Journal of Structural Biology - MITweb.mit.edu/cortiz/www/ConnorsM_JStrBio2012.pdf · 2012-02-15 · bInstitute of Bioanalytical Chemistry, Dresden University of Technology, 01069

Journal of Structural Biology 177 (2012) 314–328

Contents lists available at SciVerse ScienceDirect

Journal of Structural Biology

journal homepage: www.elsevier .com/ locate/y jsbi

Three-dimensional structure of the shell plate assembly of the chitonTonicella marmorea and its biomechanical consequences

Matthew J. Connors a, Hermann Ehrlich b, Martin Hog b, Clemence Godeffroy b, Sergio Araya c, Ilan Kallai d,Dan Gazit d,e, Mary Boyce f, Christine Ortiz a,⇑a Department of Materials Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USAb Institute of Bioanalytical Chemistry, Dresden University of Technology, 01069 Dresden, Germanyc Department of Architecture, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USAd Skeletal Biotech Laboratory, The Hebrew University – Hadassah Faculty of Dental Medicine, Ein Kerem, Jerusalem 91120, Israele Department of Surgery and Cedars-Sinai Regenerative Medicine Institute (CS-RMI), Cedars-Sinai Medical Center, Los Angeles, CA 90048, USAf Department of Mechanical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA

a r t i c l e i n f o

Article history:Received 2 September 2011Received in revised form 12 December 2011Accepted 14 December 2011Available online 10 January 2012

Keywords:ChitonShell microstructureExoskeletonNatural armorMicro-computed tomography

1047-8477/$ - see front matter � 2012 Elsevier Inc. Adoi:10.1016/j.jsb.2011.12.019

⇑ Corresponding author. Department of MateriaMassachusetts Institute of Technology, 77 MassachusCambridge, MA 02139, USA. Fax: +1 617 253 5620.

E-mail address: [email protected] (C. Ortiz).

a b s t r a c t

This study investigates the three-dimensional structure of the eight plate exoskeletal (shell) assembly ofthe chiton Tonicella marmorea. X-ray micro-computed tomography and 3D printing elucidate the mech-anism of conformational change from a passive (slightly curved, attached to surface) to a defensive(rolled, detached from surface) state of the plate assembly. The passive and defensive conformationsexhibited differences in longitudinal curvature index (0.43 vs. 0.70), average plate-to-plate overlap(�62% vs. �48%), cross-sectional overlap heterogeneity (60–82.5% vs. 0–90%, fourth plate), and plate-to-plate separation distance (100% increase in normalized separation distance between plates 4 and 5),respectively. The plate-to-plate interconnections consist of two rigid plates joined by a compliant, actu-ating muscle, analogous to a geometrically structured shear lap joint. This work provides an understand-ing of how T. marmorea achieves the balance between mobility and protection. In the passive state, themorphometry of the plates and plate-to-plate interconnections results in an approximately continuouscurvature and constant armor thickness, resulting in limited mobility but maximum protection. In thedefensive state, the underlying soft tissues gain protection and the chiton gains mobility through tidalflow, but regions of vulnerability open dorsally, due to the increase in plate-to-plate separation anddecrease in plate-to-plate overlap. Lastly, experiments using optical and scanning electron microscopy,mercury porosimetry, and Fourier-transform infrared spectroscopy explore the microstructure and spa-tial distribution of the six layers within the intermediate plates, the role of multilayering in resistingpredatory attacks, and the detection of chitin as a major component of the intra-plate organic matrixand girdle.

� 2012 Elsevier Inc. All rights reserved.

1. Introduction

Chitons (Polyplacophorans) are marine mollusks foundworldwide in all seas with more than 940 extant (Schwabe,2005; Stebbins and Eernisse, 2009) and 430 fossil species (Puchal-ski et al., 2008) which extend to the late Cambrian (Runnegar et al.,1979). Chitons are one of the earliest diverging groups of livingmollusks and are often referred to as ‘‘living fossils’’ sincetheir body plan has not significantly changed for over 300 millionyears (Sigwart, 2009). They are typically oval in shape, bilaterally

ll rights reserved.

ls Science and Engineering,etts Avenue, Room 13-4022,

symmetric, and generally range from a few mm to 15 cm in length(Eernisse and Reynolds, 1994). Chitons are of great interest from abiomechanical perspective because instead of a single continuousshell, they possess an assembly of eight dorsal aragonite-basedplates (valves) (Fig. 1a). The first (head) and eighth (tail) platesare semicircular in outline while the intermediate plates are ‘‘but-terfly’’-shaped. These plates provide protection while still allowingfor some degree of flexibility during locomotion over uneven,rough surfaces as well as when rolling defensively into a ball-likeconformation (Fig. 1b) if dislodged from a surface. In many species,the head plate nearly touches the tail plate in the rolled state(Vermeij, 1993). This defensive conformation covers and protectsthe ventral side of the chiton from imminent predation, and mayallow it to be carried out of harm’s way by a passing wave to amore favorable location to right itself (Arey and Crozier, 1919;

http://dx.doi.org/10.1016/j.jsb.2011.12.019

mailto:[email protected]


http://www.sciencedirect.com/science/journal/10478477

http://www.elsevier.com/locate/yjsbi

• z

y

x

Posterior Anterior Girdle

(a)

(b)

8

7

6 5 4

3

2

1

0.5 cm

Girdle

Fig.1. Images of Tonicella marmorea; (a) side view optical microscopy image of adried shell; (b) side view photograph of a recently thawed chiton in a defensive,curved posture.

M.J. Connors et al. / Journal of Structural Biology 177 (2012) 314–328 315

Eernisse, 2007). The curvature of the plate assembly is most oftenconvex, although less frequent concave curvatures have been ob-served. Convex flexure is provided by the ‘‘enrolling’’ muscle,which encircles the shell underneath the outer margin of the plates(Haszprunar and Wanninger, 2000). Plates 2–8 possess two ante-rior projections, the apophyses (sutural laminae), which overlapwith the ventral surface of the adjacent anterior plate. Transversemuscles lie in the overlapping region between neighboring platesand connect them together (Wingstrand, 1985). Also extendingbetween the shell plates is a thin cuticular layer (Beedham andTrueman, 1967), which laterally attaches to a thick chitinous cuti-cle that covers a leathery girdle structure surrounding the plates.

Resistance to penetration is provided by the internal, multilay-ered microstructure of the individual exoskeletal units (Bruet et al.,2008), in this case the chiton plates, which are known to possessone of the most complex multilayered structures of any mollusk(Bøggild, 1930). The multilayered structure of the plates have beenstudied by a number of authors including Haas (1972, 1976), Laghiand Russo (1978, 1980), Baxter and Jones (1981), Kreusch (1981),Poulicek and Kreusch (1983), Carter and Hall (1990), Currie(1992), and Chen (2010). In addition to the periostracum, an outerthin organic layer that is often eroded away on older shells, chitonshell plates are generally composed of seven aragonite-based lay-ers as follows (described from outer to inner):

1) Tegmentum: a granular or irregular simple prismatic layerwhich is penetrated by the aesthetes, an intricate networkof tissue-filled channels that open on each plates’ dorsal sur-face as sensory organs (Vendrasco et al., 2008).

2) Dorsal Mesostracum: a thin simple crossed layer which maybe divideddorsoventrallyinto two sub-layers by a row of aes-thete channels.

3) Articulamentum: a composite prismatic layer that constitutesthe bulk of the apophyses and insertion plates.

4) Ventral Mesostracum: a simple crossed-lamellar layer. Thedorsal and ventral mesostracum layers are collectivelyreferred to the as the Pallial Myostracum.

5) Anterior Myostracum: a granular or irregular simple pris-matic layer which stems from the impressions of the obliqueand post-apophysial latero-pedal muscles.

6) Posterior Myostracum: a granular or irregular simple pris-matic layer which lies in the impression of the transversemuscle.

7) Hypostracum: a simple crossed-lamellar layer that consti-tutes the bulk of the central callus.

The objectives of this study were to (1) to quantify the three-dimensional geometric structure of individual plates, as well asof the entire plate assemblies of chitons in passive (slightly curved,attached to surface) and defensive (rolled, detached from surface)conformations, (2) to determine the spatial distribution of micro-structures present in intermediate plates and microstructural ori-entations relative to macroscopic plate dimensions, (3) toelucidate relationships between the internal multilayering andmicrostructures of the internal plates (‘‘inherent materiality’’)and the larger length scale geometric design to better understandthe balance between threat protection and biomechanical mobil-ity, and (4) to gain insights into the functional and physiologicalconsequences related to survivability of the segmented exoskele-ton (structurally, not biologically). To accomplish these goals, thechiton Tonicella marmorea was analyzed by X-ray micro-computedtomography (microCT) to quantify biomechanically-relevant fea-tures such as the three-dimensional geometry of the individualplates, the inter-plate registration and overlap, and the curvatureand continuity of the entire plate assemblies. Three-dimensionalprinting was employed to create a scaled-up macroscopic syntheticprototype directly from the microCT data to elucidate the detailedplate morphometry and to better visualize the inter-plate articula-tion. The intra-plate microstructure was characterized via mercuryporosimetry, optical microscopy, scanning electron microscopy(SEM), and X-ray diffraction (XRD). Fourier-transform infraredspectroscopy (FTIR) and amino acid analysis (AAA) were used toidentify and compare the composition of organic materials fromthe girdle, shell plates, and inter-plate region. The obtained resultsare discussed in the context of other species with articulating nat-ural armor, in particular the balance between the local protectionmechanisms of the individual plates and the larger length scale de-sign principles of the plate-to-plate interconnections. The new sci-entific information obtained holds potential for comparativemorphometric analyses of chitons, as well as for the developmentof improved bio-inspired protective materials, for example with re-gards to human extremity protection (Arciszewski and Cornell,2006; Bruet et al., 2008; Ortiz and Boyce, 2008; Song et al., 2010;Yao et al., 2010).

