017a) published the structures

of ikaite and struvite. This is the first recorded instance of SCXRD being used in cheese,

and in dairy products at-large. The diffraction artifacts present in the preliminary work of

the aforementioned study demonstrated that the extracted crystals can dehydrate when

exposed to atmospheric conditions. This lends valuable insight into the conditions in which

these crystals grow. The surface smear of washed rind cheeses allows for the growth of

metastable forms of single crystals, which may have use in other disciplines. On a related

note, laboratory-grown single crystals of α-lactose monohydrate have been analyzed via

SCXRD. Beevers and Hansen (1971) found that α-lactose monohydrate is monoclinic.

31

Basics of Polarized Light Microscopy

THEORETICAL BASIS FOR POLARIZED LIGHT MICROSCOPY

Polarized light microscopy (PLM) is a special application of bright-field

microscopy. In simple terms, a light microscope can be retrofitted to perform PLM by the

inclusion of two polarizers within the optical pathway: one between the specimen and the

light source, and the other between the specimen and the observer (Kalab et al. 1995). The

former is referred to as the “polarizer” and the latter is known as the “analyzer”. The

polarizer produces plane polarized light. When the analyzer is in the optical pathway at an

orthogonal angle (i.e. 90°) to the polarizer this is referred to as “crossed polarizers”. In this

arrangement (Figure 3), amorphous materials and the background appear dark, while

crystalline regions will be brightly lit, depending on orientation (Stoiber and Morse 1971).

Figure 3. Schematic representation of polarized light microscopy technique.

32

Building upon this arrangement, other apparatuses may be included in the PLM

process to garner more information about the specimen under examination. Crystal

properties of interest to the present discussion include: angle of extinction, birefringence

(i.e. interference colors), and interference figures. Extinction refers to phenomenon of a

crystal “going dark” or transparent when viewed at certain angles using cross-polarizers

(Dyar et al. 2008). The angle at which extinction occurs is called the angle of extinction

(AE). To facilitate AE measurement a rotating stage is used. Extinction occurs four times

upon a full 360° rotation of the stage (once per 90° rotation). AE = 90° (i.e. 0°) is known

as parallel extinction; AE = 45° is known as symmetric extinction; AE at other values is

known as inclined or oblique extinction (Carlton 2011). Note: extinction, and the other

optical properties discussed here, are properties of anisotropic crystals. The single crystals

identified on cheese thus far have all been anisotropic. Isotropic crystals (i.e. cubic crystal

system) have not been observed as of the writing of this thesis. Birefringence refers to the

differences of refractive indices of the specimen. In other words, the crystal splits the

incoming plane polarized light into two components. The analyzer recombines these

components, and along with constructive/destructive interference, results in different

colorings (Bloss 1961). Although birefringence is being used to refer to coloring in this

thesis, a more appropriate term might be “interference coloring”—which is function of

birefringence, crystal thickness, and orientation (Carlton 2011). Retardation plates may be

put in the optical pathway in order to enhance the optical path different, thereby

intensifying birefringent effects; an example of such a plate is the 1/4λ plate (Stoiber and

Morse 1971). Birefringence/retardation can be measured quantitatively using a Berek

33

compensator, Senarmont compensator, or similar devices (Bloss 1961). Interference

figures are images of light interference viewed at the objective rear focal plane (Dyar et al.

2008). They are useful tools in optical mineralogy and can be used to differentiate crystal

types. A Bertrand lens is used between the observer and microscope objective in order to

see an interference figure. The microscope ocular can also be removed to see an

interference figure (Stoiber and Morse 1971). Interference figures appear as regions of light

and dark with specific color patterns depending on the type of crystal (Table 6). There are

many other properties that can examined during PLM that have not been described due to

being outside the scope of this discussion, such as optic sign, optic axial angle, sign of

elongation, etc. (Carlton 2011)

Table 6. Select optical properties for isotropic and anisotropic (uniaxial and biaxial) crystals.

Adapted from Carlton 2011.

Optical Property1

Isotropic

crystals

(cubic)

Uniaxial crystals

(tetragonal,

hexagonal)

Biaxial crystals

(orthorhombic,

monoclinic, triclinic)

Interference colors Absent Present Present

Birefringence Absent Present Present

Extinction angle Absent Parallel/Symmetric Inclined

Interference figure2 Absent Maltese cross Hyperbola-type 1When viewed under crossed polarizers 2Interference figures may not always be centered within the field of view, and

therefore only portions of the figure will be visible to the observer.

APPLICATION OF POLARIZED LIGHT MICROSCOPY IN DAIRY PRODUCTS

Although PLM can be applied to a wide variety of food products (e.g. starch

granules, fat crystal “bloom” in chocolate, plant cell walls, and muscle fibers; Kalab et al.

1995), this section will focus on PLM used for crystal-containing dairy products. Lactose

is perhaps the most studied crystal in dairy systems; having been studied via PLM for close

to a century (Wherry 1928), and continuing to the current era (Maher et al. 2015). Degree

34

of rotation of polarized light has long been used to estimate lactose in milk products (Goff

and Hartel 2013). However, the inclusion of sugar (i.e. sucrose) in products such ice cream

makes lactose estimation via degree of rotation of polarized light impossible. Although

crystals can be extracted and studied in isolation, PLM observations of crystals often occurs

in situ. Lactose crystals have been studied at length in products such as ice cream,

milk/whey powders, and other dairy products (Hartel 2001). Keller et al. (1943) developed

a set of methods to differentiate lactose crystals, ice crystals, and amorphous materials in

ice cream samples. Thin sections of ice cream were examined via PLM, and the differences

in birefringence made visual identification of lactose vs. ice a simple process. PLM can

also be used to measure crystals in quantitative contexts as well. Since crystals are easily

identifiable in PLM micrographs, the resulting images are suitable for image analysis

(Donhowe et al. 1991; Hartel 2001). The size and amount of lactose crystals were

determined in Dulce de Leche by Hough et al. (1990); these data were than correlated to

the sensory perception of grittiness. Similar work has been conducted in products frozen

dairy products (Donhowe et al. 1991). Crystals from cheese (e.g. cheddar, brie) have been

examined using PLM, but concrete identification using chemical or XRD techniques are

required (Brooker et al. 1975a; Brooker 1987; Morris et al. 1988). Recently, crystals from

the surface of washed rind cheeses have been observed via PLM, but quantification of

crystal size was not conducted (Tansman et al. 2017a).

35

References

Agarwal S, Powers JR, Swanson BG, et al (2006) Cheese pH, Protein Concentration, and

Formation of Calcium Lactate Crystals. J Dairy Sci 89:4144–4155. doi:

10.3168/JDS.S0022-0302(06)72459-6

Agarwal S, Powers JR, Swanson BG, et al (2008) Influence of Salt-to-Moisture Ratio on

Starter Culture and Calcium Lactate Crystal Formation. J Dairy Sci 91:2967–2980.

doi: 10.3168/JDS.2008-1101

Amano K (1950) A Note on Sodium Phosphate Crystals Found in Mild-cured Salmon.

Nippon SUISAN GAKKAISHI 15:638–641. doi: 10.2331/suisan.15.638

Amrane A, Prigent Y (2008) DIFFUSION OF CALCIUM AND INORGANIC

PHOSPHATE AT THE SURFACE OF A SOLID MODEL MEDIUM IN

RELATION WITH GROWTH OF GEOTRICHUM CANDIDUM AND

PENICILLIUM CAMEMBERTII. J Food Biochem 32:813–825. doi: 10.1111/j.1745-

4514.2008.00200.x

Anbu P, Kang C-H, Shin Y-J, So J-S (2016) Formations of calcium carbonate minerals by

bacteria and its multiple applications. Springerplus 5:250. doi: 10.1186/s40064-016-

1869-2

Arifuzzaman S., Rohani S (2004) Experimental study of brushite precipitation. J Cryst

Growth 267:624–634. doi: 10.1016/J.JCRYSGRO.2004.04.024

Babcock SM, Russell HL (1901) Curing of Cheddar Cheese, with Especial Reference to

Cold-Curing. In: Wisconsin Agricultural Experiment Station Bulletin, 94th edn.

University of Wisconsin, Madison, pp 4–44

Babcock SM, Russell HL, Vivian A, Baer US (1902) Conditions Affecting the

Development of White Specks in Cold-Cured Cheese. In: Wisconsin Agricultural

Experiment Station Bulletin. University of Wisconsin, Madison, pp 180–183

Beevers CA, Hansen HN (1971) The structure of α-lactose monohydrate. Acta Crystallogr

B27:1323–1325. doi: 10.1107/S0567740871003947

Ben Omar N, Gonzalez-Muñoz MT, Peñalver JMA (1998) Struvite crystallization on

Myxococcus cells. Chemosphere 36:475–481. doi: 10.1016/S0045-6535(97)10014-5

Beveridge TJ, Murray RG (1976) Uptake and retention of metals by cell walls of Bacillus

subtilis. J Bacteriol 127:1502–18.

Beveridge TJ, Murray RG (1980) Sites of metal deposition in the cell wall of Bacillus

subtilis. J Bacteriol 141:876–87.

36

Bischoff JL, Fitzpatrick JA, Rosenbauer RJ (1993) The Solubility and Stabilization of

Ikaite (CaCO3•6H2O) from 0° to 25°C: Environmental and Paleoclimatic

Implications for Thinolite Tufa. J Geol 101:21–33. doi: 10.1086/648194

Blaya J, Barzideh Z, LaPointe G (2017) Interaction of starter cultures and nonstarter lactic

acid bacteria in the cheese environment. J Dairy Sci. doi: 10.3168/JDS.2017-13345

Bloss FD (1961) An Introduction to the Methods of Optical Crystallography. Holt,

Rinehard, and Winston Inc., New York, NY

Bloss FD (Fred D (1994) Crystallography and crystal chemistry : an introduction.

Mineralogical Society of America

Bockelmann W (2011) Cheese | Smear-Ripened Cheeses. In: Encyclopedia of Dairy

Sciences. pp 753–766

Bokulich NA, Mills DA (2013) Facility-specific &quot;house&quot; microbiome drives

microbial landscapes of artisan cheesemaking plants. Appl Environ Microbiol

79:5214–23. doi: 10.1128/AEM.00934-13

Bonaïti C, Leclercq-Perlat M-N, Latrille E, Corrieu G (2004) Deacidification by

Debaryomyces hansenii of Smear Soft Cheeses Ripened Under Controlled

Conditions: Relative Humidity and Temperature Influences. J Dairy Sci 87:3976–

3988. doi: 10.3168/jds.S0022-0302(04)73538-9

Boutrou R, Gaucheron F, Piot M, et al (1999) Changes in the composition of juice

expressed from Camembert cheese during ripening. Lait 79:503–513. doi:

10.1051/lait:1999541

Bragg WH, Bragg WL (1913) The Reflection of X-rays by Crystals. Proc R Soc A Math

Phys Eng Sci 88:428–438. doi: 10.1098/rspa.1913.0040

Brennan NM, Cogan TM, Loessner M, Scherer S (2004) Bacterial surface-ripened cheeses.

In: Cheese: Chemistry, Physics and Microbiology. pp 199–225

Brennan NM, Ward AC, Beresford TP, et al (2002) Biodiversity of the bacterial flora on

the surface of a smear cheese. Appl Environ Microbiol 68:820–30. doi:

10.1128/AEM.68.2.820-830.2002

Brooker BE (1990) Food Structure Identification and Characterization of Cocoa Solids and

Milk Proteins in Chocolate Using X-ray Microanalysis. Food Struct 9:9–21.

Brooker BE (1987) The Crystallization of Calcium Phosphate at the Surface of Mould-

Ripened Cheeses. Food Microstruct 6:115–119.

Brooker BE, Hobbs DG, Turvey A (1975a) Observations on the microscopic crystalline

37

inclusions in Cheddar cheese. J Dairy Res 42:341. doi: 10.1017/S0022029900015363

Brooker BE, Hobbs DG, Turvey A, LAXA O (1975b) Observations on the microscopic

crystalline inclusions in Cheddar cheese. J Dairy Res 42:341. doi:

10.1017/S0022029900015363

Buchardt B, Israelson C, Seaman P, Stockmann G (2001) Ikaite Tufa Towers in Ikka Fjord,

Southwest Greenland: Their Formation by Mixing of Seawater and Alkaline Spring

Water. J Sediment Res 71:176–189. doi: 10.1306/042800710176

Cardello A V., Segars RA (1989) EFFECTS OF SAMPLE SIZE AND PRIOR

MASTICATION ON TEXTURE JUDGMENTS. J Sens Stud 4:1–18. doi:

10.1111/j.1745-459X.1989.tb00454.x

Carlton RA (2011) Polarized Light Microscopy. In: Pharmaceutical Microscopy. Springer,

New York, NY, pp 7–64

Chou Y-E, Edwards CG, Luedecke LO, et al (2003) Nonstarter Lactic Acid Bacteria and

Aging Temperature Affect Calcium Lactate Crystallization in Cheddar Cheese. J

Dairy Sci 86:2516–2524. doi: 10.3168/JDS.S0022-0302(03)73846-6

Clark GL (1929) X-Ray Applications in Every-Day Life. Sci. Mon. 28:172–178.

Clarkson JR, Price TJ, Adams CJ (1992) Role of metastable phases in the spontaneous

precipitation of calcium carbonate. J Chem Soc Faraday Trans 88:243. doi:

10.1039/ft9928800243

Conochie J, Czulak J, Lawrence AJ, Cole WF (1960) Tyrosine and calcium lactate crystals

on rindless cheese. Aust J Dairy Technol 15:120.

Conochie J, Sutherland BJ (1965) The nature and cause of seaminess in Cheddar cheese. J

Dairy Res 32:35. doi: 10.1017/S0022029900018355

Cooke DR, McSweeney PLH (2017) From Micelle to Melt: The Influence of Calcium on

Physico-chemical Properties of Cheese. In: Global Cheesemaking Technology. John

Wiley & Sons, Ltd, Chichester, UK, pp 20–44

Cosmidis J, Benzerara K, Guyot F, et al (2015) Calcium-Phosphate Biomineralization

Induced by Alkaline Phosphatase Activity in Escherichia coli: Localization, Kinetics,

and Potential Signatures in the Fossil Record. Front Earth Sci 3:84. doi:

10.3389/feart.2015.00084

Crichton RR (2012) Biomineralisation. In: Biological Inorganic Chemistry. Elsevier, pp

359–378

Cusack M, Freer A (2008) Biomineralization: Elemental and Organic Influence in

38

Carbonate Systems. Chem Rev 108:4433–4454. doi: 10.1021/cr078270o

Damodaran S, Wang S (2017) Ice crystal growth inhibition by peptides from fish gelatin

hydrolysate. Food Hydrocoll 70:46–56. doi: 10.1016/J.FOODHYD.2017.03.029

Dan H, Hayakawa F, Kohyama K (2008) Modulation of biting procedures induced by the

sensory evaluation of cheese hardness with different definitions. Appetite 50:158–

166. doi: 10.1016/J.APPET.2007.06.009

Diao Y, Myerson AS, Hatton TA, Trout BL (2011) Surface Design for Controlled

Crystallization: The Role of Surface Chemistry and Nanoscale Pores in

Heterogeneous Nucleation. Langmuir 27:5324–5334. doi: 10.1021/la104351k

Do Ngoc M, LeNoir J, Choisy C (1971) Les acides amines libres des fromages affines de

Camembert, Saint-Paulin et Gruyere de Comte. Rev Lait Fr 288:447–462.

Donhowe DP, Hartel RW, Bradley Jr. RL (1991) Determination of Ice Crystal Size

Distributions in Frozen Desserts. J Dairy Sci 74:3334–3344. doi: 10.3168/JDS.S0022-

0302(91)78521-4

Dorn FL, Dahlberg AC (1942) Identification of the White Particles Found on Ripened

Cheddar Cheese. J Dairy Sci 25:31–36. doi: 10.3168/jds.S0022-0302(42)95264-0

Dox AW (1911) THE OCCURRENCE OF TYROSINE CRYSTALS IN ROQUEFORT

CHEESE. J Am Chem Soc 33:423–425. doi: 10.1021/ja02216a024

Dumont JP, Roger S, Adda J (1974) Etude des composés volatils neutres présents dans les

fromages à pâte molle et à croûte lavée. Lait 54:31–43. doi: 10.1051/lait:1974531-

5323

Dupraz C, Reid RP, Braissant O, et al (2009) Processes of carbonate precipitation in

modern microbial mats. Earth-Science Rev 96:141–162. doi:

10.1016/J.EARSCIREV.2008.10.005

Dyar MD (Melinda D, Gunter ME, Tasa D, Mineralogical Society of America. (2008)

Mineralogy and optical mineralogy. Mineralogical Society of America

Engelen L, de Wijk RA, van der Bilt A, et al (2005) Relating particles and texture

perception. Physiol Behav 86:111–117. doi: 10.1016/j.physbeh.2005.06.022

Erdemir D, Lee AY, Myerson AS (2009) Nucleation of Crystals from Solution: Classical

and Two-Step Models. Acc Chem Res 42:621–629. doi: 10.1021/ar800217x

Ferris FG, Beveridge TJ (1986) Physicochemical roles of soluble metal cations in the outer

membrane of Escherichia coli K-12. Can J Microbiol 32:594–601. doi: 10.1139/m86-

110

39

Feurer C, Vallaeys T, Corrieu G, Irlinger F (2004) Does Smearing Inoculum Reflect the

Bacterial Composition of the Smear at the End of the Ripening of a French Soft, Red-

Smear Cheese? J Dairy Sci 87:3189–3197. doi: 10.3168/JDS.S0022-0302(04)73454-

2

Fidaleo M, Miele Nicoletta A, Mainardi S, et al (2017) Effect of refining degree on particle

size, sensory and rheological characteristics of anhydrous paste for ice creams

produced in industrial stirred ball mill. LWT - Food Sci Technol. doi:

10.1016/j.lwt.2017.01.046

Gaucheron F (2005) The minerals of milk. Reprod Nutr Dev 45:473–483. doi:

10.1051/rnd:2005030

Giannimaras EK, Koutsoukos PG (1987) The crystallization of calcite in the presence of

orthophosphate. J Colloid Interface Sci 116:423–430. doi: 10.1016/0021-

9797(87)90138-X

Goff HD, Hartel RW (2013) Ice cream, 7th edn. Springer

Gori K, Mortensen HD, Arneborg N, Jespersen L (2007) Ammonia Production and Its

Possible Role as a Mediator of Communication for Debaryomyces hansenii and Other

Cheese-Relevant Yeast Species. J Dairy Sci 90:5032–5041. doi: 10.3168/jds.2006-

750

Gorski CA, Fantle MS (2017) Stable mineral recrystallization in low temperature aqueous

systems: A critical review. Geochim Cosmochim Acta 198:439–465. doi:

10.1016/J.GCA.2016.11.013

Griffiths MW, Tellez AM (2013) Lactobacillus helveticus: the proteolytic system. Front

Microbiol 4:30. doi: 10.3389/fmicb.2013.00030

Guinard J-X, Mazzucchelli R (1996) The sensory perception of texture and mouthfeel.

Trends Food Sci Technol 7:213–219. doi: 10.1016/0924-2244(96)10025-X

Guo MR, Kindstedt PS (1995) Age-Related Changes in the Water Phase of Mozzarella

Cheese. J Dairy Sci 78:2099–2107. doi: 10.3168/JDS.S0022-0302(95)76836-9

Harper WJ, Swanson AM, Sommer HH (1953) Observations on the Chemical Composition

of White Particles in Several Lots of Cheddar Cheese. J Dairy Sci 36:368–372. doi:

10.3168/jds.S0022-0302(53)91505-7

Hartel RW (2001) Crystallization in Foods. Aspen Publishers, Inc., Gaithersburg, MD

Hartel RW (2013) Advances in Food Crystallization. Annu Rev Food Sci Technol 4:277–

92. doi: 10.1146/annurev-food-030212-182530

40

Hassan A, Johnson ME, Lucey JA (2004) Changes in the Proportions of Soluble and

Insoluble Calcium During the Ripening of Cheddar Cheese. J Dairy Sci 87:854–862.

doi: 10.3168/JDS.S0022-0302(04)73229-4

Hitchcock DI (1924) THE SOLUBILITY OF TYROSINE IN ACID AND IN ALKALI. J

Gen Physiol 6:747–57.

Hohenegger M, Guéguen M, Beduhn R, et al (2014) Biodiversity of the Surface Microbial

Consortia from Limburger, Reblochon, Livarot, Tilsit, and Gubbeen Cheeses.

