ArticlePDF AvailableLiterature Review

Modularity is the mother of invention: a review of polymorphism in bryozoans: Modularity is the mother of invention



Modularity is a fundamental concept in biology. Most taxa within the colonial invertebrate phylum Bryozoa have achieved division of labour through the development of specialized modules (polymorphs), and this group is perhaps the most outstanding exemplar of the phenomenon. We provide a comprehensive description of the diversity, morphology and function of these polymorphs and the significance of modularity to the evolutionary success of the phylum, which has >21000 described fossil and living species. Modular diversity likely arose from heterogeneous microenvironmental conditions, and cormidia (repeated clusters of associated modules) are an emergent property of the cue thresholds governing zooid plasticity. Polymorphs in a colony have, during phylogeny, transitioned into associated non‐zooidal structures (appendages), increasing colonial integration. While the level of module compartmentalization is important for the evolution of bryozoan polymorphism, it may be less influential for other colonial invertebrates.
Biol. Rev. (2018), pp. 000000. 1
doi: 10.1111/brv.12478
Modularity is the mother of invention: a
review of polymorphism in bryozoans
Carolann R. Schack1,2, Dennis P. Gordon2and Ken G. Ryan1
1School of Biological Sciences, Victoria University of Wellington, PO Box 600, Wellington, 6140, New Zealand
2National Institute of Water & Atmospheric Research, Private Bag 14901, Kilbirnie, Wellington, 6021, New Zealand
Modularity is a fundamental concept in biology. Most taxa within the colonial invertebrate phylum Bryozoa have
achieved division of labour through the development of specialized modules (polymorphs), and this group is perhaps the
most outstanding exemplar of the phenomenon. We provide a comprehensive description of the diversity, morphology
and function of these polymorphs and the significance of modularity to the evolutionary success of the phylum, which
has >21000 described fossil and living species. Modular diversity likely arose from heterogeneous microenvironmental
conditions, and cormidia (repeated clusters of associated modules) are an emergent property of the cue thresholds
governing zooid plasticity. Polymorphs in a colony have, during phylogeny, transitioned into associated non-zooidal
structures (appendages), increasing colonial integration. While the level of module compartmentalization is important
for the evolution of bryozoan polymorphism, it may be less influential for other colonial invertebrates.
Key words: modularity, polymorphism, Ctenostomata, Cheilostomata, Cyclostomata, avicularia, kenozooids, ovicells,
I. Introduction .............................................................................................. 2
(1) Bryozoan taxonomy .................................................................................. 2
(2) Autozooids ........................................................................................... 2
II. Polymorphism ............................................................................................ 4
(1) Polymorphism classification .......................................................................... 5
III. Mandibulate polymorphs ................................................................................. 6
(1) Avicularia ............................................................................................ 6
(a) B-zooids .......................................................................................... 10
(b) Evolution of avicularia ............................................................................ 11
(2) Eleozooids and aviculomorphs ....................................................................... 11
IV. Kenozooids, autozooidal appendages and extrazooidal structures ....................................... 12
(1) Space-filling and strengthening structures ............................................................ 13
(2) Stolons ............................................................................................... 13
(3) Rhizoids .............................................................................................. 15
(a) Articulated colonies ............................................................................... 15
(4) Spines ................................................................................................ 16
(a) Gymnolaemate spines ............................................................................ 16
(b) Stenolaemate spines .............................................................................. 18
(5) Frontal walls .......................................................................................... 19
V. Reproductive polymorphism ............................................................................. 19
(1) Gonochoristic zooids ................................................................................. 19
(2) Cheilostome brood chambers ........................................................................ 19
(a) Evolution of brood chambers ..................................................................... 20
* Address for correspondence (Tel: +64 4 386 0909; E-mail:
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
2Carolann R. Schack and others
(3) Ctenostome brood chambers ......................................................................... 21
(4) Cyclostome gonozooids .............................................................................. 22
(5) Brood chambers in other stenolaemates .............................................................. 24
VI. Asexual propagules ....................................................................................... 24
(1) Evolution of asexual propagules ...................................................................... 25
VII. Polymorphs of unknown function ........................................................................ 25
(1) Nanozooids ........................................................................................... 25
(2) Zooeciules ............................................................................................ 26
VIII. Development of polymorphism ........................................................................... 27
(1) Origin and evolution of polymorphism ............................................................... 27
(2) The cormidium: zooid assembly rules ................................................................ 28
(3) Incidence of polymorphism and degrees of modularity ............................................... 30
(4) Types of polymorphs ................................................................................. 33
IX. Summary ................................................................................................. 33
X. Conclusions .............................................................................................. 34
XI. Acknowledgments ........................................................................................ 34
XII. References ................................................................................................ 34
The high degree of modularity in the organization of many
colonial invertebrates is key to their evolutionary success.