2. Materials and methods

2.1. Samples and terminology

Chitons were collected on the coast of Pembroke, Maine andstored frozen until experimentation. When needed, individualplates were separated using forceps and razor blades. Plates werecleaned by sonication (Branson 1510, Danbury, CT) in deionizedwater twice for approximately 45 s.

316 M.J. Connors et al. / Journal of Structural Biology 177 (2012) 314–328

Species were identified as T. marmorea and distinguished fromTonicella rubra by the height/width aspect ratio of the fourth plate(�0.44 vs. �0.29), and appearance of the girdle (leathery to thenaked eye vs. densely clothed with club-shaped, calcareous cor-puscles), respectively (Kaas and Van Belle, 1985). Macroscopicmorphological chiton terminology used was defined and/or stan-dardized by Schwabe (2010) while chiton shell plate layer nomen-clature used was derived from Bergenhayn (1930) and Laghi andRusso (1978, 1980). Molluscan microstructure terminology usedwas defined by Taylor and Layman (1972) and Carter and Clark(1985).

2.2. Micro-computed tomography (microCT)

The complete shells (plates 1–8 and girdle) of T. marmorea werescanned with a micro-computed tomography system (lCT40,Scanco Medical AG, Switzerland) operated at 70 kV and 114 lA.Microtomographic slices were recorded every 10 lm and werereconstructed with 10 � 10 lm voxels in plane. For the chiton pre-sented in Fig. 1a, the 3D information of the plate assembly was par-titioned into contributions from each individual plate using thethreshold and contour functions of Scanco MicroCT Software. Forthe chiton presented in Fig. 1b, the 3D information of the plateassembly was segmented using the threshold and region growthfunctions of image processing software package Mimics (Material-ise, Belgium). Mimics was also used to create all three-dimensionalimages, including those with transparency effects. The three-dimensional geometric information of the plate assemblies wasconverted into three-dimensional polygonal meshes (stereo lithog-raphy) using Scanco MicroCT Software for the chiton presented inFig. 1a or Mimics for the chiton presented in Fig. 1b. Meshes werereduced in size as needed using the smooth, reduce, and remeshtools of 3-matic (Materialise, Belgium). Cross-sectional imageswere created from the surface meshes using the cut function ofthe simulation module of Mimics. Open source image analysis soft-ware ImageJ (Rasband, 1997–2011) was used to measure thedimensions of cross-sectional images. The spatial distribution ofthickness of each individual plate and of the entire plate assembly(Fig. 3a and b, respectively) of the chiton presented in Fig. 1a wascalculated using a spherical method in which each point in 3Dspace is assigned a thickness value corresponding to the diameterof the largest sphere centered on that point which can fit withinthe boundaries of the plate (Bouxsein et al., 2010).

2.3. Calculations of plate-to-plate overlap and curvature

In Fig. 2f, overlap was quantized using aerial projections (in the‘‘y’’–‘‘z’’ plane) of the plates and plate-to-plate overlapping regionsof Fig. 2d and e. For each intermediate plate, total overlap percent-age was calculated by summing the projected areas of the twooverlapping regions, and expressing the sum as a percentage ofthe projected area of the entire plate which encompasses them.For the first and eighth plates, total overlap percentage was calcu-lated by expressing the projected area of the single overlapping re-gion (the overlapping region of plates 1 and 2, and plates 7 and 8,respectively) as a percentage of the projected area of the respectiveplate. In Fig. 5c, overlap was calculated by projecting the overlap-ping regions of plates 3 and 5 onto the long axis (‘‘l’’) of plate 4,summing the projected lengths, and expressing the sum as a per-centage of the cross-sectional length of plate 4. The transverse radiiof curvature of individual plates and the longitudinal radii of cur-vature of the plate assemblies were calculated using open source‘‘Circle Fit (Pratt Method)’’ MatLab (MathWorks, MA) code(http://www.mathworks.com/matlabcentral/fileexchange/22643)written by Nikolai Chernov based on an algorithm developed byPratt (1987).

2.4. 3D-printing

A scaled-up macroscopic prototype of the chiton shell plateassembly was fabricated using a three-dimensional printer (3DP,ZPrinter� 310 Plus, ZCorporation, USA) using the stereolithography(STL) files that were created from the micro-computed tomogra-phy data. The material used was a plaster powder (ZP� 131 pow-der, ZCorporation, USA). Eighty-nine micrometer thick layerswere laid down using a commercially available binder (ZCorpora-tion, USA) at a vertical build speed of 25 mm/h. After printing,the prototype was immersed in a wax bath to ensure a smoothsurface.

2.5. Optical and scanning electron microscopy (SEM)

For cross-sectional imaging, plates were embedded in a fastcure epoxy (Loctite, USA) and sectioned with a diamond impreg-nated saw (Buehler Isomet 5000, Lake Bluff, IL) at 875 rpm. Embed-ded samples were polished stepwise with aluminum oxide padswith roughness varying from 9 to 0.1 lm (South Bay Technologies,CA) and then with 500 nm silica nanoparticles on microcloth(Buehler, IL). Some samples were slightly etched with 0.5 M EDTAfor 5 min to improve contrast between microstructurally distinctlayers. Additionally, some plates were cryofractured by immersingthem in liquid nitrogen for 5 min and then breaking them with ahammer. Optical images were taken with a Nikon ECLIPSE LV100microscope (Tokyo, Japan). Cross-sectional SEM samples werefixed on a steel support using a conductive tape and then sput-ter-coated with �5 nm of gold–palladium in a Denton VacuumDesk II (Moorestown, NJ). A JEOL JSM 6700 (Peabody, MA) scanningelectron microscope was used for imaging at an acceleration volt-age 10 kV. Demineralized SEM samples were secured in a holder,covered with carbon for 1 min using an Edwards S150B sputtercoater (West Sussex, United Kingdom) and imaged with an ESEMXL 30 Philips (Amsterdam, Netherlands) or LEO DSM 982 Gemini(Oberkochen, Germany) scanning electron microscope.

2.6. Thermogravimetric analysis (TGA)

Plates were lightly ground with a pestle and mortar. Sampleswere then vacuum dried at 110 �C overnight to remove residualwater. TGA was carried out from 110 to 500 �C at 2.5 �C/min on aTA Instruments TGA Q50 (New Castle, DE). Weight loss due to com-bustion of organic material was attributed to the temperaturerange 110–475 �C (Zaremba et al., 1998; Neves and Mano, 2005).

2.7. Mercury porosimetry

The volume percent porosity of three plates (2–4) was mea-sured by a mercury porosimeter (Autopore IV 9500, Micromeritics,USA), which operated at pressures between 3.7 kPa and 14 MPa,corresponding to pore diameters of 404 and 0.107 lm, respec-tively. A standard contact angle of 140�, which was used previouslyby Gane et al. (2004) to study the porosity of calcium carbonatestructures, was assumed in the pore size calculations.

2.8. X-ray diffraction (XRD)

The mineral phase of the plates was verified using XRD (PhilipsPANalytical X’ Pert PRO diffractometer with CuKa radiation, Neth-erlands), operating at 45 kV and 40 mA between 10� and 70� (2h).

2.9. Chitin isolation and staining

The organic matrix of chiton plates was isolated by decalcifica-tion using 3 M HCl solution as well as a 7.4 pH Osteosoft™ (Merck)

http://www.mathworks.com/matlabcentral/fileexchange/22643

0

20

40

60

80

100

0

5

10

15

20

25

30

35

1 2 3 4 5 6 7 8

Perc

enta

ge O

verla

p

Proj

ecte

d Ar

ea (m

m2 )

Plate Number

Plate Area (Fig. 2d)

Plate Area (Fig. 2e)

Plate-Plate Overlap Area (Fig. 2d)

Plate-Plate Overlap Area (Fig. 2e)

% Total Overlap (Fig. 2d)

% Total Overlap (Fig. 2e)

•

x

z

y

(a) Dorsal View

L1

T1

Anterior

z

x y x Posterior

(b) Ventral View

z

y

x

Posterior Anterior

(d)

Ventral

Dorsal 0.5 cm

(c)

y

x

z

•

Slits

(e)

(f)

z

y

x •

Posterior Anterior

T1

Fig.2. microCT images of the three-dimensional structure of the armor plate assembly of Tonicella marmorea; (a–d) correspond to the chiton of Fig. 1a while (e) correspondsto the chiton of Fig. 1b; (a) dorsal view displaying position of cross sections L1 and T1; (b) ventral view; (c) side/ventral view; (d) side view; (e) side view displaying positionof cross section T1; (a, b, d and e) include complimentary images created using a transparency effect; (f) projected area in the ‘‘y’’–‘‘z’’ plane (open circle and square symbols)and total overlap percentage (filled circle and square symbols) vs. plate number calculated from the plates and plate-to-plate overlapping regions of the transparent images of(d) and (e). Aerial overlap contributions from all adjacent plates were considered in calculations of total overlap percentage.



solution at room temperature. The procedure was monitored usingstereo, light and fluorescence microscopy (BZ-8000, Keyonce,Japan). Immersion of the isolated organic matrix, inter-plate mate-rial, and girdle in 2.5 M NaOH at 37 �C led to an immediate loss ofbrownish pigment as well as of proteins from these specimens. Thefibrous colorless materials were then washed with distilled waterfive times, dialyzed against deionized water on Roth (Germany)membranes with a MWCO of 14 kDa for 48 h at 4 �C, and finallydried at room temperature. Calcofluor White (Fluorescent Bright-ener M2R, Sigma) was used to confirm the presence of chitin.Samples were placed in 0.1 M Tris–HCl at pH 8.5 for 30 min,stained using 0.1% Calcofluor White solution for 30 min in dark-ness, rinsed five times with deionized water, dried at room temper-ature, and finally observed using fluorescence microscopy (Ehrlichet al., 2007).

