both units in three dimensions and at differ-
ent developmental stages. Tomograms of ICCs
at different growth stages were collected in a
scanning transmission electron microscope
(STEM), using the high-angle annular dark-field
(HAADF) detector, and reconstructed into 3D
volumes (figs. S5 to S7) ( 24 , 25 ). 3D analysis
of a coccolith at an early developmental stage
showed that all crystal units expose flat crys-
tallographic facets (Fig. 2). The dihedral angles
between these surfaces and the angles between
their edges are in agreement with those known
for the {104} calcite rhombohedron ( 26 ), which
suggests that only these stable crystallographic
facets are expressed (fig. S8).
We observed that R-units are situated on
their acute edges, aligned along the circum-
ference of the coccolith ring (Fig. 2A)—an
arrangement that is in agreement with ob-
servations in other species ( 15 , 27 ). This is
interesting for two reasons: (i) As a result of
geometrical considerations, and in contrast
to the conventional V/R model, aligning such
{104} rhombohedra on their acute edges en-
forces a subradial orientation of the crystals’c
axes, breaking radial symmetry and conveying
chirality to the emergent structure (Fig. 2A, cyan
arrow); (ii) it challenges the concept of epitaxy,
as the crystals should have a facet, and not an
edge, parallel to the nucleating surface (i.e., the
base plate). Although less clear in the V-units
(initial crystals appear less rhombohedral),
we also see that the crystals possess a sub-
vertical tilt of theircaxes—a result of orient-
ing the rhombohedra on their obtuse edge
(Fig. 2B and fig. S9). Given that our data lackinformation on the mechanisms that lead to
this regulated crystal orientation, the role of
the base plate as a nucleating surface remains
an open question.
To relate the morphological information to
the crystallography of the crystals, we analyzed
adjacent R-units from an ICC with annular dark-
field (ADF) STEM coupled with scanning nano-
beam electron diffraction (NBED) ( 28 ), which
collects a diffraction pattern from every point
the beam raster traverses. The analysis con-
firmed a relative tilt between the units, as well
as the subradial deviation of eachcaxis rela-
tive to the coccolith circumference (Fig. 2, C to
E). These analyses of early-stage ICCs allow us
to refine the“classical”V/R model, which cen-
ters onc-axis directions and is achiral, to a more
accurate crystallographic representation of crystalSCIENCEscience.org 15 APRIL 2022•VOL 376 ISSUE 6590 313
Fig. 1. Overview of crystal morphogenesis in developing coccoliths.
(A)AC. leptoporuscell with a complete coccolith shell (top) and a schematic
cross section of a mature coccolith (bottom). Here and in the remaining
figures, the V-units, forming the distal shield, are in orange, and the R-units,
forming the proximal shield, are in blue. Dashed lines indicate curved surfaces,
continuous lines indicate flat surfaces, and a red line indicates the putative
position of the base plate ( 13 ). (BtoL) ICCs at four growth stages. Three different
views highlight the architecture at each stage. Insets in (E) and (I) show a
magnification of the dashed areas. Arrowheads indicate flat (purple) and curved
(green) surfaces. The dark section in the schematic timeline at the bottom
represents the morphological spectrum of ICCs analyzed in the following figures.
Scale bars, 500 nm (insets, 200 nm).RESEARCH | REPORTS