Microbiol Spectr 2:219–250. doi: 10.1128/microbiolspec.CM-0010-2012

Hough G, Martinez E, Contarini A (1990) Sensory and Objective Measurement of

Sandiness in Dulce de Leche, a Typical Argentine Dairy Product. J Dairy Sci 73:604–

611. doi: 10.3168/jds.S0022-0302(90)78709-7

Hu Y-B, Wolf-Gladrow DA, Dieckmann GS, et al (2014) A laboratory study of ikaite

(CaCO3·6H2O) precipitation as a function of pH, salinity, temperature and phosphate

concentration. Mar Chem 162:10–18. doi: 10.1016/J.MARCHEM.2014.02.003

Hudson Institute of Mineralogy (2018) Mineral information and data.

https://www.mindat.org/. Accessed 17 Feb 2018

Hunziker OF (1920) The Butter Industry: Prepared for the use of creameries, dairy students

and pure food departments, by Otto F. Hunziker ... Self Published, La Grange, Ill.

Irlinger F, Layec S, Hélinck S, Dugat-Bony E (2015) Cheese rind microbial communities:

diversity, composition and origin. FEMS Microbiol Lett 362:1–11. doi:

10.1093/femsle/fnu015

Jacquet J, Saingt O (1952) Étude sur des fabrications accidentelles de Camemberts Rouges.

Lait 32:129–145. doi: 10.1051/lait:1952313-3145

Jelen P (1992) Whey Cheeses and Beverages. In: Whey and Lactose Processing. Springer

Netherlands, Dordrecht, pp 157–193

Johnson ME (2014) Mesophilic and Thermophilic Cultures Used in Traditional

Cheesemaking. In: Cheese and Microbes. American Society of Microbiology, pp 73–

94

Jokanovic V, Colovic B, Zivkovic-Sandic M, et al (2014) Calcium phosphate materials in

bone tissue engineering. Stomatol Glas Srb 61:93–101. doi: 10.2298/SGS1402093J

Kalab M, Allan-Wojtas P, Miller SS (1995) Microscopy and other imaging techniques in

food structure analysis. Trends Food Sci Technol 6:177–186. doi: 10.1016/S0924-

2244(00)89052-4

41

Karahadian C, Lindsay RC, Ades GL, et al (1987) Integrated Roles of Lactate, Ammonia,

and Calcium in Texture Development of Mold Surface-Ripened Cheese. J Dairy Sci

70:909–918. doi: 10.3168/jds.S0022-0302(87)80094-2

Keller WD, Reid WHE, Arbuckle WS, Decker CW (1943) Microscopy of Ice Cream with

Polarized Light. J Dairy Sci 26:683–688. doi: 10.3168/JDS.S0022-0302(43)92769-9

Kim D, Min KJ, Lee K, et al (2016) Effects of pH, molar ratios and pre-treatment on

phosphorus recovery through struvite crystallization from effluent of anaerobically

digested swine wastewater. Environ Eng Res 22:12–18. doi: 10.4491/eer.2016.037

Kindstedt PS, Kosikowski FV (1988) Calcium, Phosphorus, and Sodium Concentrations

in Cheddar Cheese. J Dairy Sci 71:285–289. doi: 10.3168/JDS.S0022-

0302(88)79556-9

Konhauser KO, Lalonde S V., Phoenix VR (2008) Bacterial biomineralization: Where to

from here? Geobiology 6:298–302. doi: 10.1111/j.1472-4669.2008.00151.x

Lawless HT, Heymann H (2010) Texture Evaluation. In: Sensory Evaluation of Food:

Principles and Practices, 2nd edn. Springer, New York, pp 259–281

Laxa O (1926) L’examen microscopique des fromages. Lait 6:885–903. doi:

10.1051/lait:19266033

Laxa O (1927) Sur la nature des grains microscopiques dans les fromages. Lait 7:521–525.

doi: 10.1051/lait:19276626

Le Graet Y, Brule G (1988) Migration des macro et oligo-éléments dans un fromage à pâte

molle de type Camembert. Lait 68:219–234. doi: 10.1051/lait:1988215

Le Graet Y, Lepienne A, Brule G, Ducruet P (1983) Migration du calcium et des

phosphates inorganiques dans les fromages à pâte molle de type Camembert au cours

de l’affinage. Lait 63:317–332. doi: 10.1051/lait:1983629-63019

Lopez FL, Bowles A, Gul MO, et al (2016) Effect of formulation variables on oral grittiness

and preferences of multiparticulate formulations in adult volunteers. Eur J Pharm Sci

92:156–162. doi: 10.1016/J.EJPS.2016.07.006

Lucey JA, Fox PF (1993) Importance of Calcium and Phosphate in Cheese Manufacture:

A Review. J Dairy Sci 76:1714–1724. doi: 10.3168/jds.S0022-0302(93)77504-9

Lucey JA, Horne DS (2009) Milk Salts: Technological Significance. In: Advanced Dairy

Chemistry. Springer New York, New York, NY, pp 351–389

Luft JR, Wolfley JR, Snell EH (2011) What’s in a drop? Correlating observations and

outcomes to guide macromolecular crystallization experiments. Cryst Growth Des

42

11:651–663. doi: 10.1021/cg1013945

Luo J, Chen X, Crump J, et al (2018) Interactions of fungi with concrete: Significant

importance for bio-based self-healing concrete. Constr Build Mater 164:275–285. doi:

10.1016/J.CONBUILDMAT.2017.12.233

Maher PG, Auty MAE, Roos YH, et al (2015) Microstructure and lactose crystallization

properties in spray dried nanoemulsions. Food Struct 3:1–11. doi:

10.1016/J.FOOSTR.2014.10.001

Marland G (1975) The stability of CaCO3·6H2O (ikaite). Geochim Cosmochim Acta

39:83–91. doi: 10.1016/0016-7037(75)90186-6

Martini S, Suzuki AH, Hartel RW (2008) Effect of High Intensity Ultrasound on

Crystallization Behavior of Anhydrous Milk Fat. J Am Oil Chem Soc 85:621–628.

doi: 10.1007/s11746-008-1247-5

McDowall FH, McDowell AKR (1939) The white particles in mature Cheddar cheese. J

Dairy Res 10:118. doi: 10.1017/S0022029900002788

McSweeney PLH (2007) Cheese problems solved. CRC Press

McSweeney PLH (2004) Biochemistry of cheese ripening. Int J Dairy Technol 57:127–

144. doi: 10.1111/j.1471-0307.2004.00147.x

Metche M, Fanni J (1978) Rôle de la flore fongique dans l’accumulation du calcium et du

phosphore à la surface des fromages du type camembert. Lait 58:336–354. doi:

10.1051/lait:197857721

Monrad JH (1910) Cheese Defects. In: New York Prod. Rev. Am. Cream.

https://books.google.com/books?id=6epMAQAAMAAJ&pg=PA151&lpg=PA151&

dq=What%27s+in+Those+“White+Specks”+on+Cheddar?.+Wis.+Agr.+Expt.+Sta&

source=bl&ots=m7xfOm9loA&sig=Tnv3CFTuDpFzKGN48WdEW9WAa2U&hl=e

n&sa=X&ved=0ahUKEwifpZjos7DZAhXRg-AKHUJxAPs. Accessed 18 Feb 2018

Morris HA, Holt C, Brooker BE, et al (1988) Inorganic constituents of cheese: analysis of

juice from a one-month-old Cheddar cheese and the use of light and electron

microscopy to characterize the crystalline phases. J Dairy Res 55:255. doi:

10.1017/S002202990002608X

Mullan WMA (2000) Causes and control of early gas production in cheddar cheese. Int J

Dairy Technol 53:63–68. doi: 10.1111/j.1471-0307.2000.tb02660.x

Mullin JW (2001) Crystallization, 4th edn. Butterworth-Heinemann, Oxford

Naka K, Chujo Y (2001) Control of Crystal Nucleation and Growth of Calcium Carbonate

43

by Synthetic Substrates. Chem Mater 13:3245–3259. doi: 10.1021/CM011035G

Nakajima H, Kunji ER., Poolman B, Konings WN (1998) Amino acid transport in

Lactobacillus helveticus. FEMS Microbiol Lett 158:249–253. doi: 10.1111/j.1574-

6968.1998.tb12828.x

Nickerson TA (1954) Lactose Crystallization in Ice Cream. I. Control of Crystal Size by

Seeding. J Dairy Sci 37:1099–1105. doi: 10.3168/JDS.S0022-0302(54)91373-9

Phadungath C, Metzger LE (2011) Effect of sodium gluconate on the solubility of calcium

lactate. J Dairy Sci 94:4843–9. doi: 10.3168/jds.2011-4549

Picque D, Leclercq-Perlat M-N, Corrieu G (2006) Effects of Atmospheric Composition on

Respiratory Behavior, Weight Loss, and Appearance of Camembert-Type Cheeses

During Chamber Ripening. J Dairy Sci 89:3250–3259. doi: 10.3168/jds.S0022-

0302(06)72600-5

Plummer LN, Busenberg E (1982) The solubilities of calcite, aragonite and vaterite in

CO2-H2O solutions between 0 and 90°C, and an evaluation of the aqueous model for

the system CaCO3-CO2-H2O. Geochim Cosmochim Acta 46:1011–1040. doi:

10.1016/0016-7037(82)90056-4

Prywer J, Torzewska A (2010) Biomineralization of struvite crystals by Proteus mirabilis

from artificial urine and their mesoscopic structure. Cryst Res Technol 45:1283–1289.

doi: 10.1002/crat.201000344

Quigley L, O’Sullivan O, Beresford TP, et al (2012) High-throughput sequencing for

detection of subpopulations of bacteria not previously associated with artisanal

cheeses. Appl Environ Microbiol 78:5717–23. doi: 10.1128/AEM.00918-12

Ribadeau-Dumas B (1984) Maîtrise de l’affinage des fromages de type Camembert. Lait

64:448–468. doi: 10.1051/lait:1984643-64435

Rivadeneyra MA, Delgado R, Párraga J, et al (2006) Precipitation of minerals by 22 species

of moderately halophilic bacteria in artificial marine salts media: Influence of salt

concentration. Folia Microbiol (Praha) 51:445–453. doi: 10.1007/BF02931589

Rysgaard S, Glud RN, Lennert K, et al (2012) Ikaite crystals in melting sea ice –

implications for pCO 2 and pH levels in Arctic surface waters. Cryosph 6:901–908.

doi: 10.5194/tc-6-901-2012

S-A Johnsson M, Nancollas GH (1992) The Role of Brushite and Octacalcium Phosphate

in Apatite Formation. Crit Rev Oral Biol Med 3:61–82.

Sainani MR, Vyas HK, Tong PS (2004) Characterization of Particles in Cream Cheese. J

Dairy Sci 87:2854–2863. doi: 10.3168/jds.S0022-0302(04)73414-1

44

Sanchez-Roman M, Rivadeneyra MA, Vasconcelos C, et al (2007) Biomineralization of

carbonate and phosphate by moderately halophilic bacteria. FEMS Microbiol Ecol

61:273–284. doi: 10.1111/j.1574-6941.2007.00336.x

Scharpf LG, Michnick MJ (1967) In Situ Identification of Crystals on Process Cheese by

X-ray Diffraction. J Dairy Sci 50:1989–1991. doi: 10.3168/JDS.S0022-

0302(67)87759-2

Severn DJ, Johnson ME, Olson NF (1986) Determination of Lactic Acid in Cheddar

Cheese and Calcium Lactate Crystals. J Dairy Sci 69:2027–2030. doi:

10.3168/jds.S0022-0302(86)80632-4

Shock AA, Harper WJ, Swanson AM, Sommer HH (1941) What’s in those “white specks”

on Cheddar? In: Wisconsin Agricultural Experiment Station Bulletin, 474th edn.

University of Wisconsin, Madison, WI,

Skeie S, Abrahamsen RK (2017) Brown Whey Cheese. In: Cheese. Elsevier, pp 1117–1132

Sommer HH (1930) A Case of Sandiness in Processed Cheese. J Dairy Sci 13:288–290.

doi: 10.3168/jds.S0022-0302(30)93526-6

Spinnler H-E, Gripon J-C (2004) Surface mould-ripened cheeses. Cheese Chem Phys

Microbiol 2:157–174. doi: 10.1016/S1874-558X(04)80043-5

Stocks-Fischer S, Galinat JK, Bang SS (1999) Microbiological precipitation of CaCO3.

Soil Biol Biochem 31:1563–1571. doi: 10.1016/S0038-0717(99)00082-6

Stoiber RE, Morse SA (1971) Microscopic Identification of Crystals. Ronald Press

Company, New York, NY

Stokes JR, Boehm MW, Baier SK (2013) Oral processing, texture and mouthfeel: From

rheology to tribology and beyond. Curr Opin Colloid Interface Sci 18:349–359. doi:

10.1016/J.COCIS.2013.04.010

Swiatek A, Jaworski J (1959) Histochemical studies on distribution of mineral matter in

cheeses. 15th Int Dairy Congr Proc 3:1509–1518.

Tansman GF, Kindstedt PS, Hughes JM (2015a) Crystal fingerprinting: elucidating the

crystals of Cheddar, Parmigiano-Reggiano, Gouda, and soft washed-rind cheeses

using powder x-ray diffractometry. Dairy Sci Technol 95:651–664. doi:

10.1007/s13594-015-0225-6

Tansman GF, Kindstedt PS, Hughes JM (2017a) Minerals In Food: Crystal Structures of

Ikaite and Struvite From Bacterial Smears On Washed-Rind Cheese. Can Mineral

55:89–100. doi: 10.3749/canmin.1600038

45

Tansman GF, Kindstedt PS, Hughes JM (2015b) Characterization of single crystals in the

rinds of white mold and smear ripened cheeses with single crystal X-ray

diffractometry. J Dairy Res 98:668–669.

Tansman GF, Kindstedt PS, Hughes JM (2017b) Crystallization and demineralization

phenomena in stabilized white mold cheese. J Dairy Sci 100:6074–6083. doi:

10.3168/jds.2016-12259

Tansman GF, Kindstedt PS, Hughes JM (2017c) Crystallization and demineralization

phenomena in washed-rind cheese. J Dairy Sci. doi: 10.3168/jds.2017-13067

Tian F, Qu H, Zimmermann A, et al (2010) Factors affecting crystallization of hydrates. J

Pharm Pharmacol 62:1534–1546. doi: 10.1111/j.2042-7158.2010.01186.x

Torri L, Piochi M, Marchiani R, et al (2016) A sensory- and consumer-based approach to

optimize cheese enrichment with grape skin powders. J Dairy Sci 99:194–204. doi:

10.3168/jds.2015-9922

Tuckey SL, Ruehe HA, Clark GL (1934) The Application of X-rays to Research in Dairy

Technology. J Dairy Sci 17:587–594. doi: 10.3168/JDS.S0022-0302(34)93277-X

Tuckey SL, Ruehe HA, Clark GL (1938a) An X-Ray Diffraction Analysis of Casein. J

Dairy Sci 21:767–775. doi: 10.3168/JDS.S0022-0302(38)95644-9

Tuckey SL, Ruehe HA, Clark GL (1938b) An X-Ray Diffraction Analysis of Cheddar

Cheese. J Dairy Sci 21:777–789. doi: 10.3168/JDS.S0022-0302(38)95645-0

Tuckey SL, Ruehe HA, Clark GL (1938c) X-ray diffraction analysis of white specks in

Cheddar cheese. J Dairy Sci 21:161.

Tyle P (1993) Effect of size, shape and hardness of particles in suspension on oral texture

and palatability. Acta Psychol (Amst) 84:111–118. doi: 10.1016/0001-

6918(93)90077-5

Vaisey-Genser M, Vane BK, Johnson S (1989) GRAININESS, CRYSTAL SIZE, AND

FIRMNESS OF STORED CANOLA OIL MARGARINES. J Texture Stud 20:347–

361. doi: 10.1111/j.1745-4603.1989.tb00445.x

Van Slyke LL, Publow CA (1909) The Science and Practice of Cheese Making. Orange

Judd Company, New York

Vassal L, Monnet V, Le Bars D, et al (1986) Relation entre le pH, la composition chimique

et la texture des fromages de type Camembert. Lait 66:341–351. doi:

10.1051/lait:1986422

Wherry ET (1928) The crystallography and optical properties of β-lactose. J Washingt

46

Acad Sci 18:302–304. doi: 10.2307/24522543

Wolfe BE, Button JE, Santarelli M, Dutton RJ (2014) Cheese Rind Communities Provide

Tractable Systems for In Situ and In Vitro Studies of Microbial Diversity. Cell

158:422–433. doi: 10.1016/j.cell.2014.05.041

Wu Q, Bishop PL, Keener TC (2005) A Strategy for Controlling Deposition of Struvite in

Municipal Wastewater Treatment Plants. Water Environ Res 77:199–207. doi:

10.2175/106143005X41771

Zamberlin Š, Antunac N, Havranek J, Samaržija D (2012) Mineral elements in milk and

dairy products. Mljekarstvo 62:111–125.

Zheng L, Sun D-W (2006) Innovative applications of power ultrasound during food

freezing processes—a review. Trends Food Sci Technol 17:16–23. doi:

10.1016/J.TIFS.2005.08.010

Zhu Y, Ma N, Jin W, et al (2017) Genomic and Transcriptomic Insights into Calcium

Carbonate Biomineralization by Marine Actinobacterium Brevibacterium linens

BS258. Front Microbiol 8:602. doi: 10.3389/fmicb.2017.00602

Zisu B, Sciberras M, Jayasena V, et al (2014) Sonocrystallisation of lactose in concentrated

whey. Ultrason Sonochem 21:2117–2121. doi: 10.1016/J.ULTSONCH.2014.03.031

47

CHAPTER 2: CHARACTERIZATION AND PRESUMPTIVE

IDENTIFICATION OF SURFACE CRYSTALS ON SMEAR RIPENED

CHEESE BY POLARIZED LIGHT MICROSCOPY

P.J. Polowsky,* G.F. Tansman,* P.S. Kindstedt,*1 and J.M. Hughes†

*Department of Nutrition and Food Sciences and

†Department of Geology, University of Vermont, Burlington 05405

1Corresponding author: Paul Kindstedt, 354 MLS Carrigan Wing, 109 Carrigan Drive,

Burlington, VT 05405, 802-656-2935, Paul.Kindstedt@uvm.edu

ABSTRACT

Surface crystallization and radial demineralization of Ca, P, and Mg occur in smear-ripened

cheese. Furthermore, crystals of ikaite, struvite, calcite, and brushite have been identified

in cheese smears by powder X-ray diffractometry (PXRD), and ikaite and struvite exist in

smears as single crystals. Polarized light microscopy (PLM) is a simple, inexpensive, and

well-established method in geology to detect and identify single crystals. However, use of

PLM to presumptively identify cheese crystals has not been reported previously. The

specific objectives of this research were: 1) to presumptively identify crystals in cheese

smears using selected PLM criteria; 2) to compare presumptive identification by PLM

against PXRD; and 3) to develop and evaluate a novel treatment for smear material to

improve crystal analyses by both PLM and PXRD. Duplicate wheels of four cheeses

produced by different manufacturers were obtained from retail sources. Scrapings of

surface smears were prepared and analyzed by PLM and PXRD by previously described

methods. Crystals were categorized by PLM based on angle of extinction (AE),

48

birefringence behavior under crossed polarizers and quartz filters, and size and shape

(circularity) by image analysis. Crystals observed by PLM fell almost exclusively into two

readily differentiated groups based on birefringence behavior and estimated angle of

extinction: Group 1 (n=18) were highly birefringent with AE=88-92°; group 2 (n=28) had

no birefringence with AE=13-26°. Group 2 crystals were significantly larger and more

circular than group 1 crystals. Group 1 and 2 were presumptively identified as struvite and

ikaite, respectively, based on known birefringence and AE characteristics. Struvite was

presumptively identified in all four cheeses by PLM but in only three cheeses by PXRD.

Ikaite was presumptively identified in three cheeses by PLM but in only two cheeses by

PXRD. These discrepancies occurred because the smear scrapings from one cheese

contained excessive amorphous matter that caused extreme background noise, potentially

obscuring diffractogram peaks that may have been present. To minimize noise, smear

scrapings were dispersed in aqueous NaOH (pH 10) prior to analyses, which resulted in

consistent results by PXRD and PLM. The method also rendered high quality images by

PLM. Data suggest that PLM may offer a simple and inexpensive means to identify

struvite, ikaite and possibly other single crystals in cheese smears.

Key words: cheese, crystal, polarized light microscopy

49

INTRODUCTION

Crystals have a dramatic effect on the sensorial and physical properties of many

cheese varieties (Tansman et al., 2015). For example, much research has focused on surface

crystals of calcium lactate pentahydrate and their detrimental effect on the visual

appearance of Cheddar cheese (Agarwal et al., 2006; Johnson et al., 1990). Dense, hard

crystals of tyrosine that form internally in aged Parmesan cheeses have also received much

attention because of their positive impact on the characteristic granular texture that is

considered a defining attribute of the fully mature cheese (Zannoni et al., 1994; Noël et al.,

1996). In contrast, dense, hard internal crystals of calcium lactate pentahydrate that

sometimes occur in Cheddar cheese may be viewed as defects because they are unfamiliar

to many consumers and may be mistaken for foreign objects (Tansman et al., 2015).