2.10. Fourier-transform infrared spectroscopy (FTIR)

Infrared spectra were recorded on a Nicolet 210 FT-IR Spec-trometer. Two hundred and fifty scans were recorded at a spectralresolution of 2 cm�1. All spectra were baseline corrected with atwo-point linear baseline at 845 and 1890 cm�1. An a-chitin stan-dard from the demosponge Ianthella basta (Brunner et al., 2009)was used as a control.

2.11. Amino acid analysis (AAA)

Dry, cleaned plates were ground with a pestle and mortar andweighed. The fine powder was suspended in deionized water. Ina demineralization step, stoichiometric amounts of 1 M HCl wereadded dropwise to the solution over the course of 2 days (Treveset al., 2003). After complete demineralization, the solution wascentrifuged (3000 rpm, 10 min, Eppendorf Centrifuge 5804 R) toseparate soluble material from the insoluble, precipitated material.The soluble material was dialyzed exhaustively (Spectra Pro 7 dial-ysis membranes, 3500 MWCO, part # 132112) to remove salts andsmall molecules. The insoluble material was lyophilized and driedovernight in a vacuum oven at 50 �C. Samples were hydrolyzed forAAA, which was performed using a Waters system with Breezesoftware by Bio-Synthesis, Inc. (Lewisville, TX).

3. Results

3.1. Three-dimensional structure and geometry of the plate assemblyof T. marmorea

The three-dimensional structure of the plate assemblies of tworepresentative T. marmorea chitons (Fig. 1a and b) were derivedfrom renderings of microCT data. In their curved states, the plateassemblies of the chitons seen in Fig. 1a and b have dimensions�1.7 � 0.95 � 0.45 cm and �1.66 � 1.4 � 1.38 cm (length (‘‘z’’-direction) �width (‘‘x’’-direction) � height (‘‘y’’-direction)), respec-tively. The girdle, as well as the inter-plate material (transversemuscle and cuticular layer), were very slightly or non-mineralizedand consequently did not absorb X-rays well enough to be clearlyvisible in the microCT scans. Fig. 2a and b display dorsal and ven-tral views, respectively, of the plate assembly of the smaller chitonof Fig. 1a along with complimentary transparent images highlight-ing plate-to-plate overlap (darkened regions). The broad ‘‘u’’ shapeof the overlapping regions results from each plate’s (except thefirst plate) two anterior projections, the apophyses, which imbri-cate with adjacent anterior plates and are separated by a sinus.Side views of the plate assemblies are shown in Fig. 2d and e,and are accompanied by transparent images. As illustrated by the

transparent image of Fig. 2d, in the smaller plate assembly withlower longitudinal curvature (about the ‘‘x’’-axis), plate-to-plateoverlap is not limited solely to the apophyses; a portion of eachplate’s jugal area (between the sinus, see Supplementary material1a) is also involved in overlap. In contrast, overlap in the jugalareas of each plate is absent or greatly reduced in the transparentside profile of the larger, defensively postured plate assembly withthe greater longitudinal curvature (Fig. 2e). In the plate assemblywith the lower longitudinal curvature (Fig. 2d), the total plateoverlap percentage of the intermediate plates increases from�52% in plate 2 to �65% in plate 7 (Fig. 2f). In contrast, in the plateassembly with the greater longitudinal curvature (Fig. 2e) the totaloverlap percentage of the intermediate plates increases from �44%in plate 2 to �58% in plate 7. The average total plate overlap per-centages of the two plate assemblies presented in Fig. 2d and eare 62.27 ± 11.45 and 48.00 ± 9.07, respectively.

The thickness distribution of the plate assembly of the smallerchiton presented in Fig. 2d (Fig. 3a) shows that the intermediateplates have a similar spatial distribution of thickness, in whichthe central non-overlapping regions are thickened (�0.85 mm) rel-ative to the anterior and posterior overlapping plate edges(�0.4 mm) and apophyses (�0.2 mm). The thin (green) and thick(red) regions of the plate assembly in Fig. 3b mirror the overlap-ping and non-overlapping regions, respectively, of the transparentimage of the plate assembly in Fig. 2a. The eight plates of theassembly also have both similar average thicknesses (411 ± 31)and average standard deviation of thickness (222 ± 23) (Fig. 3c).Although thickness is relatively constant across the assembly,there is an asymmetric trend in plate volume (Fig. 3c); the headplate is �40% larger than the tail plate by volume. Additional struc-tural data is available in Supplementary material 1, which also la-bels the anatomy of individual plates and defines structuralparameters. Visualization of plate morphology and imbrication isfacilitated using a macroscopic prototype of the plate assembly(Supplementary material 2).

Fig. 4a and b show the L1 cross sections of the plate assembliesof the chitons presented in Fig. 1a and b, respectively. L1 bisectsthe jugal sinus of each plate 2–8, so the apophyses are absent. Inthe plate assembly with lower longitudinal curvature (Fig. 4a),the heterogeneous cross-sectional plate geometry and plate-to-plate overlap result in a relatively constant thickness and approx-imately continuous convex curvature of the plate assembly as awhole. To approximate their longitudinal (about the ‘‘x’’-axis) radiiof curvature, 16 landmark points were selected for each assemblyfrom the two end points of each of the eight plates along their lon-gitudinal axes. As illustrated in Fig. 4c and d, circles were fit tothese points using a direct least squares fitting algorithm (Pratt,1987). The longitudinal radii of curvature of the L1 cross sectionsof Fig. 4a and b are 9.15 and 7.39, respectively. The transverse(about the ‘‘z’’-axis) radii of curvature of the fourth plates of eachassembly were approximated using the same method. Eleven land-mark points from cross section T1 were used in the calculation foreach plate: the midpoint of cross section L1, the midpoints of crosssections L2–L5 (left side) and their analogous midpoints on theright side, and left and right base points of the insertion plates(Fig. 5a and b). The transverse radii of curvature of the T1 cross sec-tions of the fourth plates seen in Fig. 5a and b are 4.60 and 6.71,respectively. To compare the curvature of the two plate assembliesand account for the size disparity, two curvature indices were de-fined (Fig. 4f). The first index, the longitudinal curvature index(L.C.I), is defined as the length of plate 4 divided by the longitudinalradius of curvature of the plate assembly in cross section L1. Thisyields L.C.I. values of 0.43 and 0.70 for the plate assemblies ofFig. 4a and b, respectively. The second index, the transverse curva-ture index (T.C.I), is defined as the width of plate 4 divided by its

0

0.2

0.4

0.6

0.8

1

0

5

10

15

20

25

1 2 3 4 5 6 7 8

Thic

knes

s (m

m)

Volu

me

(mm

3 )

Plate Number

Total Volume

Average Thickness(c)

1

2

3

4

5

6

7

8 (a)

JS APAP

(b)

Posterior

Anterior

• x

z

y

3

4

3 mm

mm0 0.94

Fig.3. Thickness distribution of the armor plate assembly of Tonicella marmorea presented in Fig. 1a; (a) dorsal view of the spatial distribution of thickness for each plate 1–8;(b) dorsal view of the spatial distribution of thickness of the entire plate assembly; (c) total volume (triangle symbols) and average thickness (circle symbols) as a function ofplate number. Error bars in (c) refer to standard deviations of the dataset. Terminology: AP, apophyses; JS, jugal sinus.


transverse radius of curvature in cross section T1. This yields T.C.Ivalues of 0.48 and 0.50 for the fourth plates of Fig. 5a and b,respectively.