Modules experience separate selection pressures within
the same organism (Kirschner & Gerhart, 1998; Carroll,
2001). The dissociative nature of modular organization
enhances adaptive potential by reducing the impact of
deleterious mutations (Kirschner & Gerhart, 1998; Carroll,
2001). Module redundancy also enhances adaptive potential,
allowing novel forms to arise in non-essential modules
(Kirschner & Gerhart, 1998). This is like the International
Space Station: each room (module) is separated by a sealed
door, which prevents breaches (deleterious mutations) from
damaging the rest of the station. The station may be
reorganized through the addition, removal, or modification
of modules (evolution). The adaptive potential provided
through modular organization may be important for
overcoming the ‘cost’ of complexity. In most organisms,
as complexity increases, the ability to evolve toward optimal
characters is reduced owing to the decreasing impact of a
single mutation (Orr, 2000; Carroll, 2001). However, the
presence of non-modular or slightly modular organisms (e.g.
arthropods: integrated individuals with serially homologous
limbs (modules) capable of differentiation) and the secondary
integration of modules (Krohs, 2009) indicates that
modularity does not universally increase adaptive potential.
Any change to a module with multiple functions must be
compatible with those functions, which limits variation and
thus adaptive potential (Krohs, 2009).
The selection pressures, developmental changes, and
canalization contributing to module differentiation as well
as the origin of modularity itself – require further study.
Bryozoans are highly modular organisms and exhibit division
of labour through a wide variety of polymorphs. The
functions, development, and evolution of many bryozoan
polymorphs are still poorly understood, but they may provide
insight into independently developed modularity and module
differentiation in different classes.
en (1977) published his review of bryozoan
polymorphism 41 years ago. Much has been done since
then and our aim is to provide an update to bryozoan
polymorphism in the context of modularity. Our companion
paper (Schack, Gordon & Ryan, 2018), provides a
classification system for cheilostome polymorphism to
facilitate morphological analyses. An appendix of terms is
provided in Table 1.
(1) Bryozoan taxonomy
Phylum Bryozoa comprises three classes: Phylactolaemata,
Stenolaemata, and Gymnolaemata. The phylactolaemates
are exclusively found in fresh water and are unmineralized,
while the opposite is true for the stenolaemates. Class
Stenolaemata dominated until the Late Cretaceous, when it
was overtaken by the radiation of gymnolaemates (McKinney
& Jackson, 1989). Today, the gymnolaemates are extremely
diverse and occur mostly in marine environments. They are
divided into two orders – the non-calcified Ctenostomata,
which are paraphyletic and the calcified, highly abundant
Cheilostomata (McKinney & Jackson, 1989; Mukai, Kiyoshi
& Reed, 1997). Since phylactolaemates lack polymorphism,
this review will focus on the stenolaemates, ctenostomes, and
the highly polymorphic cheilostomes.
(2) Autozooids
The basic (primary) module of a bryozoan colony is the
self-supporting autozooid (Sil´
en, 1977; McKinney & Jackson,
1989), which consists of a cystid and polypide (Fig. 1). The
cystid (body wall) is partially deformable and is composed
of a secreted outer layer (ectocyst, generally calcified in
cheilostomes and cyclostomes) and an inner cellular lining
(endocyst). Enclosed by the cystid is the polypide – which
includes the ciliated tentacle crown (‘lophophore’), the
U-shaped digestive tract, the intrinsic muscular system, and
parts of the nervous system (McKinney & Jackson, 1989;
Mukai et al., 1997).