In addition to affecting cheese texture and visual appearance directly, crystals may

also indirectly trigger dramatic changes in sensorial and physical properties, as in the case

of soft surface ripened white mold (bloomy rind) cheese varieties such as Camembert and

Brie. pH-induced crystallization of calcium phosphate at the surface of bloomy rind

cheeses sets up a pronounced outward migration of endogenous cheese minerals such as

calcium, magnesium, and phosphorus during ripening (Le Graet & Brule, 1988; Le Graet

et al., 1983; Tansman et al., 2017a). This process leads to demineralization of the curd body

and results in pH-dependent radial softening, significantly affecting the overall texture and

body of the resulting cheese (Metch & Fanni, 1978).

A similar process of radial softening occurs in soft surface ripened washed rind

50

(smear ripened) cheeses. However, in contrast to bloomy rind cheeses, the surface crystals

that form on soft washed rind cheese consist mostly of ikaite (calcium carbonate

hexahydrate), struvite (magnesium ammonium phosphate hexahydrate) and small amounts

of calcite (calcium carbonate), with only minor and evidently transient formation of

brushite (calcium phosphate dihydrate) (Tansman et al., 2015, 2017b,c). Anecdotal reports

from the industry and informal observations by the authors suggest that these surface

crystals can grow to sizes that exceed the sensory perception threshold and cause a gritty

or sandy texture in the rinds of soft washed rind cheeses.

Advances in analytical approaches to identify and characterize crystals in cheese

have helped to fuel a growing interest in cheese crystal research, including several recent

studies that employed new applications powder X-ray diffractometry (PXRD) (Tansman

et al., 2014, 2015; Tansman et al., 2017a,b). A limitation of this technique, however, has

been the presence of high background noise due to the difficulty of removing non-

crystalline cheese material from the harvested crystals, which can make interpretation of

X-ray diffraction patterns difficult and result in unidentified crystal phases (Bloss, 2012).

This difficulty appears to be especially problematical when analyzing surface scrapings

from soft washed rind and bloomy rind cheeses (Tansman et al., 2017a,b).

Polarized light microscopy (PLM), which is a well-known technique in optical

mineralogy and crystallography to study and identify crystals, is another promising

approach that has been recently applied to study cheese crystals (Tansman et al., 2017b,c).

PLM offers a simple and inexpensive, yet powerful analytical approach when crystals

occur as single crystals because the anisotropic nature of single crystals lends them unique

51

identifying features such as angle of extinction and birefringence when viewed using this

methodology (Nesse, 2004). Angle of extinction refers to the specific position that some

crystals go “extinct”, or dark/transparent, when viewed under cross polarized light (Nesse,

2004). Birefringence refers to the coloring some crystals take on when view under cross

polarized light. The type and intensity of birefringence can help indicate crystal type (Dyar

et al., 2008). Single crystals also tend to exhibit characteristic shapes and geometries that

can aid in identification (Bloss, 2012). Image analysis software can be used to quantify the

shape and size of crystals, allowing for robust analysis and differentiation. These crystal

morphologies, along with intensity of birefringence and angle of extinction, can all be used

to tentatively identify specific crystal species.

Previous studies confirmed that the principal surface crystals on soft washed rind

cheeses, namely ikaite and struvite, occurred as single crystals that were readily observed

using PLM under crossed polarizers (Tansman et al., 2017 b,c). The present work now aims

to apply the above criteria to presumptively identify single crystals present on the surface

of soft washed rind cheeses. The specific objectives of this work are three-fold: (1) to

presumptively identify surface crystals on soft washed rind cheese using specific PLM

criteria, (2) to compare presumptive PLM identification against PXRD, and (3) to develop

and evaluate a novel treatment for cheese surface smear material to improve the analyses

of cheese crystals by both PLM and PXRD.

52

MATERIALS AND METHODS

Cheeses Samples

Four different soft washed rind cheese varieties were purchased at local retailers in

the Burlington, VT area. Two wheels of each cheese variety with identical code date and

vat identifier (if available) were obtained and served as duplicate experimental units. All

analyses were performed identically on each of the two wheels for each of the four

varieties. All cheese samples were stored at 3°C for up to three days until analyses were

conducted. A flat-tip pH electrode (Thomas Scientific, Swedesboro, NJ) was used for

surface pH measurements. The probe was placed on three separate rind locations (top, side,

bottom) and measurements were averaged. (6 total pH measurements per cheese variety)

Moisture analyses were conducted in triplicate using a forced draft oven set at 100°C. A

wedge of cheese from each wheel was homogenized in a mortar and pestle. Approximately

2 grams of homogenized sample were weighed into pre-dried and pre-weighed aluminum

pans and then allowed to dry for 24 hours at 100°C, until a constant weight was reached.

(6 total moisture content measurements per cheese variety)

Smear Collection

Smear scrapings were collected from a 1 cm2 section of cheese rind using a metal

spatula. Care was taken to limit the fracturing of crystals. Scrapings were applied directly

to microscope and diffraction slides for PLM and PXRD analyses, respectively, using a

metal spatula and a dissection needle.

53

Alkaline Dispersion

Alkaline dispersions (pH=10) of smears were prepared by submerging a sample of

smear (collected as described above) in 15mL of 0.0001M NaOH (Fisher Scientific,

Pittsburgh, PA) in a 50mL beaker. Beakers were vortexed gently to encourage dispersion

of smear matrix. Alkaline mixtures were allowed to sit at room temperature for

approximately three hours. Crystals that settled to the bottom of the beaker were used for

PLM and PXRD analyses.

Powder X-ray Diffractometry

A zero-background diffraction slide was used for all PXRD analyses. Cheese smear

scrapings were mounted directly on the slide and overlaid with a thin film of mineral oil as

described in previous work (Tansman et al., 2017 b,c). For alkaline dispersion samples,

300μL of crystal-containing smear dispersion liquid (pipetted from the bottom of the

beaker) were applied to the diffraction slide. A tissue (Kimwipes, Kimberly-Clark

Professional, Roswell, GA) was used to absorb excess liquid. Diffractograms were

generated on a Miniflex II powder X-ray diffractometer (Rigaku, The Woodlands, TX)

using a speed of 2° 2θ/min between 5° 2θ and 50° 2θ. Diffraction patterns were compared

to existing reference diffraction patterns archived in the International Center for Diffraction

Data (ICDD) database. Diffractograms were analyzed and compared using the PDXL

diffractometer software package (Rigaku, The Woodlands, TX). The following ICDD

reference cards were used: #01-074-7147 (ikaite), #00-015-0762 (struvite), and #01-074-

3640 (brushite). The experimental diffraction patterns were compared to the reference

pattern from each ICDD card in order to account for all peaks in the experimental

54

diffraction pattern. The absence of additional unexplained peaks indicated all crystal types

were elucidated by the reference patterns within the sensitivity of the equipment and

experimental design. Some experimental diffraction patterns exhibited preferred

orientation, but did not prevent identification of the major crystal phases present in the

experimental samples. Preferred orientation results in diffraction patterns with missing

peaks or peaks of diminished intensity (Klug & Alexander, 1954).

Polarized Light Microscopy

Samples collected by the scraping method were applied to a glass microscope slide

and overlaid with a thin film of mineral oil to prevent dehydration. Samples prepared in

alkaline dispersion were analyzed directly in a beaker to prevent atmospheric exposure of

the crystalline entities. Polarized light micrographs were captured using a Nikon E200POL

microscope (Nikon Corporation, Tokyo, Japan). Samples were viewed under crossed

polarizers. Images were recorded using a SPOT Idea 1.3 MP color camera (SPOT Imaging

Solutions, Sterling Heights, MI). The microscope was equipped with a rotating stage,

which was used to analyze angles of extinction. Observed objects were determined to be

single crystals if they displayed uniform extinction. Video was captured of each field of

view while rotating the stage a full 360°. Angle of extinction measurements were

determined using these videos and an on-screen protractor (Protractor for Mac, Trilithon

Software, New York, NY). Three fields of view (FOV) were used for each duplicate cheese

(6 total images were collected per cheese variety). FOVs were chosen at random by moving

the stage random distances in the x and y directions. Random distances were determined

55

using the “random” package (Eddelbuettel, 2017) in R (version R version 3.3.2; R

Foundation for Statistical Computing; Vienna, Austria).

Image Analysis

PLM images were imported into Adobe Photoshop (Adobe Systems Incorporated,

San Jose, CA) where the crystalline entities were outlined and a black background was

applied. Overlapping crystals, crystals with edges obscured by smear material, and crystals

intersecting the FOV border were not outlined, and not used for any further analysis. The

outlined images were analyzed using Fiji (ImageJ distribution, Schindelin et al. 2012).

Images were converted to 8-bit black and white, and thresholding was applied using the

Li's Minimum Cross Entropy thresholding option (Li & Lee, 1993). The threshold images

were then analyzed using the particle analysis plugin. Metrics of interest (area and

circularity) were collected and recorded for each isolated crystal in each processed image.

Outlined entities smaller than 50 pixels2 were not included due to limited measurement

accuracy when analyzing very small objects.

Statistical Analysis

Crystal grouping data and crystal harvest technique data were analyzed via a one-

way ANOVA. Analyses were performed using JMP Pro 13 (SAS Institute Inc., Cary, NC).

A P<0.05 level of significance was used for all analyses.

56

RESULTS AND DISCUSSION

Cheeses Samples

All cheese samples had surface pH values and moisture contents that were

consistent with the soft washed rind cheese category (Table 7). Surface pH values varied

by approximately 0.5, and moisture content varied by approximately 5% across the

different cheeses. These observed differences are perhaps a result of differences in make

procedure, ripening regimen, and age of the samples at the time of observation.

Polarized Light Microscopy

Extinction angle, birefringence, and image analysis results stratified the observed

crystals into two well-defined groupings (Table 8). Examples of polarized light

micrographs of crystals isolated from smear scrapings are shown in Figure 4. “Group 1”

crystals exhibited an average size and circularity of approximately 5600μm2 and 0.662,

respectively, and were observed in all four cheeses used in this study. These crystals

exhibited angles of extinction in the range of 89.5° to 91.0° (mean = 90.1°) and high

birefringent coloring. “Group 1” crystals can be seen in Figure 4 marked with hollow

arrows. Their birefringence is manifested as bright colors such as blue, orange, teal, yellow,

and green. A common occurrence observed during PLM analyses was a color

transformation of “Group 1” crystals when the microscope stage was rotated. They often

started as bright blue and, upon rotation past extinction (i.e. transparency), transitioned to

bright orange. The use of the quartz (quarter-λ) filter enhances weak birefringent coloring,

and a similar process has been used previously to identify urate crystals, where they showed

57

a characteristic color transformation from yellow to blue (Hardy & Nation, 1984). “Group

1” crystals had an angle of extinction of approximately 90°, which is known as parallel

extinction in the optical mineralogy field (Dyar et al., 2008). The shape of “Group 1”

crystals were often trapezoidal and angular, as confirmed by circularity measurements.

“Group 1” crystals were significantly smaller and less circular than “Group 2” crystals

(p<0.05). The angle of extinction, shape, and birefringent coloring of “Group 1” crystals

are consistent with the mineral struvite, which exhibits parallel extinction (Ellis, 1962), has

a characteristic angular or needle-like shape (Wilsenach et al., 2007), and has similar

coloring (Daudon et al., 2016).

The average size and circularity of “Group 2” crystals was approximately

22000μm2 and 0.708, respectively, and were observed in cheeses A, C, and D. They were

significantly larger and rounder/more circular than “Group 1” crystals (p<0.05). They had

angles of extinction in the range of 13.5° to 24.5°, as well as little-to-no birefringent

coloring. (shown in Figure 4 – solid arrows) The absence of birefringence presents as

crystals having a white coloring. “Group 2” crystals exhibited a mean extinction angle of

18.1°. Angle of extinction not equal to 90° is known as “inclined extinction”. The range of

extinction angles observed for “Group 2” crystals was likely caused by difficulties during

angle of extinction measurement. Round crystals, fractured crystals, and crystals obscured

by smear material make proper orientation and angle measurement more difficult. The

angle of extinction, shape, and low birefringent coloring of “Group 2” crystals are

consistent with the mineral ikaite, which exhibits inclined extinction at an angle around

58

17° (Hesse et al. 1983), has a characteristic round/oval shape, and has similar coloring

(Rysgaard et al., 2012).

Powder X-ray Diffractometry

Diffractograms of smear scrapings (Figure 5) identified crystals in cheeses A, B,

and D. Smear scrapings from cheese A displayed distinctive peaks consistent with struvite

(NH4MgPO4·6H2O) and ikaite (CaCO3·6H2O). Cheese B smear scrapings were found to

only contain struvite crystals. Cheese C smear scrapings resulted in diffractograms

displaying high amorphous background noise with no identifiable crystal peaks. Cheese D

smear scrapings displayed peaks consistent with struvite and ikaite crystals. The absence

of discernable crystal peaks in the diffractograms from cheese C smears (Figure 5C) is at

odds with the “Group 1” and “Group 2” crystals identified as struvite and ikaite,

respectively, during PLM analysis (Figure 4C; Table 10). However, it is important to note

that all of the diffractograms obtained from the two wheels of cheese C displayed extremely

high background noise, even when the analyses were repeated several times. Although

high background was present, to varying degrees, across all analyzed smear samples, the

exceptionally high background noise of cheese C made complete identification more

difficult and increased the possibility of not detecting certain crystalline phases that may

have been present in small amounts. This may possibly explain the discrepancy with cheese

C, which clearly displayed crystals that were presumptively identified as ikaite and struvite

by PLM, but which showed no evidence of crystals by PXRD.

59

Significance of Ikaite, Struvite, and Brushite

Ikaite, struvite, and brushite have all been previously identified in soft washed rind

cheese (Tansman et al., 2015; Tansman et al., 2017b,c). These minerals most likely form as

a result of physiochemical changes during cheese ripening, including radial mineral

migration and gaseous changes of the ripening atmosphere (Tansman et al., 2017 b;

McSweeney, 2004). Crystal formation no doubt plays an important role in cheese softening

and texture development by sequestering important structural components of the casein

matrix such as calcium and phosphate. In addition to changes of the cheese body, crystal

formation may also have an impact on the sensory properties of the cheese surface as well.

Anecdotal reports from cheese makers and mongers point towards surface grittiness

potentially being a common defect among smear ripened cheeses (M. Kehler, Cellars at

Jasper Hill, Greensboro, VT, personal communication). Present results suggest that certain

crystals, especially ikaite, may grow large enough to surpass the sensory perception

threshold and result in the aforementioned grittiness. Vaisey-Genser et al. (1989) estimated

the crystal size threshold for grittiness in margarine spreads to be 22μm, while Hough et

al. (1990) noted a detection threshold range of 45μm - 105μm for sandiness in Dulce de

Leche. A similar threshold of 45μm has been used for the detection threshold of ice crystals

in ice cream (Goff & Hartel, 2013). Figure 4 shows many crystals which exceed the length

of the scale bars shown (250μm). Crystals of this order of size are very likely exceeding

the sensory perception threshold, but further study linking crystal size to sensory response

is needed to ascertain the threshold level of grittiness/sandiness in these cheeses.

60

Presumptive Crystal Identification via PLM

Based on the results from PXRD analyses, and the crystal groupings identified from

PLM images, we make the supposition that “Group 1” and “Group 2” crystals can

tentatively be identified as struvite and ikaite, respectively. The overall crystal types

identified by the different analysis techniques used in the present study are summarized in

Table 10. Presumptive struvite crystals were significantly smaller and less circular than

presumptive ikaite crystals (Figure 4). Presumptive struvite also exhibited high birefringent

coloring and parallel extinction, whereas presumptive ikaite had low birefringence and

inclined extinction (Table 8; Figure 4). These properties are consistent with observations

in our previous work (Tansman et al., 2017b; Tansman et al., 2017b,c) and the optical

mineralogy literature (Ellis, 1962; Hesse et al., 1983). Crystal size and circularity may

change based on the cheese age and crystals fracturing during examination, but angle of

extinction and birefringence are intrinsic properties not dependent on crystal size or shape

(Dyar et al., 2008). Although angle of extinction and birefringence are likely sufficient to

aid in the identification among the small subset of crystals known to occur on the rinds of

smear-ripened cheeses, more advanced optical techniques can be employed. One such

technique is to examine interference figures, which appear as regions of light and dark with

specific coloring depending on the type of crystal (Bloss, 1961). A Bertrand lens is inserted

between the ocular and microscope objective in order to see an interference figure.

Interference figures were generated for presumptive ikaite and struvite crystals, which were

biaxial (-) 2V~40° and biaxial (+) 2V~45°, respectively. This is in agreement with

reference values (Hudson Institute of Mineralogy, 2018). This further confirms the

61

conclusion made using observations of birefringence, extinction angle, and crystal

morphology. These criteria can be used as part of a robust methodology for presumptively

identifying crystal types via PLM, and are standard techniques in the optical

crystallographic of minerals in subdisciplines of geology.

Although PXRD will remain the gold standard for crystal identification, it has

possible drawbacks. The high capital cost and advanced experience needed to interpret

results is often exclusionary for most cheese technology professionals. The presence of

amorphous material can also make PXRD analysis difficult and prevent the detection of

crystalline entities, as evident in Figure 5C.

Alkaline Dispersion Sample Preparation

As a possible remedy to the high amorphous background issue, alkaline dispersions

of each smear scraping were prepared and analyzed via PXRD and PLM. Diffractograms

from each alkaline smear dispersion can be seen in Figure 6. Cheese A smear dispersion

displayed characteristics of struvite, ikaite, and brushite crystals. Cheese B displayed peaks

consistent with only struvite crystals being present. Cheese C displayed peaks indicative

of struvite and ikaite crystals. Cheese D was found to contain struvite and ikaite. Across

all diffractograms, less amorphous background noise was present, as can be viewed by the

lower baselines in Figure 6. Of note is the emergence of diffractogram peaks

corresponding to ikaite and struvite in the alkaline smear dispersion of cheese C (Figure

6C), which were consistent with the presumptive identifications by PLM, but were not

present in the direct smear scrapings (Figure 4C; Table 10). This suggests that the alkaline

62

treatment improved the sensitivity of the PXRD identification by separating the smear

crystals from the amorphous smear material that caused high baseline noise in the direct

smear sample. Also of interest is the emergence of diffractogram peaks corresponding to

brushite in the alkaline dispersion of cheese A. This again could be due to the improved

PXRD sensitivity, or it could indicate induced crystallization due to the alkaline

environments to which the smears were subjected. Although the latter is possible, the

absence of brushite from the three other alkaline treatments points towards better PXRD

performance as the causative factor. Previous work (Tansman et al., 2017b) has also shown

brushite to be endogenous to cheese smears when measured directly, without alkaline

treatment.

Summary data of image analysis results from crystals isolated via alkaline smear

dispersions, and also via smear scrapings, are shown in Table 9. Crystals analyzed from

the alkaline dispersions were smaller and less circular on average than crystals analyzed

via the smear scraping method. The alkaline dispersions may have caused the crystals to

partially dissolve or may have preferentially separated smaller, less circular crystals.

However, the large differences in standard deviations between the two treatments violated

the assumption of homogeneity of variance, and therefore a statistical difference tests could

not be performed on mean crystal area and circularity. Of special note is the improved

quality of micrographs captured when the alkaline dispersion method was used (Figure 7).

Smear material was present in minute quantities and crystal entities were clearer and more

easily identifiable. Given this, the alkaline dispersion technique coupled with PLM could

be a powerful presumptive identification tool, but perhaps not suitable for quantification

63

of crystal size or shape. These micrographs also allowed for observations of multi-

generational crystal growth and possible resorption (Figure 8). Multi-generational crystal

growth can occur after an initial crystal is formed (first generation), and reaction conditions

change such that new crystal growth occurs at a future time, using the first generation as a

seed crystal. This new growth is known as a second generation crystal (Klapper, 2010),

and may be the result of an influx of minerals necessary for crystal growth to occur again.

Resorption occurs when crystals dissolve back into solution, which can be indicative of a

shift in equilibrium (Ginibre et al., 2007). A struvite crystal can be seen in Figure 8 (open

arrow) with an irregular morphology and pitted surface, suggesting it may be being

resorbed into solution. This suggests that changes in the mineral content of the cheese’s

water phase resulted in struvite being dissolved back into solution, while a second

generation of ikaite was crystallizing. Both of these phenomena once again suggest that the

smears on washed rind cheeses are complicated chemical systems that can influence

complex crystallization dynamics.

CONCLUSIONS

The present study outlines a novel approach to analyzing the size and shape of surface

situated crystals on externally bacterial ripened cheeses via PLM and image analysis, as

well as a set of presumptive criteria to establish the identity of these crystals. When PXRD

analyses are not feasible due to cost or lack of trained personnel, PLM is a powerful tool

that can lend valuable insight into the types of crystals present on the surface of smear

ripened cheeses. These findings provide exciting opportunities for researchers to further

study mineral crystallization in cheese and its impact on sensory properties.

64

ACKNOWLEDGMENTS

This study was funded by USDA Hatch Project 033286. The National Science

Foundation is acknowledged for support of grant EAR-0922961 for the purchase of the

powder X-ray diffractometer.