Fig. 4e shows how the inter-plate separation distance changesas a function of gap number for the two plate assemblies seen inFig. 4a and b. Gap number ‘‘n’’ corresponds to the gap betweenplate number ‘‘n’’ and ‘‘n + 1’’. The inter-plate separation distance

of gap ‘‘n’’ was measured along the long axis of plate number‘‘n + 1’’ in cross section L1. For example, the separation distance be-tween plates 4 and 5 (Gap 4) was measured along the long axis(‘‘l’’) of plate 5 (Fig. 4b). To account for the size disparity betweenthe two assemblies, the separation distances were normalized bydividing each one on by the length of the fourth plate of therespective assembly. As seen in Fig. 4e, the normalized separation

Fig. 6a Fig. 6b

L.C.I 0.43 0.70

T.C.I 0.48 0.50

R/t (longitudinal) 10.73 6.44

R/t (transverse) 5.39 5.84

2 mm

(a) (b)

• z

y

x

L1L1

Posterior Anterior

Ventral

Dorsal

P8

P7

P6

P5

P4

P3

P2

P1

(c) (d)

(f)

ℓ

L. Radius = 9.15 mm

L. Radius = 7.39 mm

0.00

0.06

0.12

0.18

0.24

0.30

1 2 3 4 5 6 7

Nor

mal

ized

Inte

r -Pl

ate

Sepa

ratio

n

Gap Number

(Fig. 6a)

(Fig. 6b)

(e)

Fig.4. L1 (longitudinal bisector) cross-sectional structure of the plate assembly of T. marmorea; (a) cross section L1 of plate assembly of the chiton presented Fig. 1a; (b) crosssection L1 of plate assembly the chiton presented in Fig. 1b; (c and d) circles fit to 16 landmark points (the two end points of each of the eight plates along their longitudinalaxes) from (a) and (b) using a direct least squares fitting algorithm (Pratt, 1987); (e) normalized inter-plate separation distance vs. gap number. Gap number ‘‘n’’ correspondsto the gap between plate number ‘‘n’’ and ‘‘n + 1’’. The inter-plate separation distance of gap ‘‘n’’ was measured along the long axis of plate number ‘‘n + 1’’ in cross section L1and normalized by dividing by the length of the fourth plate of the respective assembly; (f) curvature indices and shell thickness ratios. The longitudinal curvature index(L.C.I) is calculated by dividing the length of plate 4 by the longitudinal radius of curvature of the plate assembly in cross section L1. The transverse curvature index (T.C.I) iscalculated by dividing the width of plate 4 by its transverse radius of curvature in cross section T1. Shell thickness ratios were calculated by dividing the appropriate radius ofcurvature by the average of the maximum thicknesses of the fourth plates in cross sections of L1–L5 (Fig. 5a and b).


distances of gaps 1 and 2 are similar in both plate assemblies.However, the normalized separation distances of gaps 3, 4, and 7are �75%, �100%, and 200% greater, respectively, in the plateassembly with a L.C.I. of 0.70 (Fig. 4b) relative to the plate assemblywith a L.C.I of 0.43 (Fig. 4a). The normalized separation distance of

gaps 5 and 6 are also larger in the plate assembly with the largerL.C.I., although the increase is not as pronounced. In both plateassemblies, the trend of normalized separation distance vs. gapnumber is asymmetric; the separation distance of gap 2 is �160%larger than that of gap 6.

(a) (b) T1

T1

• x

y

zL1L2

L3

L4

L5

P4P5

L1L2

L3

L4

L5

P4 P5

k k

L1 2 mm

L2

L3

L4

L5

P4

P5

P3

L1

L2

L3

L4

L5

Ventral

Dorsal

Posterior Anterior

P4

P5

P3

ℓ

VTC

AP

•••

••

• •

••

••

T. Radius = 4.6 mm •

••

•• •

••

••

•T. Radius = 6.71 mm

0.50

0.75

1.00

1.25

1.50

0

20

40

60

80

100

1 2 3 4 5

Max

. Thi

ckne

ss (

mm

)

% T

otal

Ove

rlap

Cross Section (L)

% Total Overlap (Fig. 7a)% Total Overlap (Fig. 7b)Max. Thickness (Fig. 7a)Max. Thickness (Fig. 7b)

(c)

Fig.5. Cross-sectional structure of the plate assembly of T. marmorea; (a and b top) T1 cross sections of the fourth plate 4 of the plate assemblies of the chitons presented inFig. 1a and b, respectively. Black circles indicate the eleven points used to calculate the transverse radius of curvature of the fourth plate of each plate assembly using a directleast squares fitting algorithm (Pratt, 1987); (a and b bottom) Morphological changes of longitudinal cross sections (L1–L5) of the fourth plate of each plate assembly. Crosssections L1–L5 were taken at 20% increments along line segment ‘‘k’’; (c) total overlap percentage and maximum thickness of plate 4 vs. longitudinal cross section number.Total overlap percentage was calculated by projecting the overlapping regions of plates 3 and 5 onto the long axis (‘‘l’’) of plate 4, summing the projected lengths, andexpressing the sum as a percentage of the cross-sectional length of plate 4. Maximum cross-sectional thickness was measured perpendicular to the long axis (‘‘l’’) of plate 4.Terminology: AP, apophyses; VTC, ventral tegmental callus.


Fig. 5a and b (top) show the dimensions and geometry of trans-verse cross section T1 of the two plate assemblies. T1 was taken inthe region in which plates four and five overlap, and thus includesboth plates. Longitudinal cross sections L2–L5 were taken at 20%increments along the line segment which runs diagonally from

the midpoint of cross section L1 to the eave of the tegmentum be-low. A similar line segment was named ‘‘k’’ by Bergenhayn (1930Fig. 1, page 14), and we adopt this notation (Fig. 5a). Fig. 5c dis-plays how the total overlap percentage of a representative inter-mediate plate (number four) evolves laterally from L1 to L5. In


the both plate assemblies, overlap is heterogeneous in nature, butthe heterogeneity is more pronounced in the plate assembly withthe higher L.C.I. The total overlap percentage of plate four of theplate assembly with the lower L.C.I. is �60% in cross sections L1and L2, and �82.5% in cross sections L3–L5. In contrast, the totaloverlap percentage of plate four of the plate assembly with thehigher L.C.I. is �0% in cross sections L1 and L2, and increasesapproximately linearly from �50% to �90% across cross sectionsL3–L5. The average total overlap percentages of the fourth platesof the plate assemblies with L.C.I. of 0.43 and 0.70 are 73.9% and44.6%, respectively.

3.2. Multilayered microstructure of the intermediate plates of T.marmorea

The intermediate plates of T. marmorea are composed of six ara-gonite-based layers (Fig. 6). The dorsal mesostracm was not ob-served in any plates used in this study, and so we refer to theventral mesostracum simply as the mesostracum. Fig. 6a and bcontain representative longitudinal (L4) and transverse (T1)cross-sections, which are each accompanied by a schematic dia-gram that illustrates the spatial distribution of the six layers. Thekey to the schematics (Fig. 6d) relates textures and colors to micro-structures and layers.

The outermost layer, the tegmentum, has a homogeneous thick-ness of 155 lm and is infiltrated with the aesthete canal system(Fig. 6a), which constitutes �7.5% of each plate by volume as mea-sured by mercury porosimetry (Supplementary material 1i). Mainaesthete channels in the tegmentum run roughly longitudinally,are separated by �20 lm, and have a diameter of �40 lm, whichlies in the lower end of the range (40–75 lm) found across genusTonicella (Vendrasco et al., 2008). Scanning electron microscopy re-vealed that the tegmentum has a fine-grained homogeneousmicrostructure (Fig. 7b and g). XRD of the dorsal shell surfacefound no preferred grain orientations (Fig. 8a). Grains are approx-imately 1 lm in diameter, but are slightly elongated along one axisventrally, below the main aesthete channels. Posteriorly, the teg-mentum folds underneath the plate and becomes the posteriormyostracum (layer 5), which lies in the impression of the trans-verse muscle (Fig. 6a). The transverse muscle, which runs longitu-dinally from the eave of the tegmentum to the tip of the apophyses,fills the gap between overlapping plates (Fig. 6c). The thickness ofthe transverse muscle ranges from �100 to �250 lm. Slit ray andjugal aesthete channels that begin in the tegmentum, pass thoughthe lower layers, and exit on the ventral shell surface are insulatedby a fine-grained homogeneous layer that is a couple of micronsthick. The second layer, the articulamentum, has a composite pris-matic microstructure (Fig. 7c and h). Bundles of prisms fan out dor-sally and ventrally from the center of the layer. The length of theprism bundles ranges from approximately 3–10 lm. Individualprisms are rectilinear in shape and have dimensions 0.3 � 0.3 lm(width � height). The thickness of the articulamentum ranges from0 mm in the jugal area to approximately 0.9 mm laterally, at thestart of the insertion plates (Fig. 6b).

The articulamentum is followed by the mesostracum, which hasa simple crossed-lamellar microstructure (layer 3) (Fig. 7d and i). Itbranches off from the hypostracum near the plates’ body diagonal,along which the slit ray aesthete channels run (Fig. 6a). The longaxes of first order lamellae are oriented at a �45� angle relativeto the ventral plate surface and have variable width between 1and 2 lm. The layer lacks well-defined second order lamellae.Third order lamellae are rectilinear in shape and have dimensions�15 � 0.3 � 0.3 lm (length �width � height); each first order col-umn is 3–6 third order lamellae wide. The angle between bundlesof third order lamellae in adjacent first order columns is �40�.Third order lamellae on the interface between adjacent first order

columns possess a ‘‘wavy’’ surface topography, which has a wave-length approximately equal to their width (0.3 lm) (Fig. 7i). Theperiodic surface elevations on one side of the interface align withsurface depressions on the opposite side. Layers 4 and 5, the ante-rior and posterior myostracum, respectively, both have a fine-grained homogeneous microstructure with grains approximately1 lm in diameter (Fig. 7e and j). The anterior myostracum stemsfrom the impressions of the oblique muscles while the posteriormyostracum lies in the impression of the transverse muscle.Together, the articulamentum, mesostracum, and anterior myos-tracum form the apophyses (Fig. 6a). The bottom layer, the hypos-tracum, has a simple crossed-lamellar microstructure identical tothat of mesostracum, except that the long axes of the first orderlamellae are oriented perpendicular to the ventral plate surface(Fig. 7f and k). This was supported by XRD of the ventral shell sur-face, which found two preferred grain orientations correspondingto the two lamellar dip directions (Fig. 8a, top dotted line). First or-der columns often slightly deviate from the perpendicular orienta-tion, become thinner and disappear, or become thicker and branch.The hypostracums’s transverse distribution of thickness is inverseto that of the articulamentum; it is approximately 0.9 mm thickin the jugal area and absent laterally, at the start of the insertionplates (Fig. 6b). Longitudinally, the hypostracum composes thebulk of the central callus and is absent from the apophyses andinsertion plates.

