rhombohedra resting on their acute/obtuse
edges. This view consolidates two crystallo-
graphic features into one basic architecture, in
which both the tiltedcaxes and the chiral ultra-
structure originate from the initial positioning
of the crystals.
To understand the ways by which coccolith
crystals grow and interlock, we analyzed the
morphology of individual crystal units in detail.
Five coccoliths, reflecting the stages along
coccolith growth that are suitable for tomog-
raphy (see Fig. 1, bottom bar), were partially
segmented (Fig. 3 and movies S1 to S5). The
derived“timeline”revealed several key aspects:
(i) Both unit types demonstrate a transition
from relatively isotropic rhombohedra to ma-
ture anisotropic crystals (Fig. 3, A and B); (ii)
both unit types feature facets that appear crys-
tallographic throughout their growth, while
some areas—the stem region, the proximal sides
of the shields, and the interfaces between neigh-
boring crystals—maintain a curved morphol-
ogy (Fig. 3C); (iii) the dihedral angles between
crystallographic-like facets throughout crystal
growth all correspond to the {104} habit (fig. S8).
This timeline demonstrates how the equivalent
{104} facets of initial crystals develop in an
anisotropic manner, giving rise to mature
{104} facets with very different sizes (Fig. 3,
D and E). These observations show that the
complex morphology of the crystals is not
the result of various types of crystallographic
planes, but rather of differential growth of
the chemically equivalent {104} facets.
The observations that the crystals are grow-
ing with the expression of {104} facets exclu-
sively (Fig. 3), and that these {104} facets grow
at different speeds, raise a critical question
regarding the factors responsible for this
symmetry breaking. This conundrum arises
from the symmetry and chemical equivalence
of all six {104} facets, such that no single facet
possesses an inherent growth rate that differs
from the others (i.e., calcium and carbonate ions
should display no association or dissociation
bias toward any specific {104} facet) ( 26 , 29 ).
To understand how the anisotropy of these
chemically equivalent facets emerges, we ana-
lyzed growth patterns of specific facets. Two
distinct patterns were observed: (i) differentialgrowth of symmetry-related facet pairs of an
individual crystal unit [e.g.,ðÞ 114 and (104)
(Fig. 4A and fig. S10)], where one facet grows
faster than its opposite and/or adjacent facets,
resulting in an anisotropic motif; and (ii) dif-
ferential growth of facets from two different
unit types (R and V) facing the same envi-
ronment (Fig. 4B). In the latter case, facets
first appear level with one another, yet end
up with the V-units repeatedly outgrowing the
R-units (Fig. 4B, compare insets). Both examples
show two chemically equivalent facets that, for
some reason, differ in their growth rates.
Within a homogeneous solution, anisotropic
growth of equivalent crystallographic facets
is incompatible with their identical growth
kinetics. However, on the atomic scale, cal-
cite growth proceeds via both acute and obtuse
steps, each having different growth kinetics
( 29 – 31 ). Therefore, nanoscale inhomogeneity
in the environment in which the crystals grow
can result in growth anisotropy. It was shown
in several coccolithophore species that crystal-
lization occurs at extreme confinement, where
only tens of nanometers separate the crystals314 15 APRIL 2022•VOL 376 ISSUE 6590 science.orgSCIENCE
Fig. 2. Alignment of crystal edges at the coccolith circumference results
in coccolith chirality.(A) 3D rendering of an early-stage ICC, viewed from the
proximal side. Yellow lines outline the acute edges on which the R-units are
situated along the circumference of the base plate. Thecaxes of the R-units
(red arrows) and their subradial tilts (note deviations from the lines illustrating
radial directions) are shown, as well as the emerging ultrastructural chirality
(cyan arrow). (B) Side view of the coccolith in (A), showing the subvertical tilt in
caxes of the V-units (angles between the vertical lines and red arrows). Inset
shows schematic {104} rhombohedra with crystallographic annotations, situated
on obtuse (R-unit) or acute (V-unit) edges. (C) STEM ADF image of an ICC
at a stage similar to that in (A); colored crystals were analyzed by NBED, and
the derivedc-axis orientations [in (E)] are indicated as colored arrows.
(D) NBED patterns associated with the crystal units marked in (C). Diffraction
scale bars, 5 nmÐ^1 .(E) The relative orientation of the three R-units in (C)
is shown in a stereographic projection of the (104) and (001) poles to the
coccolith plane.RESEARCH | REPORTS