65

REFERENCES

Agarwal, S., Powers, J.R., Swanson, B.G., Chen, S., Clark, S. (2006) Cheese pH, protein

concentration, and formation of calcium lactate crystals. J. Dairy Sci. 89:4144-

4155.

Bloss, F.D. (2012) Crystallography and crystal chemistry. Hold, Rinehart, Winston, Inc.,

New York, NY.

Bloss, F.D. (1961) An Introduction to the Methods of Optical Crystallography. Holt,

Rinehard, and Winston Inc., New York, NY

Daudon, M., Frochot, V., Bazin, D., and Jungers, P. (2016) Crystalluria analysis improves

significantly etiologic diagnosis and therapeutic monitoring of nephrolithiasis. C R

Chim. 19:1514-1526.

Dyar, M.D., Gunter, M.E., and Tasa, D. (2008) Mineralogy and Optical Mineralogy.

Mineralogical Society of America, Chantilly, VA.

Eddelbuettel, Dirk. (2017) random: True Random Numbers using RANDOM.ORG. R

package version 0.2.6. https://CRAN.R-project.org/package=random.

Ellis, S.E. (1962) Phosphate minerals in a calculus from the reticule-tureen of the sheep.

Mineral Mag. 33:600-608.

Ginibre C., Worner G., Kronz A. (2007) Crystal zoning as an archive for magma evolution.

Elements. 3:261-266.

Goff, H.D., Hartel, R.W. (2013) Ice Cream. 7th ed. Springer, New York, NY.

Hardy, R.H. & Nation, B. (1984) Acute gout and the accident and emergency department.

Arch Emerg Med, 1:89–95.

Hesse, K.F., Küppers, H., Suess, E. (1983) Refinement of the structure of Ikaite,

CaCO3·6H2O. Z Kristallogr Cryst Mater. 163:227-231.

Hough, G., Martinez, E., and Contarini, A. (1990) Sensory and objective measurement of

sandiness in Dulce de Leche, a typical Argentine dairy product. J. Dairy Sci.

73:604-611.

Hudson Institute of Mineralogy (2018) Mineral information and data.

https://www.mindat.org/. Accessed 28 Feb 2018.

Johnson, M.E., Riesterer, B.A., Chen, C., Tricomi, B., Olson, N.F. (1990) Effect of

packaging and storage conditions on calcium lactate crystallization on the surface

of cheddar cheese. J. Dairy Sci. 73:3033-3041.

66

Karahadian, C. & Lindsay, R.C. (1987) Integrated roles of lactate, ammonia, and calcium

in texture development of mold surface-ripened cheese. J. Dairy Sci. 70:909-918.

Klapper, H. (2010) Generation and propagation of defects during crystal growth. In

Springer Handbook of Crystal Growth. Springer, Berlin, Germany.

Klug, H. P., & Alexander, L. E. (1954) X-ray diffraction procedures: for polycrystalline

and amorphous materials. Vol. 2. Wiley, New York, NY.

Li, CH & Lee, CK. (1993) Minimum Cross Entropy Thresholding. Pattern Recognit. 26(4):

617-625.

Le Graet, Y., A. Lepienne, G. Brule, and P. Ducruet. (1983) Migration du calcium et des

phosphates inorganiques dans les fromages à pâte molle de type Camembert au

cours de l'affinage. Lait. 63:317-332.

Le Graet, Y. & G. Brule. (1988) Migration des macro et oligo-éléments dans un fromage à

pâte molle de type Camembert. Lait 68:219-234.

McSweeney, P. (2004) Biochemistry of cheese ripening. Int J Dairy Technol. 57:127-144.

Metche, M. & Fanni, J. (1978) Rôle de la flore fongique dans l'accumulation du calcium et

du phosphore à la surface des fromages du type camembert. Lait. 58:336-354.

Nesse, W.D. (2004). Introduction to Optical Minerology. 3rd ed. Oxford University Press,

New York, NY.

Noël, Y., Zannoni, M., Hunter, E.A. (1996) Texture of Parmigiano Reggiano cheese:

statistical relationships between rheological and sensory variates. Lait 76:243-254.

Rysgaard, S., Glud, R. N., Lennert, K., Cooper, M., Halden, N., Leakey, R.J.G.,

Hawthorne, F. C., and Barber, D. (2012) Ikaite crystals in melting sea ice –

implications for pCO2 and pH levels in Arctic surface waters. Cryosphere. 6:901–

908.

Schindelin, J., Arganda-Carreras, I. & Frise, E. et al. (2012) Fiji: an open-source platform

for biological-image analysis. Nat. Methods. 9:676-682.

Tansman, G., Kindstedt, P.S., and Hughes, J.M. (2015) Crystal fingerprinting: elucidating

the crystals of Cheddar, Parmigiano-Reggiano, Gouda, and soft washed-rind

cheeses using powder x-ray diffractometry. Dairy Sci Technol. 95:651-664.

Tansman, G., Kindstedt, P.S., and Hughes, J.M. (2017a) Crystallization and

demineralization phenomena in stabilized white mold cheese. J. Dairy Sci. 100:

6074–6083.

67

Tansman, G., Kindstedt, P.S., and Hughes, J.M. (2017b) Crystallization and

demineralization phenomena in washed-rind cheese. J. Dairy Sci. 11:8694-8704.

Tansman, G., Kindstedt, P.S., and Hughes, J.M. (2017c) Minerals in food: Crystal

structures of ikaite and struvite from bacterial smears on washed-rind cheese. Can

Mineral. 55:89-100.

Vaisey-Genser, M., Vane, B.K., and Johnson, S. (1989) Graininess, crystal size, and

firmness of stored canola oil margarines. J Texture Stud. 20:347-361.

Wilsenach, J.A., Schuurbiers, C.A.H, and van Loosdrecht, M.C.M. (2007) Phosphate and

potassium recovery from source separated urine through struvite precipitation.

Water Res. 41:458-466.

Zannoni, M., Bertozzi, L., and Hunter, E.A. (1994) Comparison of Parmigiano-Reggiano

and American Parmesan cheeses by sensory analysis of texture. Scienza e Tecnica

Lattiero Casearia. 45:505-518.

68

Table 7. Cheese sample compositional information. Results are from triplicate measurements

from duplicate wheels of cheese.

Sample Manufacture Location Surface pH Moisture Content (%)

Mean SD Mean SD

Cheese A Vermont, USA 6.50 0.04 47.3 0.92

Cheese B New York, USA 6.71 0.03 45.2 0.57

Cheese C Vermont, USA 6.84 0.04 47.9 0.92

Cheese D Vermont, USA 7.06 0.04 51.9 0.85

69

Table 8. Crystal grouping results from image analysis of polarized light micrographs. N refers to the

number of crystals analyzed in each grouping.

Crystal

Groupings N

Area (μm2) Circularity Extinction

Angle (°) Bire-

fringence

Presumptive

ID Mean SD Mean SD Mean Range

Group 1 18 5637.46b 2128.84 0.662b 0.057 90.1 98.5-91.0

(parallel) High Struvite

Group 2 28 22148.48a 13021.09 0.708a 0.077 18.1 13.5-24.5

(inclined) Low Ikaite

a,bDiffering letters within a column indicate a significant difference (P<0.05).

70

Table 9. Comparison of crystal size and shape between the two harvesting techniques.

Extraction Method Area (μm2) Circularity

Mean SD Mean SD

Alkaline dispersion 17738.57 14690.18 0.632 0.110

Smear scraping 68923.06 125800.24 0.700 0.026

71

Table 10. Summary of crystal identification results from powder X-ray diffractometry

(PXRD) and tentative identification from polarized light microscopy (PLM).

Sample

PXRD

PLM smear scraping

alkaline

dispersion

Cheese A struvite

ikaite

struvite

ikaite

brushite

struvite

ikaite

Cheese B struvite struvite struvite

Cheese C None identified struvite

ikaite

struvite

ikaite

Cheese D struvite

ikaite

struvite

ikaite

struvite

ikaite

72

Figure 4. Polarized light micrographs captured via the smear scraping method from cheeses A, B, C,

and D. Examples of presumptive ikaite crystals labeled with solid arrows. Examples of presumptive

struvite crystals labeled with open arrows. Scale bars represent

73

Figure 5. PXRD patterns of crystals isolated from cheeses A, B, C, and D via the smear scraping

method. Diffractogram peaks are labeled as being characteristic of ikaite (square) or struvite

(triangle). Y-axis is using arbitrary units (AU) of counts per second.

74

Figure 6. PXRD patterns of crystals isolated from cheeses A, B, C, and D via the alkaline dispersion

method. Diffractogram peaks are labeled as being characteristic of ikaite (square), struvite (triangle),

or brushite (circle). Y-axis is using arbitrary units (A

75

Figure 7. Polarized light micrographs comparing image quality when the smear scraping technique is

employed (A), and when the alkaline dispersion method is used (B). Scale bars represent 250μm.

Viewed under quartz filter.

76

Figure 8. Polarized light micrograph captured using the alkaline dispersion method. Example of

presumptive ikaite showing multi-generational growth. First generation (solid arrow) can be seen

encompassed by second generation (striped arrow). Presumptive struvite showing possibility of

resorption/dissolving into aqueous phase (open arrow). Scale bar represents 75μm.

77

CHAPTER 3: SIZE, SHAPE, AND IDENTITY OF SURFACE CRYSTALS

AND THEIR RELATIONSHIP TO SENSORY PERCEPTION OF

GRITTINESS IN AMERICAN ARTISANAL AND EUROPEAN PDO SOFT

SMEAR RIPENED CHEESES

P.J. Polowsky,* P.S. Kindstedt,*1 and J.M. Hughes†

*Department of Nutrition and Food Sciences and

†Department of Geology, University of Vermont, Burlington 05405

1Corresponding author: Paul Kindstedt, 354 MLS Carrigan Wing, 109 Carrigan Drive,

Burlington, VT 05405, 802-656-2935, Paul.Kindstedt@uvm.edu

ABSTRACT

Soft smear-ripened cheeses undergo extensive surface crystallization and radial

demineralization of calcium, magnesium and phosphorus, which likely contribute to radial

softening during ripening. Furthermore, anecdotal evidence suggests that grittiness is a

common characteristic of smear-ripened cheeses. The primary aims of the present study

were to evaluate the intensity of perceived grittiness in U.S. artisanal and European PDO

smear ripened cheeses, and to relate perceived grittiness to the size, shape and identity of

crystals present in the cheese surface smears. Fully ripened wheels of 24 different varieties

of smear-ripened cheeses, 16 produced in the U.S. and 8 in the E.U., were obtained from

retail sources. A trained sensory panel (n=12) was employed to evaluate intensity of

grittiness. Crystals present in the cheese smears were identified by powder X-ray

diffractometry and polarized light microscopy (PLM), and further evaluated in PLM

micrographs by image analysis for size and shape characteristics. Mean sensory scores for

78

the 24 cheeses ranged from no perceived grittiness to pronounced grittiness. Surface

crystals included ikaite, struvite, calcite and brushite, and mean crystal length and area

ranged among cheeses from 27μm to 1096μm, and 533μm2 to 213,969μm2, respectively.

Panel threshold for grittiness occurred at a mean crystal length of about 66μm and mean

crystal area of about 2913μm2. Cheeses with mean values at or below these thresholds

displayed negligible perceived grittiness. In contrast, for cheeses with mean values above

these thresholds, the mean sensory scores for grittiness were highly correlated with mean

crystal length and crystal area (r=0.93 and 0.96, respectively; P<0.001). Results suggest

that surface crystals in soft smear ripened cheeses influence sensory perception of texture

in complex ways that likely include radial softening and grittiness development. A better

understanding of factors that govern surface crystal formation may lead to improved

control over crystallization and more consistent cheese texture.

INTRODUCTION

Soft smear ripened cheeses, also known as soft washed rind cheeses, can encompass

a wide variety of products. Several underlying commonalities, however, are the surface

application of a brine solution during aging, the resulting outgrowth and proliferation of a

complex surface microbiota, and the radial softening/liquefaction of the cheese body over

time (Brennan et al. 2004). These actions bring about the characteristic reddish-orange

appearance, soft texture, and distinctive flavor that many associate with soft smear ripened

cheeses.

79

The surface ecologies of smear ripened cheeses are complex and can contain dozens

of genera of fungi and bacteria, such as Penicillium, Debaryomyces, Brevibacterium,

Corynebacterium, and many others (Wolfe et al. 2014; Dugat-Bony et al. 2016). These

microbes, and their associated metabolisms, can have dramatic effects on cheese chemistry

and ripening. The metabolization of lactate and subsequent pH rise can induce precipitation

and crystallization of minerals at the cheese surface, causing radial migration of minerals

outward (Tansman et al. 2017b). A homologous process has been observed in a closely

related cheese variety: white mold cheeses such as brie and camembert (Le Graet et al.

1983; Le Graet and Brule 1988; Tansman et al. 2017c). Another consequence of surface

microbe activity is the production of gaseous metabolites such as carbon dioxide and

ammonia (Karahadian et al. 1987; Riahi et al. 2007). These alkaline surface conditions,

combined with the aforementioned minerals and gaseous components, result in various

surface-situated crystals being formed on washed rind cheeses. Ikaite (calcium carbonate

hexahydrate - CaCO3•6H2O), Calcite (calcium carbonate - CaCO3), Struvite (magnesium

ammonium phosphate hexahydrate - NH4MgPO4•6H2O), and Brushite (dicalcium

phosphate dihydrate - CaHPO4•2H2O) have been previously identified on soft washed rind

cheeses (Tansman et al. 2017a, b).

While the identity of these crystals have been confirmed using a robust set of

methodologies including powder X-ray diffractometry (PXRD), single crystal X-ray

diffractometry, and polarized light microscopy (PLM), their incidence have only been

confirmed in a small number of cheeses from a few producers (Tansman et al. 2015, 2017a,

b; Polowsky et al. 2018). Moreover, the direct effect of these crystals on the sensorial

80

properties of washed rind cheeses remains an unresolved question. Anecdotal reports from

cheese makers, cheese mongers, and the popular press indicate that a relatively common

defect of this cheese variety is a gritty or sandy mouthfeel associated with the rind (M.

Kehler, Cellars at Jasper Hill, Greensboro, VT, personal communication). The authors have

found only one mention in scientific literature referencing this proposedly common defect

(Le Mens et al. 2000), but none studying its possible linkage to surface crystals. The present

study aims at addressing this dearth in knowledge. Previous work demonstrated that surface

crystals on washed rind cheeses can grow to large sizes, on the order of hundreds of

microns (Tansman et al. 2017a; Polowsky et al. 2018). This size of crystal most likely

exceeds the sensory perception threshold, which has been reported in food products to be

anywhere from ~20μm to ~105μm depending on product and number of crystals present

(Vaisey-Genser et al. 1989; Hough et al. 1990; Goff and Hartel 2013). A systematic study

comparing crystal size to grittiness occurrence in smear ripened rind cheese is needed to

confirm what sensorial effects these crystals contribute to this cheese variety.

Looking broadly, the occurrence of calcium lactate crystals in cheddar cheese

(Johnson et al. 1990; Agarwal et al. 2006; Tansman et al. 2015) and amino acid crystals in

Parmesan-style cheeses (Zannoni et al. 1994; Noël et al. 1996) are well known in the dairy

technology world, and their effects on cheese quality have been studied to some extent.

Such work has yet to be done in the area of smear ripened cheeses. The present study sets

forth the following goals to address these gaps in understanding: (1) to establish “grittiness”

as a sensory attribute while assessing other key attributes for washed rind cheeses, (2) to

utilize PXRD and PLM techniques to understand the prevalence of surface crystals, (3) to

81

employ image analysis software to quantify the size and shape of these surface crystals,

and (4) to correlate the occurrence of grittiness with crystal size/shape/type.

MATERIALS AND METHODS

Cheese Samples

Commercial soft washed rind cheeses (n=24) were sourced from the United States

and Europe (France, Italy, Spain, and Germany). Replicate wheels of cheese, with the same

code data and vat identifier (if available), were used for all analyses. All samples were

purchased from retail establishments or were shipped overnight on ice, and inspected upon

arrival. Samples were stored at 3°C for up to five days until analyses were conducted. Each

of the cheeses used in this study will be referred to with a single letter (A through X).

Compositional Analysis

Surface pH measurements were taken on three separate locations (bottom, side, and

top) on each wheel using a flat-tipped probe (Thomas Scientific, Swedesboro, NJ), and

averaged for each cheese (six total pH measurements per cheese). Moisture content was

determined in triplicate using a forced air draft oven set to 100C. A cross-section from

each wheel was ground to homogeneity using a mortar and pestle. Approximately two

grams of homogenized sample were weighed and transferred into pre-dried and pre-

weighed aluminum pans. Samples were dried for 24 hours, or until a constant mass was

reached (six total moisture measurements per cheese).

82

Cheese Smear Collection and Processing

A 1cm2 rind section of smear material of each wheel was collected using a metal

spatula. Gentle pressure was used to limit the possibility of crystals being fractured. For

PXRD analysis, alkaline dispersions (pH=10) of smear samples were used. Dispersions

were prepared by placing the smear sample in a 50mL beaker and covering with 15mL of

0.0001M NaOH (Fisher Scientific, Pittsburgh, PA), which were then vortexed gently to

encourage good dispersion and allowed to rest at room temperature for approximately two

hours. Crystals that settled to the bottom of the beaker during this time were used for PXRD

analysis. Alkaline dispersion was used to lower the amount of amorphous material present

in the cheese smears, thereby lowering the background noise in the diffractograms and

improving PXRD performance overall. For PLM analysis, smear samples were applied

directly to glass microscope slides. Alkaline dispersion was not used for PLM analysis,

because our previous work (Chapter 2) indicated that crystal size may be reduced when an

alkaline dispersion methodology is used to harvest crystals, perhaps due to dissolution of

the crystals in the aqueous environment.

Powder X-ray Diffractometry

A 300μL aliquot of alkaline smear dispersion liquid (pipetted from bottom of

beaker) was applied to a zero-background diffraction slide for all PXRD analyses. Care

was taken to ensure the dispersion liquid was applied to the center of the slide in order to

allow for equal distribution of crystals on slide. Excess liquid was absorbed by blotting

with a tissue (Kimwipes, Kimberly-Clark Professional, Roswell, GA). Diffraction patterns

were generated using a Miniflex II powder X-ray diffractometer (Rigaku, The Woodlands,

83

TX). Patterns were created between 5° 2θ and 50° 2θ, using a speed of 2° 2θ/min. The

PDXL software package (Rigaku, The Woodlands, TX) was used to analyze the

diffractograms. Each experimental pattern was compared to existing reference diffraction

patterns in the International Center for Diffraction Data (ICDD) database. The following

ICDD reference cards were used: 01-074-7147 (ikaite), 01-074-3640 (brushite), 00-005-

0586 (calcite), 00-015-0762 (struvite), and 00-039-1840 (L-tyrosine). Peaks from

experimental patterns were compared to reference patterns until all peaks were identified.

Preferred orientation was present in several of the experimental diffraction patterns but did

not inhibit identification of major crystalline phases (Discussed at more length in Chapter

2).

Polarized Light Microscopy

Smear scrapings applied to a glass microscope slide (as described above) were

overlaid with a thin film of mineral oil in order to delay crystal dehydration. Micrographs

of smear scrapings were captured using a Nikon E200POL microscope (Nikon

Corporation, Tokyo, Japan) equipped with a rotating stage. Samples were viewed under

polarized light (crossed polarizers) using a 1/4λ plate (quartz filter), and images were

captured using a SPOT Idea 1.3 MP color camera (SPOT Imaging Solutions, Sterling

Heights, MI). Presumptive entities were confirmed to be crystals when they displayed

uniform extinction upon rotating the stage a full 360°. Four fields of view (FOV) were

captured for each cheese wheel duplicate. FOVs were chosen at random using a systematic

random sampling method (Altunkaynak et al. 2011). Gridded microscope slide stickers

(Sigma-Aldrich, St. Louis, MO) were applied to back side of each microscope slide. Grid

84

squares were chosen at random and centered in each FOV, and an image was captured

(eight images total were captured per cheese).

Image Analysis

Images from PLM were imported into Adobe Photoshop (Adobe Systems

Incorporated, San Jose, CA), where a black background was applied and crystals were

outlined using the Quick Selection tool. Crystals that were overlapping, crystals obscured

by smear material, and crystals not entirely in the FOV were not used for any further

analysis. Outlined images were saved as PNG files and analyzed using Fiji (ImageJ

distribution; Schindelin et al. 2012). Each image was converted to black and white (8 bit

images), and thresholding was applied using the Li's Minimum Cross Entropy thresholding

option (Li and Lee 1993). Each image was then analyzed using the particle analysis plugin.

Area, perimeter, and circularity measurements were collected and recorded for each

isolated crystal in each processed image. A minimum size threshold of 50 pixels2 was used,

objects smaller than this could not be accurately measured and were not used in any further

analysis.