3.3. Chemical composition of the intra-plate organic matrix, girdle, andinter-plate material of T. marmorea

The intermediate plates of T. marmorea have an organic matrixcontent of �2.6 wt.% (aragonite content of �97.4 wt.%), as mea-sured by thermogravimetric analysis (Supplementary material1i). The organic matrix, which was obtained after a 24 h gentledemineralization using Osteosoft™ solution at room temperature,showed a strong resistance to alkali treatment (Supplementarymaterial 3a). It remained stable after submersion in 2.5 M NaOHat 37 �C for 30 days. The girdle and inter-plate material, which iscomposed of the transverse muscle and cuticular layer, displayedsimilar properties. After alkali dissolution of their proteinaceouscomponents, Calcofluor White staining (Supplementary material3b) and FT-IR spectroscopy indicated the presence of a-chitinwithin the shell plate organic matrix, girdle, and inter-plate mate-rial (Fig. 8b). Amino acid analysis of the intra-plate organic matrixrevealed that the total amount of protein in the shell plates is0.17 wt.% (Supplementary material 4b).

4. Discussion

In segmented natural exoskeleton systems, there exists aninterplay between the ‘‘inherent materiality’’ of the individual ar-mor sub-units (i.e. microstructure, multilayering, anisotropy, grad-ing, etc.) and the larger length scale morphometric or geometricdesign to simultaneously achieve the necessary protection andbiomechanical flexibility/mobility. This interplay can vary dramat-ically among different systems. Here, we explore this topic in amodel system of the chiton T. marmorea by using a suite of exper-imental techniques to quantify critical design aspects such as (fromsmallest to largest length scale): (1) the microstructure, multilay-ering and material properties of the individual armor sub-units,(2) the geometry of the individual armor sub-units, (3) theinterconnections between armor sub-units, (4) the degree andtype of armor unit overlap, and the interaction and correlationbetween these design aspects. We will discuss each of theseaspects following.

Layer Name Microstructure Symbol(s)

1 Tegmentum Granular

2 Articulamentum Composite Prismatic

3 Mesostracum Crossed-Lamellar a) b)

4 Ant. Myostracum Granular

5 Post. Myostracum Granular

6 Hypostracum Crossed-Lamellar a) b)

- Muscle Tissue -

0.5 mm

(a) 1

3 6

L4

21

2

Dorsal

Ventral Anterior Posterior

4

Transverse Muscle

(b)

x x

y

z

2 mm

T16

2

SR

IP IP

(c)

AP

SL

0.5 mm

Transverse Muscle

(d)

Oblique Muscle

5

1

34

Fig.6. Cross-sectional structure of intermediate plates of Tonicella marmorea by dark-field optical microscopy; (a) composite image of polished longitudinal cross-section (L4)and complimentary schematic; (b) composite image of polished transverse cross-section (T1) and complimentary schematic; (c) longitudinal cross-section displayingtransverse muscle between two adjacent intermediate plates; (d) table relating textures in schematics to layers and microstructures. Terminology: AP, apophyses; IP,insertion plate; SR, slit ray.


In order to understand functional design, it is important to con-sider the environment and types of predatory attacks against forwhich specific natural armor systems are designed. Langer(1978) observed that T. marmorea from the northwestern Atlanticis found most abundantly on subtidal calcareous algae coveredrock, and displays cryptic coloration. Field observations led Langer(1978) to suspect that the primary predators of T. marmorea were

the sea star Leptasterias littoralis in shallow water (less than 6 m)and the wrasse Tautogolabrus adspersus and winter flounder Pseu-dopleuronectes americanus at greater depths; secondary predatorsinclude the crabs Cancer borealis and Cancer irrorarus and the lob-ster Homarus americanus. Possible predatory attacks include biting(fish), pinching with claws (lobsters and crabs), drilling with rad-ulas (sea snails), inserting the cardiac stomach underneath or

1- T

egm

entu

m

2- A

rticu

lam

entu

m

4 3

Dorsal

6- H

ypos

tracu

m

Mus

cle

50 µm Ventral

Polished & Etched Sample Cryofractured Samples

c

d

e

f

b

3 µm

0.5 µm

1.5 µm

1.5 µm

1 µm 1 µm

1 µm

1 µm

1 µm

1 µm

0.3 µm

(a) (b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

(k)

L4

Fig.7. Spatial distribution of microstructures of intermediate plates of Tonicella marmorea by SEM; (a) composite image of the boxed (white dashes) region of longitudinalcross section L4 of Fig. 6a, which was polished and etched with 0.5 M EDTA for 5 min to increase image contrast. Images (b–f) correspond to magnified regions of (a) andpreserve microstructure orientation with respect to cross section L4. Images (g–k) were obtained from cryofractured samples and display true microstructure dimensions andgeometry; (b and g) layer 1, tegmentum, fine-grained homogenous; (c and h) layer 2, articulamentum, composite prismatic; (d and i) layer 3, mesostracum, simple crossedlamellar; (e and j) layer 4, anterior myostracum, fine-grained homogenous; (f and k) layer 6, hypostracum, simple crossed lamellar.


between the shell plates (sea stars), and smashing against rocks(birds).

4.1. Comparison of the shell plate microstructures of T. marmorea withother Polyplacophora and Mollusca

The mesostracum was originally named by Bergenhayn (1930)for its position between the tegmentum and articulamentum inAcanthopleura spinigera. He observed that it was similar in struc-ture to the hypostracum, which is now recognized as having a sim-ple crossed-lamellar microstructure. The mesostracum has been

observed lining both the dorsal and ventral sides of the articula-mentum in Chiton olivaceous (Kreusch, 1981;Laghi and Russo,1978; Poulicek and Kreusch, 1983), Leopidopleurus cajentanus(Laghi and Russo, 1980), Liolophura gainardi (Kreusch, 1981;Poulicek and Kreusch, 1983), and Liolophura japonica (Chen,2010). However, we only observed it lining the articulamentumventrally in T. marmorea, and Baxter and Jones (1981) did not ob-serve it at all in Lepidochitona cinereus. In addition, Kreusch(1981) observed the mesostracum within the articulamentum in-stead of lining it dorsoventrally in Cryptochiton stelleri. Laghi andRusso (1978, 1980) and Kreusch (1981) note that in species which

25 30 35 40 45 50 55

Inte

nsity

Two-Theta (deg.)

VentralSurface

DorsalSurface

Aragonite

(102

) (1

12)

(a) (1

11)

1628

(b)

1543

1374

1151

1107

1061

1008

947

890

6508501050125014501650

Inte

nsity

Wavenumber (cm-1)

α-Chitin standard from Organic Matrix after NaOH TreatmentNative Inter-Plate MaterialInter-Plate Material after NaOH TreatmentNative GirdleGirdle after NaOH Treatment

I. basta

Fig.8. Chemical composition of intermediate plates of Tonicella marmorea; (a) X-raydiffraction pattern of the dorsal and ventral surfaces of an intermediate plate; (b)infrared spectra of the intra-plate organic matrix, girdle, and inter-plate materialbefore and after NaOH treatment to remove proteins.


contain the dorsal mesostracum, it is split dorsoventrally into twosub-layers by a row of aesthete channels. Thus, while the interme-diate plates of T. marmorea have six layers, those of other chiton

species that contain the dorsal mesostracum may be regarded ashaving seven, or possibly eight. Laghi and Russo (1978, 1980),Kreusch (1981), and Poulicek and Kreusch (1983) refer to the dor-sal and ventral mesostracum layers collectively as the pallial myos-tracum. With the sole exception of the dorsal mesostracum, thespatial distribution of microstructures in T. marmorea, C. olivaceous(see longitudinal and transverse cross-sectional schematics inFigs. 2 and 4, Laghi and Russo, 1978), and L. gainardi (see longitu-dinal schematic in Fig. 12, Kreusch, 1981) are very similar. Poulicekand Kreusch (1983) note that in the evolutionary series of chitons,the tegmentum becomes thinner (disappearing almost completelyin the more evolved Cryptochiton), the articulamentum becomesthicker, and the hypostracum, mesostracum, and myostracal layersshow little variation in proportions.

Similar orientation, thickness, and branching irregularities ofthe first order lamellae of the crossed-lamellar layers of the shellplates of T. marmorea have also been reported in the shells of othermollusks (Currey and Kohn, 1976; Taylor and Reid, 1990; Wilmotet al., 1992). The width of first order lamellae in the crossed-lamel-lar layers of the shell plates of T. marmorea (1–2 lm), Acanthople-ura brevispinosa (1.7–5 lm) (Wilmot et al., 1992), Chiton olivaceus(2–3 lm) (Laghi and Russo, 1978), Lepidopleurus cajetanus(�3 lm) (Laghi and Russo, 1980), Liolophura japonica (�3 lm)(Chen, 2010), as well as the range found in 36 chiton species rep-resenting 19 genera (3–5 lm) (Haas, 1972) lies in the extreme low-er end of the range (1–40 lm) found in crossed-lamellarmicrostructures across Mollusca (Carter and Clark, 1985; Wilmotet al., 1992). As previously observed by Haas (1976, 1981), Poulicekand Kreusch (1983), and Carter and Hall (1990), the simplecrossed-lamellar structures present in chiton shell plates are de-void of well-defined second order lamellae. The width of third or-der lamellae found in the crossed-lamellar layers of T. marmorea,Chiton olivaceus (Laghi and Russo, 1978), and Lepidopleurus cajet-anus (Laghi and Russo, 1980) are identical (�300 nm), and similarto the range (200–300 nm) observed by Haas (1972). The angle be-tween the two dip directions of third order lamellae in adjacentcolumns of first order lamellae in T. marmorea is approximately40�, which is similar to the dip angles previously observed in Chi-ton olivaceus (�50�) (Laghi and Russo, 1978), Lepidopleurus cajet-anus (�45�) (Laghi and Russo, 1980), and Acanthopleurabrevispinosa (32–45�) (Wilmot et al., 1992), and much less thanthe range (60–85�) observed by Haas (1972).