Descriptive Sensory Analysis

A trained sensory panel (n=12) evaluated all samples in duplicate using the

SpectrumTM method (Meilgaard et al. 2007). A 15-point intensity scale was used for each

attribute. Sensory panelists were recruited from the staff of The Cellars at Japer Hill

(Greensboro, VT), and had extensive prior experience evaluating washed rind cheeses.

85

Each panelist underwent an additional 15 hours of training specific to this study. Attribute

definitions, evaluation techniques, and attribute references can be seen in Table 11.

A half-inch cross section of each cheese, consisting of two pieces of rind separated by the

cheese body, were stored in small soufflé cups labeled with random three digit codes. All

samples were evaluated at a temperature of 15C. Panelists expectorated all samples and

were provided with deionized water and unsalted crackers as palate cleansers. RedJade

software (RedJade Software Solutions, LLC, San Francisco, CA) was used to organize

descriptive sensory data and to track panelist performance. Grittiness threshold was

determined in a manner similar to that as described by Imai et al. (1995); based on the

number of panelists that perceived no grittiness. All sensory analysis was performed in

compliance of University of Vermont’s Institutional Review Board guidelines for human

subjects.

Statistical Analysis

All statistical analyses were performed using R (version 3.3.2; R Core Team, 2016).

Compositional data, crystal size/shape data, and descriptive sensory data were analyzed via

one-way ANONVA (using the aov function) with Tukey’s Honest Significance Difference

(HSD) test used to check for significant differences between means (P<0.05). The

agricolae package was used for the Tukey HSD test (version 1.2-8; de Mendiburu, 2017).

Principal component analysis (PCA) was also applied to these data using the FactoMineR

package (version 1.34; Lê, Josse, & Husson, 2008). PCA was carried out using the

correlation matrix. Pearson correlation coefficients were generated using the cor function.

86

Grittiness occurrence and PXRD data were analyzed via a chi-square test, using the vcd

package (version 1.4-4; Meyer, Zeileis, & Hornik, 2017).

RESULTS

Cheese Composition

Place of origin, moisture content, and pH values of the cheeses used in this study

are shown in Table 12. Moisture content ranged from approximately 40% to 54%, however

the majority of cheeses were at the higher end of that range. Although the moisture content

ranged widely, all cheeses exhibited the characteristic softening associated with soft

washed rind cheeses. pH ranged from approximately 6.6 to 8.0, which is also consistent

with this cheese variety. These differences could have resulted from differences in make

schedule, affinage practices, or the age at which the cheeses were examined.

Powder X-ray Diffraction

A representative diffractogram obtained from Cheese E can be seen in Figure 9,

which highlights the crystal peaks identified in a single cheese. Individual diffractograms

from all 24 cheeses are available as Supplement 1. Results from PXRD crystal

identification are summarized in Table 13. Crystals were detected in 23 out of the 24

(~96%) cheeses, with only one cheese (Cheese P) showing no identifiable peaks present in

its diffractogram. Overall, the following crystal types were identified: struvite, ikaite,

brushite, calcite, and the amino acid tyrosine. Of the samples with crystals present, struvite,

ikaite, brushite, calcite, and tyrosine had approximate rates of occurrence of 100%, 70%,

87

26%, 17%, and 4%, respectively. Worthy of mention is the occurrence of multiple crystal

types in a single cheese, which the majority (17 out of 24) of samples exhibited.

Polarized Light Microscopy and Image Analysis

The results of presumptive crystal identification via PLM, and crystal metrics

resulting from image analysis of PLM micrographs are shown in Table 13. PLM

presumptive identification of ikaite and struvite crystals was carried out using the criteria

described in Chapter 2, namely birefringence and angle of extinction. Such criteria have

not been developed for the other crystal types present in this study. Overall, there was good

agreement between PLM presumptive identification and PXRD identification. However,

there were 7 instances (Cheeses B, G, H, L, Q, V, and X) where PLM analysis failed to

detect ikaite and/or struvite when those crystal types were detected by PXRD. Of special

interest are Cheeses S and T, where PLM presumptively identified struvite/ikaite which

was not identified during PXRD analysis. Possible reasons for these discrepancies will be

discussed below.

Crystal length and crystal area varied by two to three orders of magnitude,

respectively. Crystal length ranged from approximately 27μm to 1100μm, with an overall

average value of approximately 235μm. Crystal area ranged from approximately 500μm2

to 200,000um2, with an overall average value of approximately 44000μm2. Crystal

circularity ranged from approximately 0.480 to 0.960, with an overall average value of

approximately 0.760. Circularity, in ImageJ, is calculated using the following formula:

4pi(area/perimeter^2), where circularity of 1.00 indicates a perfect circle, and circularity

88

of 0.00 indicates an increasingly extended polygon. Figure 10 shows an example of PLM

micrographs that were used for presumptive crystal identification and image analysis.

Intra- and inter-sample differences in crystal shape, crystal size, and coloring are clearly

visible. These results indicate a diverse collection of crystal sizes and morphologies existed

on the surfaces of the soft washed rind cheeses used in this study.

Descriptive Sensory Analysis

Table 14 contains the results from the descriptive sensory analysis. All attributes,

except grittiness, were detected in each of the cheeses evaluated. Mean grittiness scores

varied widely depending on cheese, with a range from approximately 0.0 to 7.4. Panelists

determined that two of the cheeses (P and Q) contained no perceptible grittiness, and had

mean values of 0 for grittiness. Five other cheeses (H, K, L, R, and V) had very low

grittiness scores (below 0.5), with a majority of the panelists indicating 0 for grittiness. The

sensory threshold of grittiness was established in a manner similar to that described by Imai

et al. (1995). Cheeses where at least 75% of panelists perceived zero grittiness in both

replicates were considered below the threshold and therefore not gritty. Mean scores for

ammonia aroma, bitterness, and salt also varied somewhat widely, with approximate ranges

of 1.1 to 8.5, 1.4 to 9.2, and 2.7 to 7.0, respectively.

Principal component analysis (PCA) of sensory data, cheese compositional data,

and crystal size/shape data was performed (Figure 11). Principal component (PC) 1

accounted for approximately 32% of the variation in the data, and was composed mainly

of crystal area, crystal length, and grittiness. PC 2 explained approximately 24% of the

89

variation, and was composed mainly of ammonia aroma, grittiness, and salt. PC 3 and 4

were composed of acid, bitter, salt, moisture and crystal circularity (Appendix B).

Approximately 25% of the variation in the data was explained by PC 3 and 4 (data not

shown).

DISCUSSION

Significance of Crystal Types Identified

Ikaite, calcite, struvite, and brushite were found on multiple cheeses used in this

study. Each of these crystal types have been identified previously on soft washed rind

cheese (Tansman et al. 2017a, b). Previous studies have included only a few cheeses,

largely from one producer. This study incorporated 24 cheeses from 17 different producers

located in the U.S. and Europe. The results of the present work indicate that these crystals

are common among soft washed rind cheeses, and are not unique to a certain producer or

region. Instead, these crystals are the result of physico/bio-chemical reactions common in

this cheese variety, namely surface microbe metabolism and mineral migration.

A few consequences of this microbe metabolism include: deacidification of the

cheese surface and subsequent mineral precipitation, as well as proteolysis (McSweeney

2004). The surface precipitation/crystallization establishes concentration gradients, which

encourages radial migration of mineral elements, such as calcium, magnesium, and

phosphorus (Tansman et al. 2017b). Gaseous products, including carbon dioxide and

ammonia, are also formed during ripening of washed rind cheeses (Gori et al. 2007; Hélias

et al. 2007). These surface-situated minerals (i.e. Ca, Mg, and PO4) and gaseous products

90

(CO2 and NH3) are the requisite components of the lattice structures of the various observed

crystals. Large concentrations of these components accumulate on the cheese surface and

most likely surpass the solubility threshold and form precipitates and crystals.

The pronounced occurrence of struvite is notable; every sample that contained

crystals contained struvite. Struvite has been reported in other biological systems, notably

waste water processing (Kim et al. 2016), urinary stones (Wilsenach et al. 2007), and fish

products (Amano 1950). While a main component of struvite, magnesium, occurs in cheese

at much lower amounts than calcium, these crystals were omnipresent in the samples used

in this study. Large amounts of phosphate and ammonia, along with struvite’s low

solubility at neutral/high pH (Wu et al. 2005), might account for this. Calcite, and its

metastable form ikaite were present in many of the samples. These crystals are both of the

type calcium carbonate, which are also sparingly soluble at the neutral/high pH range. As

already mentioned, the surface accumulation of calcium has been well established in

surface ripened cheeses (Tansman et al. 2017c, b). Carbon dioxide content of cheese

ripening spaces can reach high levels (Picque et al. 2006), which can dissolve into the water

phase at the cheese surface. At high enough concentration, these components exceed

solubility thresholds and crystallize out. Of special note is the formation of metastable

ikaite, which has rarely been seen in nature (Rysgaard et al. 2012). Ikaite and calcite are

known to transform into each other depending on reaction conditions, and the presence of

phosphate can stabilize metastable ikaite (Hu et al. 2014). It is possible that the surface

smear of washed rind cheeses presents an optimal environment for ikaite to form. More

research is needed to determine if that’s the case.

91

Brushite crystals are likely the easiest to account for on the surface of washed rind

cheeses. Calcium and phosphate are major structural components of the casein matrix

(Lucey and Fox 1993), and undergo the aforementioned radial migration. Brushite is also

insoluble in the pH range of many washed rind cheeses (Arifuzzaman and Rohani 2004).

While brushite is a metastable form of calcium phosphate, certain chemical kinetic

conditions can encourage its formation, such as concentration of ions and pH (Arifuzzaman

and Rohani 2004). Tyrosine crystals were found on only one sample (Cheese N). While

this crystal type has not been identified previously on soft washed rind cheese, it commonly

occurs in Grana-style cheeses (Tansman et al. 2015). A consequence of the surface

microbiota’s proteolytic behavior could be the accumulation of free amino acids (Brennan

et al. 2002), such as tyrosine. High concentration of free tyrosine, coupled with its low

solubility at neutral/alkaline pH (Hitchcock 1924), may explain its presence. However, it

is unclear why tyrosine was only found on one sample. A possible explanation may be the

specific microbial metabolism of Cheese N resulted in unusually high amounts of free

tyrosine.

Grittiness Threshold and Descriptive Sensory Analysis

To the authors’ knowledge, surface grittiness in washed rind cheese has only been

mentioned once in the dairy technology literature (Le Mens et al. 2000), but never

associated with crystal growth. Although the occurrence of this attribute has not been

thoroughly confirmed, it is a recognized attribute in the cheese industry. In order to

accurately measure grittiness as a descriptive sensory attribute, a novel method was needed

to establish reference samples for grittiness. Mixtures of vegetable shortening with

92

different grain sizes of sugar were used in the present work as anchor points (Table 11).

Vegetable shortening was chosen rather than other dairy-containing matrices (i.e. cream

cheese) due to its low moisture content, which ensured crystals did not dissolve over time.

Sugar was chosen as the crystalline inclusion due to its mild sweet taste, which would not

interfere with other common taste attributes associated with washed rind cheeses (e.g. salt

and bitter). The descriptive panel agreed that these reference samples allowed for consistent

grittiness evaluation, with minimal chance of carryover effects when used in conjunction

with evaluating cheese samples.

Perceived grittiness varied by a large amount across the 24 samples evaluated in

this study (Figure 12), from no grittiness to pronounced grittiness. Seven cheeses met the

criteria for the absence of grittiness (Cheeses P, Q, K, H, V, L, and R). Coincidently, these

seven cheeses had mean grittiness values below 0.5, whereas cheeses above the threshold

had grittiness values of at least 1.5. Figure 12 indicates this threshold with a dashed line.

In the present study, 71% of cheeses were considered gritty. This relatively high occurrence

of grittiness confirms what anecdotal evidence suggests; that is, grittiness is a common

sensorial attribute associated with soft washed rind cheeses. PCA of crystal metrics and

grittiness data (Figure 11) indicate a strong association between crystal size and grittiness.

Comparing the occurrence of grittiness to mean crystal size allows for an approximate

estimate of the threshold size of crystals, above which grittiness could occur. Figure 13

shows the mean size distribution (length and area) of crystals across all 24 cheese samples.

The grittiness threshold is indicated with a dashed line, which occurred at approximately

at 66μm and 2900 μm2. The grittiness threshold is indicated with a dashed line. The seven

93

cheeses that fell below the sensory threshold for grittiness also had the lowest mean crystal

lengths and areas of the 24 surveyed cheeses. These crystal sizes serve as approximate size

thresholds, and mean crystal values at or above this level would likely result in gritty

cheese. It’s important to note that results from a descriptive panel don’t necessarily equate

to how consumers perceive grittiness. Further studies are needed to determine at what

crystal size consumers perceive grittiness, and at what grittiness intensity are cheeses

considered objectionable.

Other trends of interest include the association between ammonia aroma and

surface pH, as well as ammonia aroma and salty taste (Figure 11). These metrics were

major contributors to PC 1 and PC 2. There was a strong, highly significant positive

correlation between ammonia aroma and surface pH (r=0.79; P<0.0001). This is an

expected result since surface ammonia production is considered a main cause of rind

alkalinization (McSweeney 2004). The relationship between ammonia aroma and salty

taste showed a moderate, very significant positive correlation (r=0.60; P=0.002). This

indicates that ammonia aroma might have altered the perception of saltiness as perceived

by the panelists. Aroma is known to influence taste perception and can increase saltiness

intensity depending on the context (Lawrence et al. 2009, 2011). This could be the case for

ammonia aroma and perceived saltiness of washed rind cheeses, and this potential

relationship warrants further study.

Occurrence of Grittiness

Grittiness occurring at a crystal size threshold of approximately 66um is in-line

94

with thresholds established in other products. Grittiness in ice cream is caused by ice

crystals that exceed 45μm (Goff and Hartel 2013). Hough et al. (1990) found sandiness

occurred in Dulce de Leche when lactose crystals were in the size range of 45μm to 105μm,

depending on the number of crystals. The sensory perception of grittiness is a complex

physiological phenomenon. Particle size, shape, hardness, and viscosity of dispersion

medium can influence grittiness perception (Engelen et al. 2005).

For cheeses above the grittiness threshold, grittiness was highly correlated with

crystal length and crystal area (Table 15). Pearson correlation coefficients for grittiness and

crystal length and crystal area were 0.927 and 0.957, respectively. Both of these

correlations also had high statistical significance (P<0.001). Of note is a higher correlation

of crystal area to grittiness, compared to crystal length and grittiness. This might indicate

that capturing crystal size data in two dimensions better represents how it is perceived in a

three-dimensional context (i.e. sensory evaluation). A downside of this approach is the

necessity to employ image analysis, which requires advanced training of personnel. One-

dimensional measurements (e.g. length, diameter) can be gathered directly by the

experimenter, using a microscope and calibrated ocular. Looking at PCA and correlation

coefficients, crystal circularity was poorly associated with grittiness (r=-0.341; P=0.181).

This could indicate crystal shape was not a causative factor for grittiness occurrence, and

crystal size played a much larger role.

A chi-square test was employed in order to gauge if crystal type had a large

influence on grittiness perception (Table 16). The presence of each crystal type, along with

grittiness occurrence, was coded in a binary format (i.e. yes or no) for each of the 23

95

cheeses that contained crystals. Phi-coefficients and likelihood ratio probabilities were

determined for the influence of ikaite, calcite, brushite, and struvite on grittiness. Tyrosine

was not used due to it occurring in only one sample. Each crystal type had relatively low

phi-coefficients (<0.3) indicating small effect sizes across the various crystal types. This

could indicate that grittiness will occur regardless of crystal type, and that crystal size is a

larger determining factor. The occurrence of multiple crystals types in a single product is

unique in the dairy product landscape. Understanding the dynamic relationship between

crystal type and crystal growth rate would aid in determining how to control potential

grittiness, and warrants further study.

CONCLUSION

The present work indicates surface-situated crystals are common to the soft washed rind

cheese category. These crystals induce a grittiness sensory sensation, which is largely

influenced by crystal size. Further research is needed to determine how crystal growth can

be controlled in a deliberate fashion, and tailored to prevent grittiness. Crystals are

composed of major structural minerals, and manipulating their growth dynamics could aid

in controlling texture development, such as softening and liquefaction.

ACKNOWLEDGMENTS

This study was funded by USDA Hatch Project 033286. The National Science

Foundation is acknowledged for support of grant EAR-0922961 for the purchase of the

powder X-ray diffractometer. Adèle Berrier and the rest of the staff at The Cellars at Jasper

Hill are gratefully acknowledged for their support with sensory evaluation.

96

REFERENCES

Agarwal S, Powers JR, Swanson BG, et al (2006) Cheese pH, Protein Concentration, and

Formation of Calcium Lactate Crystals. J Dairy Sci 89:4144–4155. doi:

10.3168/JDS.S0022-0302(06)72459-6

Altunkaynak BZ, Onger ME, Altunkaynak ME, et al (2011) A Brief Introduction to

Stereology and Sampling Strategies: Basic Concepts of Stereology.

NeuroQuantology. doi: 10.14704/nq.2012.10.1.427

Amano K (1950) A Note on Sodium Phosphate Crystals Found in Mild-cured Salmon.

Nippon SUISAN GAKKAISHI 15:638–641. doi: 10.2331/suisan.15.638

Arifuzzaman S., Rohani S (2004) Experimental study of brushite precipitation. J Cryst

Growth 267:624–634. doi: 10.1016/J.JCRYSGRO.2004.04.024

Brennan NM, Cogan TM, Loessner M, Scherer S (2004) Bacterial surface-ripened cheeses.

In: Cheese: Chemistry, Physics and Microbiology. pp 199–225

Brennan NM, Ward AC, Beresford TP, et al (2002) Biodiversity of the bacterial flora on

the surface of a smear cheese. Appl Environ Microbiol 68:820–30. doi:

10.1128/AEM.68.2.820-830.2002

de Mendiburu F (2017) agricolae: Statistical Procedures for Agricultural Research.

Dugat-Bony E, Garnier L, Denonfoux J, et al (2016) Highlighting the microbial diversity

of 12 French cheese varieties. Int J Food Microbiol 238:265–273. doi:

10.1016/j.ijfoodmicro.2016.09.026

Engelen L, de Wijk RA, van der Bilt A, et al (2005) Relating particles and texture

perception. Physiol Behav 86:111–117. doi: 10.1016/j.physbeh.2005.06.022

Goff HD, Hartel RW (2013) Ice cream, 7th edn. Springer

Gori K, Mortensen HD, Arneborg N, Jespersen L (2007) Ammonia Production and Its

Possible Role as a Mediator of Communication for Debaryomyces hansenii and Other

Cheese-Relevant Yeast Species. J Dairy Sci 90:5032–5041. doi: 10.3168/jds.2006-

750

Hélias A, Mirade P-S, Corrieu G, et al (2007) Modeling of Camembert-type cheese mass

loss in a ripening chamber: main biological and physical phenomena. J Dairy Sci

90:5324–33. doi: 10.3168/jds.2007-0272

Hitchcock DI (1924) THE SOLUBILITY OF TYROSINE IN ACID AND IN ALKALI. J

Gen Physiol 6:747–57.

97

Hough G, Martinez E, Contarini A (1990) Sensory and Objective Measurement of

Sandiness in Dulce de Leche, a Typical Argentine Dairy Product. J Dairy Sci 73:604–

611. doi: 10.3168/jds.S0022-0302(90)78709-7

Hu Y-B, Wolf-Gladrow DA, Dieckmann GS, et al (2014) A laboratory study of ikaite

(CaCO3·6H2O) precipitation as a function of pH, salinity, temperature and phosphate

concentration. Mar Chem 162:10–18. doi: 10.1016/J.MARCHEM.2014.02.003

Imai E, Hatae K, Shimada A (1995) Oral perception of grittiness: effect of particle size and

concentration of the dispersed particles and the dispersion medium. J Texture Stud

26:561–576. doi: 10.1111/j.1745-4603.1995.tb00804.x

Johnson ME, Riesterer BA, Chen C, et al (1990) Effect of Packaging and Storage

Conditions on Calcium Lactate Crystallization on the Surface of Cheddar Cheese. J

Dairy Sci 73:3033–3041. doi: 10.3168/jds.S0022-0302(90)78990-4

Karahadian C, Lindsay RC, Ades GL, et al (1987) Integrated Roles of Lactate, Ammonia,

and Calcium in Texture Development of Mold Surface-Ripened Cheese. J Dairy Sci

70:909–918. doi: 10.3168/jds.S0022-0302(87)80094-2

Kim D, Min KJ, Lee K, et al (2016) Effects of pH, molar ratios and pre-treatment on

phosphorus recovery through struvite crystallization from effluent of anaerobically

digested swine wastewater. Environ Eng Res 22:12–18. doi: 10.4491/eer.2016.037

Lawrence G, Salles C, Palicki O, et al (2011) Using cross-modal interactions to

counterbalance salt reduction in solid foods. Int Dairy J 21:103–110. doi:

10.1016/J.IDAIRYJ.2010.09.005

Lawrence G, Salles C, Septier C, et al (2009) Odour–taste interactions: A way to enhance

saltiness in low-salt content solutions. Food Qual Prefer 20:241–248. doi:

10.1016/J.FOODQUAL.2008.10.004

Lê S, Josse J, Husson F (2008) FactoMineR: An R Package for Multivariate Analysis. J

Stat Softw 25:1–18. doi: 10.18637/jss.v025.i01

Le Graet Y, Brule G (1988) Migration des macro et oligo-éléments dans un fromage à pâte

molle de type Camembert. Lait 68:219–234. doi: 10.1051/lait:1988215

Le Graet Y, Lepienne A, Brule G, Ducruet P (1983) Migration du calcium et des

phosphates inorganiques dans les fromages à pâte molle de type Camembert au cours

de l’affinage. Lait 63:317–332. doi: 10.1051/lait:1983629-63019

Le Mens P, Parguel P, Tormo H, et al (2000) Description sensorielle de trois fromages

fermiers : le Banon, la Brique du Forez et le Rogeret. Renc Rech Ruminants 7:300–

303.

98

Li CH, Lee CK (1993) Minimum cross entropy thresholding. Pattern Recognit 26:617–

625. doi: 10.1016/0031-3203(93)90115-D

Lucey JA, Fox PF (1993) Importance of Calcium and Phosphate in Cheese Manufacture:

A Review. J Dairy Sci 76:1714–1724. doi: 10.3168/jds.S0022-0302(93)77504-9

McSweeney PLH (2004) Biochemistry of cheese ripening. Int J Dairy Technol 57:127–

144. doi: 10.1111/j.1471-0307.2004.00147.x

Meilgaard M, Civille GV, Carr BT (2007) Sensory evaluation techniques. Taylor & Francis

Meyer D, Zeileis A, Hornik K (2017) vcd: Visualizing Categorical Data.

Noël Y, Zannoni M, Hunter EA (1996) Texture of Parmigiano Reggiano cheese: Statistical

relationships between rheological and sensory variates. Lait 76:243–254. doi:

10.1051/lait:1996320

Picque D, Leclercq-Perlat M-N, Corrieu G (2006) Effects of Atmospheric Composition on

Respiratory Behavior, Weight Loss, and Appearance of Camembert-Type Cheeses

During Chamber Ripening. J Dairy Sci 89:3250–3259. doi: 10.3168/jds.S0022-

0302(06)72600-5

Polowsky PJ, Tansman GF, Kindstedt PS, Hughes JM (2018) Characterization and

presumptive identification of Surface Crystals on Smear Ripened Cheese by Polarized

Light Microscopy. Manuscript in Preparation.

R Core Team (2016) R: A language and environment for statistical computing.

Riahi MH, Trelea IC, Leclercq-Perlat M-N, et al (2007) Model for changes in weight and

dry matter during the ripening of a smear soft cheese under controlled temperature

and relative humidity. Int Dairy J 17:946–953. doi: 10.1016/J.IDAIRYJ.2006.11.002

Rysgaard S, Glud RN, Lennert K, et al (2012) Ikaite crystals in melting sea ice –

implications for pCO 2 and pH levels in Arctic surface waters. Cryosph 6:901–908.

doi: 10.5194/tc-6-901-2012

Schindelin J, Arganda-Carreras I, Frise E, et al (2012) Fiji: an open-source platform for

biological-image analysis. Nat Methods 9:676–682. doi: 10.1038/nmeth.2019

Tansman GF, Kindstedt PS, Hughes JM (2015) Crystal fingerprinting: elucidating the

crystals of Cheddar, Parmigiano-Reggiano, Gouda, and soft washed-rind cheeses

using powder x-ray diffractometry. Dairy Sci Technol 95:651–664. doi:

10.1007/s13594-015-0225-6

Tansman GF, Kindstedt PS, Hughes JM (2017a) Minerals In Food: Crystal Structures of

Ikaite and Struvite From Bacterial Smears On Washed-Rind Cheese. Can Mineral

99

55:89–100. doi: 10.3749/canmin.1600038

Tansman GF, Kindstedt PS, Hughes JM (2017b) Crystallization and demineralization

phenomena in washed-rind cheese. J Dairy Sci. doi: 10.3168/jds.2017-13067

Tansman GF, Kindstedt PS, Hughes JM (2017c) Crystallization and demineralization

phenomena in stabilized white mold cheese. J Dairy Sci 100:6074–6083. doi:

10.3168/jds.2016-12259

Vaisey-Genser M, Vane BK, Johnson S (1989) GRAININESS, CRYSTAL SIZE, AND

FIRMNESS OF STORED CANOLA OIL MARGARINES. J Texture Stud 20:347–

361. doi: 10.1111/j.1745-4603.1989.tb00445.x

Wilsenach JA, Schuurbiers CAH, van Loosdrecht MCM (2007) Phosphate and potassium

recovery from source separated urine through struvite precipitation. Water Res

41:458–466. doi: 10.1016/J.WATRES.2006.10.014

Wolfe BE, Button JE, Santarelli M, Dutton RJ (2014) Cheese Rind Communities Provide

Tractable Systems for In Situ and In Vitro Studies of Microbial Diversity. Cell

158:422–433. doi: 10.1016/j.cell.2014.05.041

Wu Q, Bishop PL, Keener TC (2005) A Strategy for Controlling Deposition of Struvite in

Municipal Wastewater Treatment Plants. Water Environ Res 77:199–207. doi:

10.2175/106143005X41771

Zannoni M, Bertozzi L, Hunter EA (1994) Comparison of Parmigiano-Reggiano and

American Parmesan cheeses by sensory analysis of texture. Sci e Tec Latt Casearia

45:505–518.

100

Table 11. Attribute lexicon used for descriptive sensory analysis. .

Attribute Definition Technique References (0-15 scale)

Ammonia

Aroma

Aroma associate with

NH3 and surface

microbe metabolism

Open sample container,

waft for 3 seconds

0 = water

5 = 2.5% ammonia solution1

10 = 10% ammonia solution

Grittiness

The size of the small,

firm particulates which

don’t dissolve

immediately upon

manipulation in the

mouth

Place sample with rind

positioned against

molars, upon the first

chew where grittiness is

perceived, chew two

more times. Average the

grittiness felt during

chewing

0 = cream cheese2

8 = 75g vegetable shortening3

mixed with 3g super-fine sugar4

12 = 75g vegetable shortening

mixed with 3g granulated sugar5

Bitter Basic taste elicited by

caffeine

Assess the taste of the

rind

0 = water

1.5 = 0.015% caffeine6 solution

4.0 = 0.030% caffeine solution

7.5 = 0.050% caffeine solution

9.0 = 0.075% caffeine solution

Acid Basic taste elicited by

lactic acid

Assess the taste of the

cheese body

0 = water

4 = 0.075% lactic acid7 solution

8 = 0.100% lactic acid solution

Salt Basic taste elicited by

sodium chloride

Assess the taste of the

rind

0 = water

2 = 0.25% NaCl8 solution

5 = 0.50% NaCl solution

11 = 1.00% NaCl solution 1Hannaford Ammonia Clear; Delhaize Group, Brussels, Belgium 2Philadelphia Cream Cheese; Kraft Heinz, Chicago, IL 3Crisco Vegetable Shortening; J.M. Smucker Company, Orrville, OH 4Domino Superfine Sugar; American Sugar Refining, Inc., West Palm Beach, FL 5Domino Granulated Sugar; American Sugar Refining, Inc., West Palm Beach, FL 6Caffeine Powder (USP/FCC); Fisher Scientific, Hampton, NH 7Lactic Acid (88% Liquid); Nelson-Jameson, Marshfield, WI 8Morton Salt (uniodized); Morton Salt, Chicago, IL

101

Table 12. Production location and compositional information of cheeses surveyed in this

study. Means and standard deviations (SD) are results from triplicate measurements from

duplicate wheels of cheese.

Cheese Production Location Surface pH Moisture Content (%)

Mean SD Mean SD

A Vermont, USA 7.23 0.16 47.6 0.39

B New York, USA 8.03 0.19 44.8 0.80

C Virginia, USA 7.28 0.02 48.0 0.69

D Vermont, USA 7.91 0.11 51.3 0.13

E Vermont, USA 7.70 0.12 53.8 0.30

F Vermont, USA 6.96 0.07 47.9 0.89

G California, USA 6.49 0.09 46.9 0.49

H Wisconsin, USA 7.18 0.09 45.9 0.76

I Minnesota, USA 7.67 0.06 47.3 0.11

J Wisconsin, USA 7.53 0.03 49.1 0.83

K California USA 7.45 0.05 47.8 0.14

L Indiana, USA 7.66 0.08 52.6 0.13

M New York, USA 7.98 0.09 49.7 0.20

N Colorado, USA 7.50 0.01 50.4 0.35

O Colorado, USA 7.58 0.04 40.2 0.14

P Indiana, USA 7.67 0.16 53.2 0.35

Q France 7.19 0.08 48.6 0.08

R France 7.19 0.01 54.5 0.20

S Italy 7.10 0.03 50.5 0.37

T Italy 7.32 0.05 52.9 0.78

U France 7.12 0.04 49.4 0.12

V France 6.63 0.11 46.3 0.08

W Spain 6.82 0.05 53.2 0.09

X Germany 6.64 0.04 48.1 0.19

HSD1 0.36 1.84 1 Means that differ by HSD within each column are significantly different (P<0.05)

102

Table 13. Summary of PLM data (crystal length, crystal area, circularity) and PXRD

observations (crystals types). SD = standard deviation.

Cheese

Crystal Length

(μm) Crystal Area (μm2) Circularity Crystal Types

Mean SD Mean SD Mean SD PXRD PLM3

A 161.93 47.2 14336.10 5604.0 0.758 0.059 Ikaite, Struvite Ikaite, Struvite

B 103.36 9.7 7737.09 1174.9 0.847 0.006 Struvite *

C 339.40 83.5 69735.42 42263.3 0.702 0.046 Ikaite, Struvite,

Calcite, Brushite Ikaite, Struvite

D 262.75 27.2 34670.89 3078.3 0.729 0.035 Ikaite, Struvite,

Brushite Ikaite, Struvite

E 491.63 113.0 132371.38 39588.7 0.671 0.049 Ikaite, Struvite Ikaite, Struvite

F 1096.25 470.2 213969.24 52565.5 0.667 0.113 Ikaite, Struvite,

Brushite Ikaite, Struvite

G 276.05 34.1 52369.18 8138.2 0.812 0.019 Ikaite, Struvite,

Brushite *

H 48.85 14.0 1511.81 712.8 0.844 0.041 Ikaite, Struvite Struvite

I 133.73 6.2 10541.91 605.6 0.815 0.007 Struvite Struvite

J 295.40 26.3 49304.84 11916.1 0.755 0.013 Ikaite, Struvite,

Brushite Ikaite, Struvite

K 65.53 10.8 2912.76 752.1 0.903 0.012 Struvite Struvite

L 34.83 4.2 888.01 157.9 0.958 0.017 Ikaite, Struvite Struvite

M 332.95 46.9 65930.50 23531.8 0.763 0.028 Ikaite, Struvite,

Calcite Ikaite, Struvite

N 84.46 9.4 4980.13 863.5 0.849 0.008 Ikaite, Struvite,

Tyrosine Ikaite, Struvite

O 436.32 34.7 57119.53 16612.2 0.481 0.022 Ikaite, Struvite,

Calcite Ikaite, Struvite

P - - - - - - ND2 ND

Q 27.23 1.6 532.83 96.6 0.887 0.006 Ikaite, Struvite Struvite

R 33.47 3.3 728.96 146.7 0.848 0.015 Ikaite, Struvite Ikaite, Struvite

S 192.58 42.7 20211.37 6361.8 0.747 0.051 Ikaite Ikaite, Struvite

T 181.92 5.7 21442.45 695.5 0.798 0.007 Struvite Ikaite, Struvite

U 86.08 15.0 5265.65 1341.7 0.800 0.007 Struvite, Calcite Struvite

V 48.58 2.2 1644.98 179.0 0.867 0.000 Struvite *

W 443.61 142.1 140537.66 86623.9 0.808 0.028 Ikaite, Struvite *

X 465.48 73.3 152602.10 39580.2 0.822 0.002 Struvite, Brushite Struvite

HSD1 437.96 106254 0.144

1Means that differ by HSD within each column are significantly different (P<0.05) 2No crystals detected 3Crystals identified via polarized light microscopy methodology as described in Chapter 2.

*Crystals were detected, but no conclusive identification could be made

103

Table 14. Summary of descriptive sensory analysis data. Data are the means and standard

deviations (SD) of panel duplicates.

Cheese Ammonia Aroma Grittiness Bitter Acid Salt

Mean SD Mean SD Mean SD Mean SD Mean SD

A 3.94 0.08 2.56 0.16 9.11 0.39 3.69 0.35 4.22 0.63

B 8.47 0.20 1.64 1.30 6.42 1.37 2.89 0.47 5.56 0.00

C 2.94 0.08 3.37 0.57 5.89 1.81 3.81 0.82 3.94 0.86

D 7.47 0.51 3.06 1.10 9.03 1.06 4.06 0.55 5.75 0.20

E 6.72 0.24 3.94 0.63 5.69 0.27 3.81 0.04 5.89 0.31

F 3.08 0.98 7.36 0.43 6.00 0.47 3.17 0.24 3.89 0.31

G 2.44 0.47 2.83 0.55 3.94 0.08 3.00 0.47 5.28 0.24

H1 3.83 0.39 0.11 0.00 3.42 0.12 2.42 0.27 4.42 0.04

I 4.89 0.16 1.78 0.79 8.17 0.55 4.22 0.47 3.78 0.31

J 7.39 0.24 3.72 1.65 6.64 0.59 5.22 1.41 6.67 0.00

K1 4.75 0.12 0.11 0.00 3.53 0.20 4.03 0.27 4.72 0.00

L1 4.56 0.16 0.22 0.16 2.61 0.16 5.17 0.55 7.00 0.79

M 7.28 0.08 2.67 0.47 5.06 0.55 4.39 0.08 6.36 0.12

N 4.56 0.16 1.92 0.27 3.57 0.81 5.72 0.24 6.56 0.16

O 4.64 0.51 3.61 0.55 2.58 0.35 5.36 0.12 6.72 0.16

P1 7.36 0.82 0.00 0.00 7.00 0.94 5.06 0.86 5.86 0.27

Q1 5.50 0.00 0.00 0.00 4.00 1.89 4.67 0.47 6.17 0.35

R1 6.67 0.24 0.42 0.59 9.21 0.18 3.00 0.24 6.50 0.47

S 2.25 0.59 2.13 0.18 1.42 0.24 3.17 0.24 3.67 0.24

T 4.00 0.00 2.04 1.24 2.33 0.71 3.75 0.12 5.46 0.41

U 5.75 1.77 1.58 0.35 7.83 0.47 2.92 0.59 3.67 0.00

V1 1.08 0.12 0.17 0.00 1.46 0.77 4.58 1.30 2.67 0.24

W 4.92 0.35 5.46 0.65 3.63 0.06 4.67 0.94 5.67 0.24

X 1.58 0.12 5.33 1.18 3.67 0.24 4.88 0.41 4.67 0.47

HSD2 2.12 2.87 3.16 2.43 1.49

1Cheeses with grittiness intensity below sensory threshold. (i.e. At least 75% of panelists indicated

“0” for grittiness intensity) 2Means that differ by HSD within each column are significantly different (P<0.05)

104

Table 15. Pearson correlation coefficients and correlation probabilities (in parentheses below) of

crystal size/shape data and grittiness sensory intensity for cheeses above grittiness threshold.

Measurement Grittiness Crystal Length Crystal Area Crystal Circularity

Grittiness 1.000 0.927 0.957 -0.341

(<0.001) (<0.001) (<0.001) (0.181)

Crystal Length 0.927 1.000 0.922 -0.487

(<0.001) (<0.001) (<0.001) (0.048)

Crystal Area 0.957 0.922 1.000 -0.287

(<0.001) (<0.001) (<0.001) (0.264)

Crystal

Circularity

-0.341 -0.487 -0.287 1.000

(0.181) (0.048) (0.264) (<0.001)

105

Table 16. Chi-square test results: phi-coefficients and (significance

probabilities).

Ikaite Calcite Brushite Struvite

Grittiness 0.037 0.273 0.178 0.127

(0.858) (0.100) (0.375) (0.431)

106

Figure 9. An example PXRD pattern captured from crystals isolated from cheese E. Characteristic

peaks of ikaite and struvite are marked with squares and triangles, respectively. AU=arbitrary units.

See Appendix A for diffractograms from each cheese.

5 10 15 20 25 30 35 40 45 50

Inte

ns

ity

(A

U)

2-theta (°)

Diffractogram – Cheese E

107

Figure 10. Polarized light micrographs of crystals isolated from cheese smear material. Presumptive

crystals of ikaite (solid arrows) and struvite (hollow arrows) are shown. Panel A contains crystals

high in circularity (rounder crystals) while panel B contains crystals lower in circularity (more boxy

or needle-like crystals). Scale bars represent 250μm.

108

Figure 11. Principal component analysis (PCA) bipot of cheese compositional data, sensory data, and

crystal size/shape metrics. Letters represent cheese samples. PC 1 and PC 2 are shown.

109

Figure 12. Mean grittiness scores for cheeses used in this study. Error bars show one standard

deviation from mean. Cheeses with grittiness value below sensory threshold had at least 75% of

panelists indicated ”0” for grittiness.

110

Figure 13. Mean crystal area and mean crystal length measurements for cheeses used in this study.

Error bars show one standard deviation from mean.

111

COMPREHENSIVE WORKS CITED

Agarwal S, Powers JR, Swanson BG, et al (2006) Cheese pH, Protein Concentration, and

Formation of Calcium Lactate Crystals. J Dairy Sci 89:4144–4155. doi:

10.3168/JDS.S0022-0302(06)72459-6

Agarwal S, Powers JR, Swanson BG, et al (2006) Cheese pH, Protein Concentration, and

Formation of Calcium Lactate Crystals. J Dairy Sci 89:4144–4155. doi:

10.3168/JDS.S0022-0302(06)72459-6

Agarwal S, Powers JR, Swanson BG, et al (2008) Influence of Salt-to-Moisture Ratio on

Starter Culture and Calcium Lactate Crystal Formation. J Dairy Sci 91:2967–2980.

doi: 10.3168/JDS.2008-1101

Altunkaynak BZ, Onger ME, Altunkaynak ME, et al (2011) A Brief Introduction to

Stereology and Sampling Strategies: Basic Concepts of Stereology.

NeuroQuantology. doi: 10.14704/nq.2012.10.1.427

Amano K (1950) A Note on Sodium Phosphate Crystals Found in Mild-cured Salmon.

Nippon SUISAN GAKKAISHI 15:638–641. doi: 10.2331/suisan.15.638

Amrane A, Prigent Y (2008) DIFFUSION OF CALCIUM AND INORGANIC

PHOSPHATE AT THE SURFACE OF A SOLID MODEL MEDIUM IN

RELATION WITH GROWTH OF GEOTRICHUM CANDIDUM AND

PENICILLIUM CAMEMBERTII. J Food Biochem 32:813–825. doi: 10.1111/j.1745-

4514.2008.00200.x

Anbu P, Kang C-H, Shin Y-J, So J-S (2016) Formations of calcium carbonate minerals by

bacteria and its multiple applications. Springerplus 5:250. doi: 10.1186/s40064-016-

1869-2

Arifuzzaman S., Rohani S (2004) Experimental study of brushite precipitation. J Cryst

Growth 267:624–634. doi: 10.1016/J.JCRYSGRO.2004.04.024

Babcock SM, Russell HL (1901) Curing of Cheddar Cheese, with Especial Reference to

Cold-Curing. In: Wisconsin Agricultural Experiment Station Bulletin, 94th edn.

University of Wisconsin, Madison, pp 4–44

Babcock SM, Russell HL, Vivian A, Baer US (1902) Conditions Affecting the

Development of White Specks in Cold-Cured Cheese. In: Wisconsin Agricultural

Experiment Station Bulletin. University of Wisconsin, Madison, pp 180–183

Beevers CA, Hansen HN (1971) The structure of α-lactose monohydrate. Acta Crystallogr

B27:1323–1325. doi: 10.1107/S0567740871003947

112

Ben Omar N, Gonzalez-Muñoz MT, Peñalver JMA (1998) Struvite crystallization on

Myxococcus cells. Chemosphere 36:475–481. doi: 10.1016/S0045-6535(97)10014-5

Beveridge TJ, Murray RG (1976) Uptake and retention of metals by cell walls of Bacillus

subtilis. J Bacteriol 127:1502–18.

Beveridge TJ, Murray RG (1980) Sites of metal deposition in the cell wall of Bacillus

subtilis. J Bacteriol 141:876–87.