4.2. Chemical composition of the shell plate organic matrix of T.marmorea

The intermediate plates of T. marmorea were determined tohave an organic matrix content �2.6 wt.% by TGA. Poulicek andKreusch (1983) found that the shell plate organic matrix contentranged from 0.48 to 1.55 wt.% (greatest in Tonicella lineata) in 11species; however, they only considered the insoluble portion ofthe matrix in their measurements. In contrast to the proteinaceouscomposition of organic matrices of other mollusk shells, those ofchiton shell plates possess a very large amount of the aminopoly-saccharide chitin (Furuhashi et al., 2009). An earlier study foundthe range of the weight percent of chitin in the insoluble portionof the shell plate organic matrix, girdle, and organic matrix of themineralized girdle spicules to be 16–41, 1.6–16, and 0.2–15,respectively (Poulicek and Kreusch, 1983). In addition, the shellplate organic matrix of Acanthopleura villantii was found to containa chitin/protein ratio of 6.9, which is over a hundred times largerthan that found in shells of other mollusks, specifically bivalves(Furuhashi et al., 2009). Our results are consistent with these stud-ies. Because of its resistance to alkali treatment, it is very simple toisolate chitinous material from most composite biomaterials,including cases where chitin is bound to proteins or pigments


(Ehrlich, 2010a; Ehrlich et al., 2010). After alkali treatment, theshell plate organic matrix, girdle, and inter-plate material of T.marmorea displayed FTIR spectra very similar to an a-chitin stan-dard from the demosponge I. basta (Fig. 8b). The protein content(0.17 wt.%) of the shell plates of T. marmorea is similar to the rangefound in the composite prismatic spines and scales (0.07–0.23 wt.%) of Acanthopleura vaillantii, Acanthopleura spinigera,Nuttalina fluxa, and Ischnochitonina sp. (Treves et al., 2003). Thisis consistent with results of Taylor and Layman (1972), who foundthat non-nacreous microstructures including crossed-lamellar,composite prismatic, and homogeneous have a protein content lessthan 0.4 wt.%. The amino acid profile of the shell plate organic ma-trix of T. marmorea (Supplementary material 4) is consistent withthose of Chiton marmoratus and Acanthopleura granulata (Haas,1972), with the sole exception of glycine content, which is about8 mol% greater in C. marmoratus and A. granulata. Calcification ofthe shell plates is likely controlled by the templating activity ofacidic proteins (Ehrlich, 2010b), which may be attached to the chi-tinous network of the organic matrix (Supplementary material 3c).

4.3. The biomechanical role of the inherent materiality of theindividual armor sub-units

The multilayered structure of the individual armor sub-unitsplays a significant role in direct penetration resistance in complexmultiaxial loading configurations by predators (Bruet et al., 2008;Wang et al., 2009). The sequence and spatial heterogeneity of thelayers of the intermediate plates of T. marmorea suggests a struc-tural response to different loading conditions experienced by dif-ferent regions of the plates. The transverse distribution ofcrossed-lamellar hypostracum (�900 lm thick in the jugal areaand decreasing to zero laterally, Fig. 8b) and orientation of first or-der lamellae (long axes perpendicular to the ventral plate surface)may function to resist transverse bending (about the ‘‘z’’-axis).Three-point bending tests by Currey and Kohn (1976) on samplesof Conus spp. showed that the flexural strength of the crossed-lamellar microstructure is highly anisotropic; they determinedthe flexural strength to be �70 MPa in the axial direction (perpen-dicular to long axes of first order lamellae) and �200 MPa in thetransverse direction (parallel to long axes of first order lamellae).Bending in the axial direction can break the layer by simply pullingthe first order lamellae apart from each other, while bending in thetransverse direction cannot break the layer without breaking eachfirst order lamellae across its long axis (Currey and Kohn, 1976).The bending tendency of the plates will depend on the boundaryconditions at the ambitus (Telford, 1985). Possible ‘‘abutment’’ ef-fects of the girdle and concave substratum, and ‘‘tie rod’’ effects ofthe muscular system will both increase the bending tendency.

Multiple ‘‘channel’’ cracking between first order lamellae isknown to be important to overall damage tolerance of shells withinner crossed-lamellar layers (Kuhn-Spearing et al., 1996; Kamatet al., 2000), and is likely an important energy dissipation mecha-nism in the shell plates of T. marmorea during longitudinal bendingdeformation. The structure of the first order lamellae found in thetwo crossed-lamellar layers, the inner hypostracum and middlemesostracum, of the shell plates of T. marmorea permits twotoughening mechanisms that may generate high channel crackdensities during fracture: crack bridging by irregular first orderlamellae (Kuhn-Spearing et al., 1996), and crack bridging by theroughness (‘‘waviness’’) of the third order lamellae on first orderinterfaces (Kamat et al., 2004). The lateral area of the plates, whichlikely experiences the largest longitudinal bending deformations,contain two middle layers (the articulamentum and mesostracum)whose first order interfaces differ in orientation from those of thehypostracum. This arrangement permits crack deflection andbridging in the middle layers – a mechanism which is responsible

for a large portion of the energy dissipated during fracture of theshell of Strombus gigas (Kamat et al., 2000; Kamat et al., 2004).The aforementioned surface waviness of the third order lamellaemay generate a strain hardening mechanism similar to that foundin nacre, in which the ‘‘dovetails’’ of individual nacre tablets pro-duce a progressive tablet locking in tension that is responsiblefor a strain at failure (�1%) that is an order of magnitude greaterthan that of non-biogenic aragonite (Barthelat et al., 2007).

The transverse thickness distribution (�900 lm thick laterallyand decreasing to zero centrally towards the jugal area, Fig. 8b)of the composite prismatic articulamentum, suggests that it func-tions to resist circumferential (roughly longitudinal, parallel totransverse ‘‘z’’-axis in Fig. 8b) tensile stress and to provide a stiff,hard layer to resist side penetration and pinching. While the prismbundles in the articulamentum are aligned parallel to interfaces ofadjacent layers (the tegmentum dorsally and mesostracum ven-trally) in the middle of the articulamentum, they are alignedapproximately perpendicular to theses interfaces at the layer junc-tions. This likely serves to provide discrete inter-prism structuralpathways for cracks to propagate ventrally for easy arrest by thecrossed-lamellar layers (Wang et al., 2009). The inner articulamen-tum and hypostracum layers may also function to resist drilling at-tacks. Chiton tuberculatus has been observed alive with drill holesthat bore through the tegementum, but left the inner layers intact(Arey and Crozier, 1919).

In contrast to the inner shell plate layers, the outer tegmen-tum’s primary role is likely sensory rather than mechanical in nat-ure; the aesthetes, a complex sensory network of tissue-filledchannels, are embedded in the tegmentum. In a secondarymechanical role, the tegmentum may operate as a brittle outershield, in which the main aesthete channels act as sacrificial stressconcentrators that distributes a concentrated load (e.g. biting at-tack) over a large area of the plate. The homogeneous microstruc-ture of the tegmentum implies that it may also function to resistdissolution. Walter and Morse (1984) showed that finely crystal-line microstructures have higher dissolution rates than coarserones, which lead Taylor and Reid (1990) to suspect that coarsegranular or irregular prismatic microstructures are moredissolution resistant than the finely crystalline crossed-lamellarmicrostructure. As previously proposed by Haas (1981), the homo-geneous myostracal layers’ location in the impressions of the late-ro-pedal, oblique, and transverse muscles suggests that thehomogeneous microstructure may be best suited for muscleattachment.

4.4. Geometric design aspects of the armor sub-units of T. marmorea

The transverse curvature (curvature of the chiton plates as ob-served in the ‘‘x’’–‘‘y’’ plane sections) provide an arch-enhancedstiffness to resist bending from both top and lateral loading con-ditions; this geometry enhancement to mechanical robustness isconsistent with the thickness distribution of the articulamentumand hypostracum layers. In less curved conformations (e.g. whenattached to a flat substrate) the longitudinal cross-sectional shapeof each plate provides a curved outer surface that achieves anearly continuous outer surface of the plate-assembly. Thecross-sectional shape (centrally thickened) and thickness distribu-tion of each unit imparts an overall spatially uniform thicknessdistribution in the plate-assembly. This effect has also been ob-served in other segmented natural armor systems including theplate assembly of the marine three spine stickleback Gasterosteusaculeatus (Song et al., 2010) and is likely a universal design prin-cipal. Together these features provide the geometric enhancementand coverage for protection from penetration, bending, andpinching attacks.