Bischoff JL, Fitzpatrick JA, Rosenbauer RJ (1993) The Solubility and Stabilization of

Ikaite (CaCO3•6H2O) from 0° to 25°C: Environmental and Paleoclimatic

Implications for Thinolite Tufa. J Geol 101:21–33. doi: 10.1086/648194

Blaya J, Barzideh Z, LaPointe G (2017) Interaction of starter cultures and nonstarter lactic

acid bacteria in the cheese environment. J Dairy Sci. doi: 10.3168/JDS.2017-13345

Bloss FD (1961) An Introduction to the Methods of Optical Crystallography. Holt,

Rinehard, and Winston Inc., New York, NY

Bloss, F.D. (2012) Crystallography and crystal chemistry. Hold, Rinehart, Winston, Inc.,

New York, NY.

Bockelmann W (2011) Cheese | Smear-Ripened Cheeses. In: Encyclopedia of Dairy

Sciences. pp 753–766

Bokulich NA, Mills DA (2013) Facility-specific &quot;house&quot; microbiome drives

microbial landscapes of artisan cheesemaking plants. Appl Environ Microbiol

79:5214–23. doi: 10.1128/AEM.00934-13

Bonaïti C, Leclercq-Perlat M-N, Latrille E, Corrieu G (2004) Deacidification by

Debaryomyces hansenii of Smear Soft Cheeses Ripened Under Controlled

Conditions: Relative Humidity and Temperature Influences. J Dairy Sci 87:3976–

3988. doi: 10.3168/jds.S0022-0302(04)73538-9

Boutrou R, Gaucheron F, Piot M, et al (1999) Changes in the composition of juice

expressed from Camembert cheese during ripening. Lait 79:503–513. doi:

10.1051/lait:1999541

Bragg WH, Bragg WL (1913) The Reflection of X-rays by Crystals. Proc R Soc A Math

Phys Eng Sci 88:428–438. doi: 10.1098/rspa.1913.0040

Brennan NM, Cogan TM, Loessner M, Scherer S (2004) Bacterial surface-ripened cheeses.

In: Cheese: Chemistry, Physics and Microbiology. pp 199–225

113

Brennan NM, Ward AC, Beresford TP, et al (2002) Biodiversity of the bacterial flora on

the surface of a smear cheese. Appl Environ Microbiol 68:820–30. doi:

10.1128/AEM.68.2.820-830.2002

Brooker BE (1987) The Crystallization of Calcium Phosphate at the Surface of Mould-

Ripened Cheeses. Food Microstruct 6:115–119.

Brooker BE (1990) Food Structure Identification and Characterization of Cocoa Solids and

Milk Proteins in Chocolate Using X-ray Microanalysis. Food Struct 9:9–21.

Brooker BE, Hobbs DG, Turvey A (1975a) Observations on the microscopic crystalline

inclusions in Cheddar cheese. J Dairy Res 42:341. doi: 10.1017/S0022029900015363

Buchardt B, Israelson C, Seaman P, Stockmann G (2001) Ikaite Tufa Towers in Ikka Fjord,

Southwest Greenland: Their Formation by Mixing of Seawater and Alkaline Spring

Water. J Sediment Res 71:176–189. doi: 10.1306/042800710176

Cardello A V., Segars RA (1989) EFFECTS OF SAMPLE SIZE AND PRIOR

MASTICATION ON TEXTURE JUDGMENTS. J Sens Stud 4:1–18. doi:

10.1111/j.1745-459X.1989.tb00454.x

Carlton RA (2011) Polarized Light Microscopy. In: Pharmaceutical Microscopy. Springer,

New York, NY, pp 7–64

Chou Y-E, Edwards CG, Luedecke LO, et al (2003) Nonstarter Lactic Acid Bacteria and

Aging Temperature Affect Calcium Lactate Crystallization in Cheddar Cheese. J

Dairy Sci 86:2516–2524. doi: 10.3168/JDS.S0022-0302(03)73846-6

Clark GL (1929) X-Ray Applications in Every-Day Life. Sci. Mon. 28:172–178.

Clarkson JR, Price TJ, Adams CJ (1992) Role of metastable phases in the spontaneous

precipitation of calcium carbonate. J Chem Soc Faraday Trans 88:243. doi:

10.1039/ft9928800243

Conochie J, Czulak J, Lawrence AJ, Cole WF (1960) Tyrosine and calcium lactate crystals

on rindless cheese. Aust J Dairy Technol 15:120.

Conochie J, Sutherland BJ (1965) The nature and cause of seaminess in Cheddar cheese. J

Dairy Res 32:35. doi: 10.1017/S0022029900018355

Cooke DR, McSweeney PLH (2017) From Micelle to Melt: The Influence of Calcium on

Physico-chemical Properties of Cheese. In: Global Cheesemaking Technology. John

Wiley & Sons, Ltd, Chichester, UK, pp 20–44

Cosmidis J, Benzerara K, Guyot F, et al (2015) Calcium-Phosphate Biomineralization

Induced by Alkaline Phosphatase Activity in Escherichia coli: Localization, Kinetics,

114

and Potential Signatures in the Fossil Record. Front Earth Sci 3:84. doi:

10.3389/feart.2015.00084

Crichton RR (2012) Biomineralisation. In: Biological Inorganic Chemistry. Elsevier, pp

359–378

Cusack M, Freer A (2008) Biomineralization: Elemental and Organic Influence in

Carbonate Systems. Chem Rev 108:4433–4454. doi: 10.1021/cr078270o

Damodaran S, Wang S (2017) Ice crystal growth inhibition by peptides from fish gelatin

hydrolysate. Food Hydrocoll 70:46–56. doi: 10.1016/J.FOODHYD.2017.03.029

Dan H, Hayakawa F, Kohyama K (2008) Modulation of biting procedures induced by the

sensory evaluation of cheese hardness with different definitions. Appetite 50:158–166.

doi: 10.1016/J.APPET.2007.06.009

Daudon, M., Frochot, V., Bazin, D., and Jungers, P. (2016) Crystalluria analysis improves

significantly etiologic diagnosis and therapeutic monitoring of nephrolithiasis. C R

Chim. 19:1514-1526.

de Mendiburu F (2017) agricolae: Statistical Procedures for Agricultural Research.

Diao Y, Myerson AS, Hatton TA, Trout BL (2011) Surface Design for Controlled

Crystallization: The Role of Surface Chemistry and Nanoscale Pores in

Heterogeneous Nucleation. Langmuir 27:5324–5334. doi: 10.1021/la104351k

Do Ngoc M, LeNoir J, Choisy C (1971) Les acides amines libres des fromages affines de

Camembert, Saint-Paulin et Gruyere de Comte. Rev Lait Fr 288:447–462.

Donhowe DP, Hartel RW, Bradley Jr. RL (1991) Determination of Ice Crystal Size

Distributions in Frozen Desserts. J Dairy Sci 74:3334–3344. doi: 10.3168/JDS.S0022-

0302(91)78521-4

Dorn FL, Dahlberg AC (1942) Identification of the White Particles Found on Ripened

Cheddar Cheese. J Dairy Sci 25:31–36. doi: 10.3168/jds.S0022-0302(42)95264-0

Dox AW (1911) THE OCCURRENCE OF TYROSINE CRYSTALS IN ROQUEFORT

CHEESE. J Am Chem Soc 33:423–425. doi: 10.1021/ja02216a024

Dugat-Bony E, Garnier L, Denonfoux J, et al (2016) Highlighting the microbial diversity

of 12 French cheese varieties. Int J Food Microbiol 238:265–273. doi:

10.1016/j.ijfoodmicro.2016.09.026

Dumont JP, Roger S, Adda J (1974) Etude des composés volatils neutres présents dans les

fromages à pâte molle et à croûte lavée. Lait 54:31–43. doi: 10.1051/lait:1974531-

5323

115

Dupraz C, Reid RP, Braissant O, et al (2009) Processes of carbonate precipitation in

modern microbial mats. Earth-Science Rev 96:141–162. doi:

10.1016/J.EARSCIREV.2008.10.005

Dyar, M.D., Gunter, M.E., and Tasa, D. (2008) Mineralogy and Optical

Mineralogy.Mineralogical Society of America, Chantilly, VA.

Eddelbuettel, Dirk. (2017) random: True Random Numbers using RANDOM.ORG. R

package version 0.2.6. https://CRAN.R-project.org/package=random.

Ellis, S.E. (1962) Phosphate minerals in a calculus from the reticule-tureen of the sheep.

Mineral Mag. 33:600-608.

Engelen L, de Wijk RA, van der Bilt A, et al (2005) Relating particles and texture

perception. Physiol Behav 86:111–117. doi: 10.1016/j.physbeh.2005.06.022

Erdemir D, Lee AY, Myerson AS (2009) Nucleation of Crystals from Solution: Classical

and Two-Step Models. Acc Chem Res 42:621–629. doi: 10.1021/ar800217x

Ferris FG, Beveridge TJ (1986) Physicochemical roles of soluble metal cations in the outer

membrane of Escherichia coli K-12. Can J Microbiol 32:594–601. doi: 10.1139/m86-

110

Feurer C, Vallaeys T, Corrieu G, Irlinger F (2004) Does Smearing Inoculum Reflect the

Bacterial Composition of the Smear at the End of the Ripening of a French Soft, Red-

Smear Cheese? J Dairy Sci 87:3189–3197. doi: 10.3168/JDS.S0022-0302(04)73454-

2

Fidaleo M, Miele Nicoletta A, Mainardi S, et al (2017) Effect of refining degree on particle

size, sensory and rheological characteristics of anhydrous paste for ice creams

produced in industrial stirred ball mill. LWT - Food Sci Technol. doi:

10.1016/j.lwt.2017.01.046

Gaucheron F (2005) The minerals of milk. Reprod Nutr Dev 45:473–483. doi:

10.1051/rnd:2005030

Giannimaras EK, Koutsoukos PG (1987) The crystallization of calcite in the presence of

orthophosphate. J Colloid Interface Sci 116:423–430. doi: 10.1016/0021-

9797(87)90138-X

Ginibre C., Worner G., Kronz A. (2007) Crystal zoning as an archive for magma evolution.

Elements. 3:261-266.

Goff, H.D., Hartel, R.W. (2013) Ice Cream. 7th ed. Springer, New York, NY.

116

Gori K, Mortensen HD, Arneborg N, Jespersen L (2007) Ammonia Production and Its

Possible Role as a Mediator of Communication for Debaryomyces hansenii and Other

Cheese-Relevant Yeast Species. J Dairy Sci 90:5032–5041. doi: 10.3168/jds.2006-

750

Gorski CA, Fantle MS (2017) Stable mineral recrystallization in low temperature aqueous

systems: A critical review. Geochim Cosmochim Acta 198:439–465. doi:

10.1016/J.GCA.2016.11.013

Griffiths MW, Tellez AM (2013) Lactobacillus helveticus: the proteolytic system. Front

Microbiol 4:30. doi: 10.3389/fmicb.2013.00030

Guinard J-X, Mazzucchelli R (1996) The sensory perception of texture and mouthfeel.

Trends Food Sci Technol 7:213–219. doi: 10.1016/0924-2244(96)10025-X

Guo MR, Kindstedt PS (1995) Age-Related Changes in the Water Phase of Mozzarella

Cheese. J Dairy Sci 78:2099–2107. doi: 10.3168/JDS.S0022-0302(95)76836-9

Hardy, R.H. & Nation, B. (1984) Acute gout and the accident and emergency department.

Arch Emerg Med, 1:89–95.

Harper WJ, Swanson AM, Sommer HH (1953) Observations on the Chemical Composition

of White Particles in Several Lots of Cheddar Cheese. J Dairy Sci 36:368–372. doi:

10.3168/jds.S0022-0302(53)91505-7

Hartel RW (2001) Crystallization in Foods. Aspen Publishers, Inc., Gaithersburg, MD

Hartel RW (2013) Advances in Food Crystallization. Annu Rev Food Sci Technol 4:277–

92. doi: 10.1146/annurev-food-030212-182530

Hassan A, Johnson ME, Lucey JA (2004) Changes in the Proportions of Soluble and

Insoluble Calcium During the Ripening of Cheddar Cheese. J Dairy Sci 87:854–862.

doi: 10.3168/JDS.S0022-0302(04)73229-4

Hélias A, Mirade P-S, Corrieu G, et al (2007) Modeling of Camembert-type cheese mass

loss in a ripening chamber: main biological and physical phenomena. J Dairy Sci

90:5324–33. doi: 10.3168/jds.2007-0272

Hesse, K.F., Küppers, H., Suess, E. (1983) Refinement of the structure of Ikaite,

CaCO3·6H2O. Z Kristallogr Cryst Mater. 163:227-231.

Hitchcock DI (1924) THE SOLUBILITY OF TYROSINE IN ACID AND IN ALKALI. J

Gen Physiol 6:747–57.

117

Hohenegger M, Guéguen M, Beduhn R, et al (2014) Biodiversity of the Surface Microbial

Consortia from Limburger, Reblochon, Livarot, Tilsit, and Gubbeen Cheeses.

Microbiol Spectr 2:219–250. doi: 10.1128/microbiolspec.CM-0010-2012

Hough G, Martinez E, Contarini A (1990) Sensory and Objective Measurement of

Sandiness in Dulce de Leche, a Typical Argentine Dairy Product. J Dairy Sci 73:604–

611. doi: 10.3168/jds.S0022-0302(90)78709-7

Hu Y-B, Wolf-Gladrow DA, Dieckmann GS, et al (2014) A laboratory study of ikaite

(CaCO3·6H2O) precipitation as a function of pH, salinity, temperature and phosphate

concentration. Mar Chem 162:10–18. doi: 10.1016/J.MARCHEM.2014.02.003

Hudson Institute of Mineralogy (2018) Mineral information and data.

https://www.mindat.org/. Accessed 17 Feb 2018

Hunziker OF (1920) The Butter Industry: Prepared for the use of creameries, dairy students

and pure food departments, by Otto F. Hunziker ... Self Published, La Grange, Ill.

Imai E, Hatae K, Shimada A (1995) Oral perception of grittiness: effect of particle size and

concentration of the dispersed particles and the dispersion medium. J Texture Stud

26:561–576. doi: 10.1111/j.1745-4603.1995.tb00804.x

Irlinger F, Layec S, Hélinck S, Dugat-Bony E (2015) Cheese rind microbial communities:

diversity, composition and origin. FEMS Microbiol Lett 362:1–11. doi:

10.1093/femsle/fnu015

Jacquet J, Saingt O (1952) Étude sur des fabrications accidentelles de Camemberts Rouges.

Lait 32:129–145. doi: 10.1051/lait:1952313-3145

Jelen P (1992) Whey Cheeses and Beverages. In: Whey and Lactose Processing. Springer

Netherlands, Dordrecht, pp 157–193

Johnson ME (2014) Mesophilic and Thermophilic Cultures Used in Traditional

Cheesemaking. In: Cheese and Microbes. American Society of Microbiology, pp 73–

94

Johnson ME, Riesterer BA, Chen C, et al (1990) Effect of Packaging and Storage

Conditions on Calcium Lactate Crystallization on the Surface of Cheddar Cheese. J

Dairy Sci 73:3033–3041. doi: 10.3168/jds.S0022-0302(90)78990-4

Jokanovic V, Colovic B, Zivkovic-Sandic M, et al (2014) Calcium phosphate materials in

bone tissue engineering. Stomatol Glas Srb 61:93–101. doi: 10.2298/SGS1402093J

Kalab M, Allan-Wojtas P, Miller SS (1995) Microscopy and other imaging techniques in

food structure analysis. Trends Food Sci Technol 6:177–186. doi: 10.1016/S0924-

2244(00)89052-4

118

Karahadian C, Lindsay RC, Ades GL, et al (1987) Integrated Roles of Lactate, Ammonia,

and Calcium in Texture Development of Mold Surface-Ripened Cheese. J Dairy Sci

70:909–918. doi: 10.3168/jds.S0022-0302(87)80094-2

Karahadian C, Lindsay RC, Ades GL, et al (1987) Integrated Roles of Lactate, Ammonia,

and Calcium in Texture Development of Mold Surface-Ripened Cheese. J Dairy Sci

70:909–918. doi: 10.3168/jds.S0022-0302(87)80094-2

Keller WD, Reid WHE, Arbuckle WS, Decker CW (1943) Microscopy of Ice Cream with

Polarized Light. J Dairy Sci 26:683–688. doi: 10.3168/JDS.S0022-0302(43)92769-9

Kim D, Min KJ, Lee K, et al (2016) Effects of pH, molar ratios and pre-treatment on

phosphorus recovery through struvite crystallization from effluent of anaerobically

digested swine wastewater. Environ Eng Res 22:12–18. doi: 10.4491/eer.2016.037

Kindstedt PS, Kosikowski FV (1988) Calcium, Phosphorus, and Sodium Concentrations

in Cheddar Cheese. J Dairy Sci 71:285–289. doi: 10.3168/JDS.S0022-

0302(88)79556-9

Klapper, H. (2010) Generation and propagation of defects during crystal growth. In

Springer Handbook of Crystal Growth. Springer, Berlin, Germany.

Klug, H. P., & Alexander, L. E. (1954) X-ray diffraction procedures: for polycrystalline

and amorphous materials. Vol. 2. Wiley, New York, NY.

Konhauser KO, Lalonde S V., Phoenix VR (2008) Bacterial biomineralization: Where to

from here? Geobiology 6:298–302. doi: 10.1111/j.1472-4669.2008.00151.x

Lawless HT, Heymann H (2010) Texture Evaluation. In: Sensory Evaluation of Food:

Principles and Practices, 2nd edn. Springer, New York, pp 259–281

Lawrence G, Salles C, Palicki O, et al (2011) Using cross-modal interactions to

counterbalance salt reduction in solid foods. Int Dairy J 21:103–110. doi:

10.1016/J.IDAIRYJ.2010.09.005

Lawrence G, Salles C, Septier C, et al (2009) Odour–taste interactions: A way to enhance

saltiness in low-salt content solutions. Food Qual Prefer 20:241–248. doi:

10.1016/J.FOODQUAL.2008.10.004s

Laxa O (1926) L’examen microscopique des fromages. Lait 6:885–903. doi:

10.1051/lait:19266033

Laxa O (1927) Sur la nature des grains microscopiques dans les fromages. Lait 7:521–525.

doi: 10.1051/lait:19276626

119

Le Graet Y, Brule G (1988) Migration des macro et oligo-éléments dans un fromage à pâte

molle de type Camembert. Lait 68:219–234. doi: 10.1051/lait:1988215

Le Graet Y, Lepienne A, Brule G, Ducruet P (1983) Migration du calcium et des

phosphates inorganiques dans les fromages à pâte molle de type Camembert au cours

de l’affinage. Lait 63:317–332. doi: 10.1051/lait:1983629-63019

Le Graet, Y. & G. Brule. (1988) Migration des macro et oligo-éléments dans un fromage

à pâte molle de type Camembert. Lait 68:219-234.

Le Graet, Y., A. Lepienne, G. Brule, and P. Ducruet. (1983) Migration du calcium et des

phosphates inorganiques dans les fromages à pâte molle de type Camembert au cours

de l'affinage. Lait. 63:317-332.

Le Mens P, Parguel P, Tormo H, et al (2000) Description sensorielle de trois fromages

fermiers : le Banon, la Brique du Forez et le Rogeret. Renc Rech Ruminants 7:300–

303.

Lê S, Josse J, Husson F (2008) FactoMineR: An R Package for Multivariate Analysis. J

Stat Softw 25:1–18. doi: 10.18637/jss.v025.i01

Li CH, Lee CK (1993) Minimum cross entropy thresholding. Pattern Recognit 26:617–

625. doi: 10.1016/0031-3203(93)90115-D

Lopez FL, Bowles A, Gul MO, et al (2016) Effect of formulation variables on oral

grittiness and preferences of multiparticulate formulations in adult volunteers. Eur J

Pharm Sci 92:156–162. doi: 10.1016/J.EJPS.2016.07.006

Lucey JA, Fox PF (1993) Importance of Calcium and Phosphate in Cheese Manufacture:

A Review. J Dairy Sci 76:1714–1724. doi: 10.3168/jds.S0022-0302(93)77504-9

Lucey JA, Horne DS (2009) Milk Salts: Technological Significance. In: Advanced Dairy

Chemistry. Springer New York, New York, NY, pp 351–389

Luft JR, Wolfley JR, Snell EH (2011) What’s in a drop? Correlating observations and

outcomes to guide macromolecular crystallization experiments. Cryst Growth Des

11:651–663. doi: 10.1021/cg1013945

Luo J, Chen X, Crump J, et al (2018) Interactions of fungi with concrete: Significant

importance for bio-based self-healing concrete. Constr Build Mater 164:275–285. doi:

10.1016/J.CONBUILDMAT.2017.12.233

Maher PG, Auty MAE, Roos YH, et al (2015) Microstructure and lactose crystallization

properties in spray dried nanoemulsions. Food Struct 3:1–11. doi:

10.1016/J.FOOSTR.2014.10.001

120

Marland G (1975) The stability of CaCO3·6H2O (ikaite). Geochim Cosmochim Acta

39:83–91. doi: 10.1016/0016-7037(75)90186-6

Martini S, Suzuki AH, Hartel RW (2008) Effect of High Intensity Ultrasound on

Crystallization Behavior of Anhydrous Milk Fat. J Am Oil Chem Soc 85:621–628.

doi: 10.1007/s11746-008-1247-5

McDowall FH, McDowell AKR (1939) The white particles in mature Cheddar cheese. J

Dairy Res 10:118. doi: 10.1017/S0022029900002788

McSweeney PLH (2004) Biochemistry of cheese ripening. Int J Dairy Technol 57:127–

144. doi: 10.1111/j.1471-0307.2004.00147.x

McSweeney PLH (2007) Cheese problems solved. CRC Press, New York, NY.

McSweeney, P. (2004) Biochemistry of cheese ripening. Int J Dairy Technol. 57:127-144.

Meilgaard M, Civille GV, Carr BT (2007) Sensory evaluation techniques. Taylor &

Francis

Metche M, Fanni J (1978) Rôle de la flore fongique dans l’accumulation du calcium et du

phosphore à la surface des fromages du type camembert. Lait 58:336–354. doi:

10.1051/lait:197857721

Meyer D, Zeileis A, Hornik K (2017) vcd: Visualizing Categorical Data.

Monrad JH (1910) Cheese Defects. In: New York Prod. Rev. Am. Cream.

https://books.google.com/books?id=6epMAQAAMAAJ&pg=PA151&lpg=PA151&

dq=What%27s+in+Those+“White+Specks”+on+Cheddar?.+Wis.+Agr.+Expt.+Sta&

source=bl&ots=m7xfOm9loA&sig=Tnv3CFTuDpFzKGN48WdEW9WAa2U&hl=e

n&sa=X&ved=0ahUKEwifpZjos7DZAhXRg-AKHUJxAPs. Accessed 18 Feb 2018

Morris HA, Holt C, Brooker BE, et al (1988) Inorganic constituents of cheese: analysis of

juice from a one-month-old Cheddar cheese and the use of light and electron

microscopy to characterize the crystalline phases. J Dairy Res 55:255. doi:

10.1017/S002202990002608X

Mullan WMA (2000) Causes and control of early gas production in cheddar cheese. Int J

Dairy Technol 53:63–68. doi: 10.1111/j.1471-0307.2000.tb02660.x

Mullin JW (2001) Crystallization, 4th edn. Butterworth-Heinemann, Oxford

Naka K, Chujo Y (2001) Control of Crystal Nucleation and Growth of Calcium Carbonate

by Synthetic Substrates. Chem Mater 13:3245–3259. doi: 10.1021/CM011035G

121

Nakajima H, Kunji ER., Poolman B, Konings WN (1998) Amino acid transport in

Lactobacillus helveticus. FEMS Microbiol Lett 158:249–253. doi: 10.1111/j.1574-

6968.1998.tb12828.x

Nesse, W.D. (2004). Introduction to Optical Minerology. 3rd ed. Oxford University Press,

New York, NY.

Nickerson TA (1954) Lactose Crystallization in Ice Cream. I. Control of Crystal Size by

Seeding. J Dairy Sci 37:1099–1105. doi: 10.3168/JDS.S0022-0302(54)91373-9

Noël Y, Zannoni M, Hunter EA (1996) Texture of Parmigiano Reggiano cheese: Statistical

relationships between rheological and sensory variates. Lait 76:243–254. doi:

10.1051/lait:1996320

Phadungath C, Metzger LE (2011) Effect of sodium gluconate on the solubility of calcium

lactate. J Dairy Sci 94:4843–9. doi: 10.3168/jds.2011-4549

Picque D, Leclercq-Perlat M-N, Corrieu G (2006) Effects of Atmospheric Composition on

Respiratory Behavior, Weight Loss, and Appearance of Camembert-Type Cheeses

During Chamber Ripening. J Dairy Sci 89:3250–3259. doi: 10.3168/jds.S0022-

0302(06)72600-5

Plummer LN, Busenberg E (1982) The solubilities of calcite, aragonite and vaterite in

CO2-H2O solutions between 0 and 90°C, and an evaluation of the aqueous model for

the system CaCO3-CO2-H2O. Geochim Cosmochim Acta 46:1011–1040. doi:

10.1016/0016-7037(82)90056-4

Polowsky PJ, Tansman GF, Kindstedt PS, Hughes JM (In Preparation) Characterization

and presumptive identification of Surface Crystals on Smear Ripened Cheese by

Polarized Light Microscopy. Manuscript in Preparation.

Prywer J, Torzewska A (2010) Biomineralization of struvite crystals by Proteus mirabilis

from artificial urine and their mesoscopic structure. Cryst Res Technol 45:1283–1289.

doi: 10.1002/crat.201000344

Quigley L, O’Sullivan O, Beresford TP, et al (2012) High-throughput sequencing for

detection of subpopulations of bacteria not previously associated with artisanal

cheeses. Appl Environ Microbiol 78:5717–23. doi: 10.1128/AEM.00918-12

R Core Team (2016) R: A language and environment for statistical computing.

Riahi MH, Trelea IC, Leclercq-Perlat M-N, et al (2007) Model for changes in weight and

dry matter during the ripening of a smear soft cheese under controlled temperature and

relative humidity. Int Dairy J 17:946–953. doi: 10.1016/J.IDAIRYJ.2006.11.002

122

Ribadeau-Dumas B (1984) Maîtrise de l’affinage des fromages de type Camembert. Lait

64:448–468. doi: 10.1051/lait:1984643-64435

Rivadeneyra MA, Delgado R, Párraga J, et al (2006) Precipitation of minerals by 22 species

of moderately halophilic bacteria in artificial marine salts media: Influence of salt

concentration. Folia Microbiol (Praha) 51:445–453. doi: 10.1007/BF02931589

Rysgaard, S., Glud, R. N., Lennert, K., Cooper, M., Halden, N., Leakey, R.J.G.,

Hawthorne, F. C., and Barber, D. (2012) Ikaite crystals in melting sea ice –

implications for pCO2 and pH levels in Arctic surface waters. Cryosphere. 6:901–908.

S-A Johnsson M, Nancollas GH (1992) The Role of Brushite and Octacalcium Phosphate

in Apatite Formation. Crit Rev Oral Biol Med 3:61–82.

Sainani MR, Vyas HK, Tong PS (2004) Characterization of Particles in Cream Cheese. J

Dairy Sci 87:2854–2863. doi: 10.3168/jds.S0022-0302(04)73414-1

Sanchez-Roman M, Rivadeneyra MA, Vasconcelos C, et al (2007) Biomineralization of

carbonate and phosphate by moderately halophilic bacteria. FEMS Microbiol Ecol

61:273–284. doi: 10.1111/j.1574-6941.2007.00336.x

Scharpf LG, Michnick MJ (1967) In Situ Identification of Crystals on Process Cheese by

X-ray Diffraction. J Dairy Sci 50:1989–1991. doi: 10.3168/JDS.S0022-

0302(67)87759-2

Schindelin J, Arganda-Carreras I, Frise E, et al (2012) Fiji: an open-source platform for

biological-image analysis. Nat Methods 9:676–682. doi: 10.1038/nmeth.2019

Severn DJ, Johnson ME, Olson NF (1986) Determination of Lactic Acid in Cheddar

Cheese and Calcium Lactate Crystals. J Dairy Sci 69:2027–2030. doi:

10.3168/jds.S0022-0302(86)80632-4

Shock AA, Harper WJ, Swanson AM, Sommer HH (1941) What’s in those “white specks”

on Cheddar? In: Wisconsin Agricultural Experiment Station Bulletin, 474th edn.

University of Wisconsin, Madison, WI,

Skeie S, Abrahamsen RK (2017) Brown Whey Cheese. In: Cheese. Elsevier, pp 1117–

1132

Sommer HH (1930) A Case of Sandiness in Processed Cheese. J Dairy Sci 13:288–290.

doi: 10.3168/jds.S0022-0302(30)93526-6

Spinnler H-E, Gripon J-C (2004) Surface mould-ripened cheeses. Cheese Chem Phys

Microbiol 2:157–174. doi: 10.1016/S1874-558X(04)80043-5

123

Stocks-Fischer S, Galinat JK, Bang SS (1999) Microbiological precipitation of CaCO3.

Soil Biol Biochem 31:1563–1571. doi: 10.1016/S0038-0717(99)00082-6

Stoiber RE, Morse SA (1971) Microscopic Identification of Crystals. Ronald Press

Company, New York, NY

Stokes JR, Boehm MW, Baier SK (2013) Oral processing, texture and mouthfeel: From

rheology to tribology and beyond. Curr Opin Colloid Interface Sci 18:349–359. doi:

10.1016/J.COCIS.2013.04.010

Swiatek A, Jaworski J (1959) Histochemical studies on distribution of mineral matter in

cheeses. 15th Int Dairy Congr Proc 3:1509–1518.

Tansman GF, Kindstedt PS, Hughes JM (2015a) Crystal fingerprinting: elucidating the

crystals of Cheddar, Parmigiano-Reggiano, Gouda, and soft washed-rind cheeses

using powder x-ray diffractometry. Dairy Sci Technol 95:651–664. doi:

10.1007/s13594-015-0225-6

Tansman GF, Kindstedt PS, Hughes JM (2015b) Characterization of single crystals in the

rinds of white mold and smear ripened cheeses with single crystal X-ray

diffractometry. J Dairy Res 98:668–669.

Tansman GF, Kindstedt PS, Hughes JM (2017b) Minerals In Food: Crystal Structures of

Ikaite and Struvite From Bacterial Smears On Washed-Rind Cheese. Can Mineral

55:89–100. doi: 10.3749/canmin.1600038

Tansman GF, Kindstedt PS, Hughes JM (2017c) Crystallization and demineralization

phenomena in stabilized white mold cheese. J Dairy Sci 100:6074–6083. doi:

10.3168/jds.2016-12259

Tansman GF, Kindstedt PS, Hughes JM (2017d) Crystallization and demineralization

phenomena in washed-rind cheese. J Dairy Sci. doi: 10.3168/jds.2017-13067

Tansman GF, Kindstedt PS, Hughes JM (2017c) Crystallization and demineralization

phenomena in washed-rind cheese. J Dairy Sci. doi: 10.3168/jds.2017-13067

Tian F, Qu H, Zimmermann A, et al (2010) Factors affecting crystallization of hydrates. J

Pharm Pharmacol 62:1534–1546. doi: 10.1111/j.2042-7158.2010.01186.x

Torri L, Piochi M, Marchiani R, et al (2016) A sensory- and consumer-based approach to

optimize cheese enrichment with grape skin powders. J Dairy Sci 99:194–204. doi:

10.3168/jds.2015-9922

Tuckey SL, Ruehe HA, Clark GL (1934) The Application of X-rays to Research in Dairy

Technology. J Dairy Sci 17:587–594. doi: 10.3168/JDS.S0022-0302(34)93277-X

124

Tuckey SL, Ruehe HA, Clark GL (1938a) An X-Ray Diffraction Analysis of Casein. J

Dairy Sci 21:767–775. doi: 10.3168/JDS.S0022-0302(38)95644-9

Tuckey SL, Ruehe HA, Clark GL (1938b) An X-Ray Diffraction Analysis of Cheddar

Cheese. J Dairy Sci 21:777–789. doi: 10.3168/JDS.S0022-0302(38)95645-0

Tuckey SL, Ruehe HA, Clark GL (1938c) X-ray diffraction analysis of white specks in

Cheddar cheese. J Dairy Sci 21:161.

Tyle P (1993) Effect of size, shape and hardness of particles in suspension on oral texture

and palatability. Acta Psychol (Amst) 84:111–118. doi: 10.1016/0001-

6918(93)90077-5

Vaisey-Genser, M., Vane, B.K., and Johnson, S. (1989) Graininess, crystal size, and

firmness of stored canola oil margarines. J Texture Stud. 20:347-361.

Van Slyke LL, Publow CA (1909) The Science and Practice of Cheese Making. Orange

Judd Company, New York

Vassal L, Monnet V, Le Bars D, et al (1986) Relation entre le pH, la composition chimique

et la texture des fromages de type Camembert. Lait 66:341–351. doi:

10.1051/lait:1986422

Wherry ET (1928) The crystallography and optical properties of β-lactose. J Washingt

Acad Sci 18:302–304. doi: 10.2307/24522543

Wilsenach JA, Schuurbiers CAH, van Loosdrecht MCM (2007) Phosphate and potassium

recovery from source separated urine through struvite precipitation. Water Res

41:458–466. doi: 10.1016/J.WATRES.2006.10.014

Wolfe BE, Button JE, Santarelli M, Dutton RJ (2014) Cheese Rind Communities Provide

Tractable Systems for In Situ and In Vitro Studies of Microbial Diversity. Cell

158:422–433. doi: 10.1016/j.cell.2014.05.041

Wu Q, Bishop PL, Keener TC (2005) A Strategy for Controlling Deposition of Struvite in

Municipal Wastewater Treatment Plants. Water Environ Res 77:199–207. doi:

10.2175/106143005X41771

Zamberlin Š, Antunac N, Havranek J, Samaržija D (2012) Mineral elements in milk and

dairy products. Mljekarstvo 62:111–125.

Zannoni M, Bertozzi L, Hunter EA (1994) Comparison of Parmigiano-Reggiano and

American Parmesan cheeses by sensory analysis of texture. Sci e Tec Latt Casearia

45:505–518.

125

Zheng L, Sun D-W (2006) Innovative applications of power ultrasound during food

freezing processes—a review. Trends Food Sci Technol 17:16–23. doi:

10.1016/J.TIFS.2005.08.010

Zhu Y, Ma N, Jin W, et al (2017) Genomic and Transcriptomic Insights into Calcium

Carbonate Biomineralization by Marine Actinobacterium Brevibacterium linens

BS258. Front Microbiol 8:602. doi: 10.3389/fmicb.2017.00602

Zisu B, Sciberras M, Jayasena V, et al (2014) Sonocrystallisation of lactose in concentrated

whey. Ultrason Sonochem 21:2117–2121. doi: 10.1016/J.ULTSONCH.2014.03.031

126

APPENDIX A: SUPPLEMENTARY DIFFRACTORGRAMS

Cheese Diffractogram

A

B

10 20 30 40 50

0

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

2400

2600

2800

3000

0

50

100Struvite, syn

10 20 30 40 50 0

50

100Ikaite

2-theta (deg)

Meas. data:A1/Data 1

Struvite, syn

Ikaite

Inte

nsity (

cps)

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

10 20 30 40 50 0

20

40

60

80

100Struvite, syn

2-theta (deg)

Meas. data:B2/Data 1

Struvite, syn

Inte

nsity (

cp

s)

127

C

D

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

500

550

0

50

100Struvite, syn

0

50

100Ikaite

0

50

100Calcite, syn

10 20 30 40 50 0

50

100Brushite, syn

2-theta (deg)

Meas. data:C2/Data 1

Struvite, syn

Ikaite

Calcite, syn

Brushite, syn

Inte

nsity (

cps)

10 20 30 40 50

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

0

50

100Struvite, syn

0

50

100Ikaite

10 20 30 40 50 0

50

100Brushite, syn

2-theta (deg)

Meas. data:D2/Data 1

Struvite, syn

Ikaite

Brushite, syn

Inte

nsity (

cps)

128

E

F

10 20 30 40 50

0

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

2400

0

20

40

60

80

100Ikaite

10 20 30 40 50 0

20

40

60

80

100Struvite, syn

2-theta (deg)

Meas. data:E1/Data 1

Ikaite

Struvite, syn

Inte

nsity (

cps)

10 20 30 40 50

0

1000

2000

3000

4000

5000

6000

7000

8000

0

50

100Ikaite

0

50

100Struvite, syn

10 20 30 40 50 0

50

100Brushite, syn

2-theta (deg)

Meas. data:F2/Data 1

Ikaite

Struvite, syn

Brushite, syn

Inte

nsity (

cps)

129

G

H

10 20 30 40 50

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

22000

24000

0

50

100Ikaite

0

50

100Struvite, syn

10 20 30 40 50 0

50

100Brushite, syn

2-theta (deg)

Meas. data:G1/Data 1

Ikaite

Struvite, syn

Brushite, syn

Inte

nsity (

cps)

10 20 30 40 50

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

6500

7000

7500

8000

8500

0

50

100Struvite, syn

10 20 30 40 50 0

50

100Ikaite

2-theta (deg)

Meas. data:H2/Data 1

Struvite, syn

Ikaite

Inte

nsity (

cps)

130

I

J

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

750

800

10 20 30 40 50 0

20

40

60

80

100Struvite, syn

2-theta (deg)

Meas. data:I2/Data 1

Struvite, syn

Inte

nsity (

cps)

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

0

50

100Struvite, syn

0

50

100Ikaite

10 20 30 40 50 0

50

100Brushite, syn

2-theta (deg)

Meas. data:J2/Data 1

Struvite, syn

Ikaite

Brushite, syn

Inte

nsity (

cps)

131

K

L

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

10 20 30 40 50 0

20

40

60

80

100Struvite, syn

2-theta (deg)

Meas. data:K2/Data 1

Struvite, syn

Inte

nsity (

cps)

10 20 30 40 50

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0

20

40

60

80

100Struvite, syn

10 20 30 40 50 0

20

40

60

80

100Ikaite

2-theta (deg)

Meas. data:L1/Data 1

Struvite, syn

Ikaite

Inte

nsity (

cps)

132

M

N

10 20 30 40 50

0

100

200

300

400

500

600

700

800

900

0

50

100Struvite, syn

0

50

100Ikaite

10 20 30 40 50 0

50

100Calcite, syn

2-theta (deg)

Meas. data:M1/Data 1

Struvite, syn

Ikaite

Calcite, syn

Inte

nsity (

cps)

10 20 30 40 50

0

50

100

150

200

250

300

350

0

50

100Struvite, syn

0

50

100Ikaite

10 20 30 40 50 0

50

100L-tyrosine

2-theta (deg)

Meas. data:N1/Data 1

Struvite, syn

Ikaite

L-tyrosine

Inte

nsity (

cps)

133

O

P

10 20 30 40 50

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0

50

100Calcite, syn

0

50

100Struvite, syn

10 20 30 40 50 0

50

100Ikaite

2-theta (deg)

Meas. data:O1/Data 1

Calcite, syn

Struvite, syn

Ikaite

Inte

nsity (

cps)

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

500

550

600

10 20 30 40 50 0

20

40

60

80

100

2-theta (deg)

Meas. data:P1/Data 1

Inte

nsity (

cps)

134

Q

R

10 20 30 40 50

0

50

100

150

200

250

300

350

400

0

20

40

60

80

100Ikaite

10 20 30 40 50 0

20

40

60

80

100Struvite, syn

2-theta (deg)

Meas. data:Q1/Data 1

Ikaite

Struvite, syn

Inte

nsity (

cps)

10 20 30 40 50

0

100

200

300

400

500

600

700

800

900

0

50

100Ikaite

10 20 30 40 50 0

50

100Struvite, syn

2-theta (deg)

Meas. data:S1/Data 1

Ikaite

Struvite, syn

Inte

nsity (

cps)

135

S

T

10 20 30 40 50

0

50

100

150

200

250

300

350

400

10 20 30 40 50 0

20

40

60

80

100Ikaite

2-theta (deg)

Meas. data:S2/Data 1

Ikaite

Inte

nsity (

cp

s)

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

500

550

600

10 20 30 40 50 0

20

40

60

80

100Struvite, syn

2-theta (deg)

Meas. data:T2/Data 1

Struvite, syn

Inte

nsity (

cps)

136

U

V

10 20 30 40 50

0

50

100

150

200

250

300

350

400

0

50

100Calcite, syn

10 20 30 40 50 0

50

100Struvite, syn

2-theta (deg)

Meas. data:U1/Data 1

Calcite, syn

Struvite, syn

Inte

nsity (

cps)

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

750

800

10 20 30 40 50 0

20

40

60

80

100Struvite, syn

2-theta (deg)

Meas. data:V1/Data 1

Struvite, syn

Inte

nsity (

cps)

137

W

X

10 20 30 40 50

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

6500

7000

7500

8000

0

50

100Struvite, syn

10 20 30 40 50 0

50

100Ikaite

2-theta (deg)

Meas. data:W2/Data 1

Struvite, syn

Ikaite

Inte

nsity (

cps)

10 20 30 40 50

0

50

100

150

200

250

300

350

400

450

0

50

100Struvite, syn

10 20 30 40 50 0

50

100Brushite, syn

2-theta (deg)

Meas. data:X1/Data 1

Struvite, syn

Brushite, syn

Inte

nsity (

cps)

138

APPENDIX B: SUPPLEMENTARY PCA FIGURES

Figure 14. Contribution of variables to PC 1. Red line indicates 10% contribution.

Figure 15. Contribution of variables to PC 2. Red line indicates 10% contribution.

139

Figure 16. Contribution of variables to PC 3. Red line indicates 10% contribution.

Figure 17. Contribution of variables to PC 4. Red line indicates 10% contribution.