4.5. The interconnections between armor units

In contrast to physically interconnected joints observed in someother natural exoskeletons (e.g. the peg-and-socket joint of thescales of Polypterus senegalus (Pearson, 1981), the sliding hinge/ellipsoidal joint of the lateral plates of Gasterosteus aculeatus (Songet al., 2010), the ball-and-socket joint of sea urchin spines (Takah-ashi, 1967)), T. marmorea utilizes an actuating sandwich structureconsisting of two rigid armor plates separated by a more compliantand actuating transverse muscle to achieve biomechanical motion.The transverse muscles are protected by the ventral tegmental cal-lus, which probably functions to reduce the dorsal separation dis-tance between intermediate plates. The contact of the apophyses ofthe intermediate plates with ventral surface of their anteriorneighbors migrates laterally from the L3 cross section to the L5section as the L.C.I. of the plate assembly increases. The close prox-imity of the overlapping regions of adjacent plates appears tostrongly limit lateral translation (along the ‘‘x’’-axis) and rotationabout the ‘‘y’’ axis by the intermediate plates. Longitudinal transla-tion (‘‘z’’ direction) of the plates is likely constrained by the compli-ance of the underlying muscular system and surrounding girdle.Although the defensive conformation (Fig. 1b) protects the under-lying soft parts of T. mamorea from potential predators, it createsvulnerability by dramatically increasing the separation distanceof gaps 3 and 4 of the plate assembly by 75% and 100%, respec-tively, relative to the passive, slightly curved conformation(Fig. 1a). Sea stars have been observed inserting their cardiac stom-achs into the gaps between the plates of chitons that rolled into aball after being captured (Town, 1979).

4.6. The degree and type of armor unit overlap

As the L.C.I. of the plate assembly increases, plate-to-plate over-lap decreases and becomes more heterogeneous. In the two plateassemblies with a L.C.I. of 0.43 and 0.70, the total overlap percent-age of the fourth plate (relative to its long axis ‘‘l’’) ranges fromapproximately 60–82.5% (average = 73.9%) and 0–90% (aver-age = 44.6%), respectively, between cross sections L1 and L5. Inaddition, the average plate-to-overlap percentages (calculatedfrom projections of the plates and overlapping regions onto the‘‘y’’–‘‘z’’ plane) of the two plate assemblies with L.C.I. of 0.43 and0.70 are 62.27 ± 11.45 and 48.00 ± 9.07, respectively. The reductionof plate-to-plate overlap, along with the aforementioned increasein plate-to-plate separation distance, reduces the protection ofplate assembly in the defensive conformation relative to lesscurved states. Species Cryptochiton stelleri and Chiton virgulatushave occasionally been observed exhibiting a righting behavior inwhich the plate assemblies undergo a concave curvature and slighttwisting motion when the chitons were dislodged and turned up-side down by hand, or an aquarium water jet. However, Cryptochi-ton stelleri always rolled into ball when dislodged and vigorouslystimulated by touch. Perhaps in some species the defensive confor-mation serves as a last resort in the presence of imminent threatswhen the chiton does not have enough time and/or security toright itself when dislodged.

5. Conclusions

In summary, this study investigates the three-dimensionalstructure and conformational transition of the eight plate shellassembly of the chiton T. marmorea, and considers its biomechan-ical consequences, as follows. One contribution of this work is theelucidation of the detailed mechanism of conformational changefrom a passive (flattened or slightly curved, attached to surface)to a defensive (rolled, detached from surface) state of the shell

plate assembly. The passive and defensive conformations exhibiteddifferences in longitudinal curvature index (0.43 vs. 0.70), averageplate-to-plate overlap (�62% vs. �48%), cross-sectional overlapheterogeneity (60–82.5% vs. 0–90%, fourth plate), and plate-to-plate separation distance (100% increase in normalized separationdistance between plates 4 and 5), respectively. In contrast to phys-ically interconnected joints observed in some other natural exo-skeletons (e.g. peg-and-socket, ball-and-socket, etc.), T. marmoreautilizes an actuating sandwich structure consisting of two rigid ar-mor plates separated by a more compliant and actuating trans-verse muscle to achieve biomechanical motion, analogous to a‘‘shear lap joint,’’ albeit one that is geometrically structured. Thiswork also provides an understanding of how T. marmorea achievesthe required balance between biomechanical mobility and protec-tion in the passive and defensive states. In the passive state, themorphometry of the individual shell plates and plate-to-plateinterconnections results in an approximately continuous curvatureand constant armor thickness and, hence, spatially homogeneousprotection; mobility is limited but armor coverage and protectionis maximized. When the animal is detached from a surface in thedefensive state, the underlying soft tissues of the foot gain greatercoverage and protection by the shell plates and the animal gainsmobility through tidal flow, but regions of vulnerability are openeddorsally, due to the increase in plate-to-plate separation distanceand decrease in plate-to-plate overlap. Lastly, experiments usingoptical microscopy, scanning electron microscopy, X-ray diffrac-tion, mercury porosimetry, fourier-transform infrared spectros-copy, and amino acid analysis explore the microstructure andspatial distribution of the six layers within intermediate shellplates, the role of multilayering in resisting predatory attacks,and the detection of chitin as a major component of the intra-plateorganic matrix and girdle.

Acknowledgments

We gratefully acknowledge support of the National ScienceFoundation MIT Center for Materials Science and Engineering(DMR-0819762), the US Army Research Office through the MIT Insti-tute for Soldier Nanotechnologies (Contract W911NF-07-D-0004),the National Security Science and Engineering Faculty FellowshipProgram (N00244-09-1-0064), and the MIT International Scienceand Technology Initiatives (MISTI-Israel). This work was also par-tially supported by the DFG (Grant No. EH 394/1-1). The authorswould like to thank MIT graduate students Samuel Crawford, ChrisNg, and Simon Choong for their assistance with photography, circlefitting algorithms, and mercury porosimetry, respectively. We alsothank E. Rosseeva for great technical assistance.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.jsb.2011.12.019.

References

Arciszewski, T., Cornell, J., 2006. Bio-inspiration: learning creative design principia.Lect. Notes Artif. Int. 4200, 32–53.

Arey, L.B., Crozier, W.J., 1919. The sensory responses of chiton. J. Exp. Zool. 29, 157–260.

Barthelat, F., Tang, H., Zavattieri, P.D., Li, C.-M., Espinosa, H.D., 2007. On themechanics of Mother-of-Pearl: a key feature in the material hierarchicalstructure. J. Mech. Phys. Solids 55, 306–337.

Baxter, J.M., Jones, A.M., 1981. Valve structure and growth in the ChitonLepidochitona cinereus (Polyplacophora: Ischnochitonidae). J. Mar. Biol. Ass.UK 61, 65–78.

Beedham, G.E., Trueman, E.R., 1967. The relationship of the mantle and shell of thepolyplacophora in comparison with that of other mollusca. J. Zool., Lond. 151,215–231.



Bergenhayn, J.R.M., 1930. Die Loricaten von Dr. Sixten Bocks Pazifik-Expedition1917–1918, mit spezieller Berücksichtigung der Perinotumbildungen und derSchalenstruktur. Göteborgs Kungliga Vetenskaps-och Vitterhets-SamhällesHandlingar Ser. B. 1, 1–52, pls. 1–3.

Bøggild, O.B., 1930. The Shell Structure of Mollusks. Det Kongelige DanskeVidenskabernes Selskabs Skrifter (Naturvidenskabelig og MathematiskAfdeling) 9, 235–326.

Bouxsein, M.L., Boyd, S.K., Christiansen, B.A., Guldberg, R.E., Jepsen, K.J., et al., 2010.Guidelines for assessment of bone microstructure in rodents using micro-computed tomography. J. Bone Miner. Res. 25, 1468–1486.

Bruet, B.J.F., Song, J., Boyce, M.C., Ortiz, C., 2008. Materials design principles ofancient fish armour. Nature Mater. 7, 748–756.

Brunner, E., Ehrlich, H., Schupp, P., Hedrich, R., Hunold, S., et al., 2009. Chitin-basedscaffolds are an integral part of the skeleton of the marine DemospongeIanthella basta. J. Struct. Biol. 168, 539–547.

Carter, J.G., Clark II, G.R., 1985. Classification and phylogenetic significance ofmolluscan shell microstructure. In: Bottjer, D.J., Hickman, C.S., Ward, P.D.,Broadhead, T.W. (Eds.), Mollusks: Notes for a Short Course. University ofTennessee Department of Geological Sciences Studies in Geology, Knoxville, TN,USA, pp. 50–71.

Carter, J.G., Hall, R.M., 1990. Polyplacophora, Scaphopoda, Archaeogastropoda andParagastropoda (Mollusca). In: Carter, J.G. (Ed.), Skeletal Biomineralization:Patterns, Processes and Evolutionary Trends, vol. 2, Atlas and Index, VanNostrand Reinhold, New York, pp. 29–31, 49, pls. 122–125

Chen, D., 2010. Microstructure and element composition of the shell plates ofLiolophura japonica Wuhan Univ. J. Nat. Sci. 15, 176–184.

Currey, J.D., Kohn, A.J., 1976. Fracture in the crossed-lamellar structure of Conusshells. J. Mater. Sci. 11, 1615–1623.

Currie, D.R., 1992. Aesthete channel morphology in three species of Australianchitons (Mollusca: Polyplacophora). J. Malac. Soc. Aust. 13, 3–14.

Eernisse, D.J., Reynolds, P.D., 1994. Polyplacophora. In: Harrison, F.W., Kohn, A.J.(Eds.), Microscopic Anatomy of Invertebrates, Volume 5. Wiley-Liss, New York,pp. 56–110.

Eernisse, D.J., 2007. Chitons. In: Denny, M.W., Gaines, S.D. (Eds.), Encyclopedia ofTidepools and Rocky Shores. University of California Press, Berkeley, California,pp. 127–133.

Ehrlich, H., Maldonado, M., Spindler, K.-D., Eckert, C., Hanke, T., et al., 2007. Firstevidence of chitin as a component of the skeletal fibers of marine sponges. PartI. Verongidae (Demospongia: Porifera). J. Exp. Zool. (Mol. Dev. Evol.) 308, 347–356.

Ehrlich, H., 2010a. Biological Materials of Marine Origin. Invertebrates. Springer-Verlag.

Ehrlich, H., 2010b. Chitin and collagen as universal and alternative templates inbiomineralization. Int. Geol. Rev. 52, 661–699.

Ehrlich, H., Ilan, M., Maldonado, M., Muricy, G., Bavestrello, G., et al., 2010. Threedimensional chitin-based scaffolds from verongida sponges (Demospongiae:Porifera). Part I. Isolation and identification of chitin. Int. J. Biol. Macromol. 47,132–140.

Furuhashi, T., Beran, A., Blazso, M., Czegeny, Z., Schwarzinger, C., et al., 2009.Pyrolysis GC/MS and IR spectroscopy in chitin analysis of molluscan shells.Biosci. Biotechnol. Biochem. 73, 93–103.

Gane, P.A.C., Ridgway, C.J., Lehtinen, E., Valiullin, R., Furo9, I., Schoelkopf, J.,Paulapuro, H., Daicic, J., 2004. Comparison of NMR cryoporometry, mercuryintrusion porosimetry, and DSC thermoporosimetry in characterizing pore sizedistributions of compressed finely ground calcium carbonate structures. Ind.Eng. Chem. Res. 43, 7920–7927.

Haas, W., 1972. Untersuchungenüber die Mikro- und Ultrastruktur derPolyplacophorenschale. Biomineralisation 5, 3–52.

Haas, W., 1976. Observations on the shell and mantle of the placophora. In: Watabe,N., Wilbur, K.M. (Eds.), The Mechanisms of Mineralization in the Invertebratesand Plants. University of South Carolina Press, Columbia, pp. 389–402.

Haas, W., 1981. Evolution of Calcareous Hardparts in Primitive Molluscs.Malacologia 21, 403–418.

Haszprunar, G., Wanninger, A., 2000. Molluscan muscle systems in developmentand evolution. J. Zool. Syst. Evol. Res. 38, 157–163.

Kaas, P., Van Belle, R.A., 1985. Monograph of Living Chitons (Mollusca:Polyplacophora), vol. 2, Suborder Ischnochitonina, Ischnochitonidae:Schizoplacinae, Callochitoninae and Lepidochitoninae. E.J. Brill, Leiden.

Kamat, S., Su, X., Ballarini, R., Heuer, A.H., 2000. Structural basis for the fracturetoughness of the shell of the conch Strombus gigas. Nature 405, 1036–1040.

Kamat, S., Kessler, H., Ballarini, R., Nassirou, M., Heuer, A.H., 2004. Fracturemechanisms of the Strombus gigas conch shell: II-micromechanics analyses ofmultiple cracking and large-scale crack bridging. Acta Mater. 52, 2395–2406.

Kreusch, B., 1981. Etude Comparée de la Structure et de la Composition Chimiquedes Formations Squelettiques chez les Mollusques Polyplacophores. Mémoirede Licence en Sciences Zoologiques, Faculté des Sciences, University de Liége, p.50.

Kuhn-Spearing, L.T., Kessler, H., Chateau, E., Ballarini, R., Heur, A.H., et al., 1996.Fracture mechanisms of the Strombus gigas conch shell: implications for thedesign of brittle laminates. J. Mater. Sci. 31, 6583–6594.

Laghi, G.F., Russo, F., 1978. Structtura ed architettura delle piastre di Chiton olivaceusSpengler (Polyplacophora, Mollusca). Bollettinodella Società PaleontologicaItaliana 17, 272–291.

Laghi, G.F., Russo, F., 1980. Struttura ed architettura delle piastre di Lepidopleuruscajetanus (Poli) (Mollusca: Polyplacophora). Ann. Univ. Ferrara Sezione IX Sci.Geol. Paleontol. 6, 321–338.

Langer, P.D., 1978. Some aspects of the biology of three Northwestern AtlanticChitons: Tonicella rubra, Tonicella marmorea, and Ischnochiton albus (Mollusca:Polyplacophora). Ph.D. Thesis, Department of Zoology, University of NewHampshire, p. 172.

Neves, N.M., Mano, J.F., 2005. Structure/mechanical behavior relationships incrossed-lamellar sea shells. Mater. Sci. Eng. C. 25, 113–118.

Ortiz, C., Boyce, M.C., 2008. Materials science – Bioinspired structural materials.Science 319, 1053–1054.

Pearson, D.M., 1981. Functional aspects of the integument in polypterid fishes. Zoo.J. Linn. Soc. 72, 93–106.

Poulicek, E., Kreusch, B., 1983. Evolutionary trends in the skeletal structures of thePolyplacophora. Proceedings of the Eighth International Malacological Congress(Budapest, Hungary), 207–212.

Pratt, V., 1987. Direct least-squares fitting of algebraic surfaces. Computer Graphics21, 145–152.

Puchalski, S., Eernisse, D.J., Johnson, C.C., 2008. The effect of sampling bias on thefossil record of chitons (Mollusca, Polyplacophora). Am. Malac. Bull. 25, 87–95.

Rasband, W.S., 1997–2011. ImageJ, US National Institutes of Health, Bethesda,Maryland, USA. http://www.imagej.nih.gov/ij/.

Runnegar, B., Pojeta, J., Taylor, M.E., Collins, D., 1979. New species of the Cambrianand Ordovician chitons Matthevia and Chelodes from Wisconsin andQueensland: evidence for the early history of polyplacophoran mollusks. J.Paleont. 53, 1374–1394.

Schwabe, E., 2005. A catalogue of recent and fossil chitons (Mollusca:Polyplacophora) addenda. Novapex 6, 89–105.

Schwabe, E., 2010. Illustrated summary of chiton terminology. Spixiana 33, 171–194.

Sigwart, J.D., 2009. Morphological cladistic analysis as a model for characterevaluation in primitive living chitons (Polyplacophora, Lepidopleurina). Am.Malac. Bull. 27, 95–104.

Song, J., Reichert, S., Kallai, I., Gazit, D., Wund, M., et al., 2010. Quantitativemicrostructural studies of the armor of the marine threespine stickleback(Gasterosteus aculeatus). J. Struct. Biol. 171, 318–331.

Stebbins, T.D., Eernisse, D.J., 2009. Chitons (Mollusca: Polyplacophora) known frombenthic monitoring programs in the Southern California Bight. The Festivus 41,53–100.

Takahashi, K., 1967. The ball-and-socket joint of the sea-urchin spine: geometry andits functional implications. J. Fac. Sci. Univ. Tokyo. Sec. IV 11, 131–135.

Taylor, J.D., Layman, M., 1972. The mechanical properties of bivalve (Mollusca) shellstructures. J. Paleontol. 15, 73–87.

Taylor, J.D., Reid, D.G., 1990. Shell microstructure and mineralogy of theLittorinidae: ecological and evolutionary significance. Hydrobiologia 193,199–215.

Telford, M., 1985. Domes, arches, and urchins: the skeletal architecture of echnoids(Echinodermata). Zoomorphology 105, 114–124.

Town, J.C., 1979. Aspects of the biology of Astrostole scabra (Hutton, 1872). Ph.D.Thesis, Department of Zoology, University of Canterbury, p. 200.

Treves, K., Traub, W., Weiner, S., Addadi, L., 2003. Aragonite formation in the chiton(Mollusca) girdle. Helvetica Chimica Acta 86, 1101–1112.

Vendrasco, M.J., Fernandez, C.Z., Eernisse, D.J., Runnegar, B., 2008. Aesthete canalmorphology in the Mopaliidae (Polyplacophora). Am. Malac. Bull 25, 51–69.

Vermeij, G.J., 1993. A Natural History of Shells. Princeton University Press,Princteon, NJ.

Walter, L.M., Morse, J.W., 1984. Reactive surface area of skeletal carbonates duringdissolution: effect of grain size. J. Sediment. Petrol. 54, 1081–1090.

Wang, L., Song, J., Ortiz, C., Boyce, M.C., 2009. Anisotropic design of a multilayeredbiological exoskeleton. J. Mater. Res. 24, 3477–3494.

Wilmot, N.V., Barber, D.J., Taylor, J.D., Graham, A.L., 1992. Electron microscopy ofmolluscan crossed-lamellar microstructure. Phil. Trans. R. Soc. Lond. Biol. 337,21–35.

Wingstrand, K.G., 1985. On the anatomy and relationships of recentMonoplacophora. Galathea Rep. 16, 7–94.

Yao, H., Dao, M., Imholt, T., Huang, J.M., Wheeler, K., et al., 2010. Protectionmechanisms of the iron-plated armor of a deep-sea hydrothermal ventGastropod. Proc. Natl. Acad. Sci. 107, 987–992.

Zaremba, C.M., Morse, D.E., Mann, S., Hansma, P.K., Stucky, G.D., 1998. Aragonite-hydroxyapatite conversion in Gastropod (Abalone) Nacre. Chem. Mater. 10,3813–3824.

http://www.imagej.nih.gov/ij/

Documents

Journal of Structural Biology - MITweb.mit.edu/cortiz/www/ConnorsM_JStrBio2012.pdf · 2012-02-15 · bInstitute of Bioanalytical Chemistry, Dresden University of Technology, 01069