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 3
Table 1. Appendix of terms.
Adventitious avicularia: avicularia that are borne on the frontal side of the autozooid. Can be sessile or pedunculated.
Alveoli: hollow structures in cyclostomes that develop between autozooids, autozooidal rows, or in association with gonozooids.
Walls have a similar structure to autozooids and share communication pores with neighbouring autozooids. Typically described
as extrazooidal but may be kenozooidal. See also: Cancelli,Dactylethrae,Kenozooid ,Mesozooid,Metapore,Nematopore,Tergopore.
Ancestrula: the first zooid in a colony, resulting from the metamorphosis and settlement of a bryozoan larva. Usually a single zooid
but may be a complex of multiple zooids.
Anascan: a cheilostome bryozoan lacking an ascus.
Appendage: a projection from the cystid of an autozooid that is not separated from the coelom by a pore plate. See also: Autozooidal
Ascus: a compensation sac/space that allows ascophoran bryozoans to evert their tentacle crowns via hydrostatic displacement.
Ascophoran: a cheilostome bryozoan possessing an ascus or compensation space.
Autozooid: the main unit of a bryozoan colony, capable of feeding and other functions necessary for life.
Autozooidal polymorph: zooid that retains a protrusible tentacle crown (which may or may not feed) but differs from an autozooid in
the form of its cystid, polypide, or both. This includes reproductive zooids, appendaged autozooids, and nanozooids.
Avicularia: a heterozooid with a reduced, non-feeding polypide. Its operculum is enlarged into a mandible. These structures have
mechanosensory capabilities and are thought to have a mainly defensive function. See also: B-zooid, Vibraculum.
B-zooid: a heterozooid with a feeding polypide and an enlarged operculum. Thought to be an evolutionary precursor to
avicularia. Found in some Steginoporellidae and Calloporidae. See also: Avicularia.
Cancelli: an ambiguous term for structures in Cancellata (Cyclostomata) that may be vacuoles, extrazooidal skeleton lacking
skeletal walls, or a kenozooidal chamber. See also: Alveoli,Dactylethrae,Kenozooid ,Mesozooid,Metapore,Nematopore,Tergopore.
Communication pore: a hole in interzooidal walls that allows zooid-to-zooid communication. Communication may be direct or via a
porecell complex (rosette).
Cormidium: a ‘colony within a colony’ that consists of a feeding zooid and associated polymorphs. The cormidium is capable of
performing most functions vital for life. Multiple types of cormidia may be present within a colony.
Cheilostome: bryozoans belonging to the gymnolaemate order Cheilostomata.
Cribrimorph:seeFrontal shield.
Cystid: zooidal body wall, whether calcified or uncalcified, enclosing the coelomic cavity and housing a polypide (when present).
The cystid wall is composed of a secreted outer layer (ectocyst) and an inner cellular lining (endocyst).
Dactylethrae: an ambiguous term for vacuoles in stenolaemate colonies that are closed by a terminal diaphragm which may be
extrazooidal or kenozooidal. See also Alveoli,Cancelli,Kenozooid ,Mesozooid ,Metapore,Nematopore,Tergopore.
Distal: in the direction of colony growth (away from the ancestrula and origin of growth).
Encrusting: a mode of growth where the colony grows along the surface of the substratum.
Erect: a mode of growth where the colony is held away from the substratum.
Exilazooid: a vacuole with internal flask-shaped structures. May be an extrazooidal structure or zooid with a reduced polypide.
Extrazooidal structures: structures that are external to zooidal boundaries. See also: Mesopore.
Frontal: the side of a zooid bearing the orifice (the hole through which the tentacle crown is protruded).
Frontal shield: a calcified frontal surface in cheilostomes. This can constitute calcification of the frontal wall itself (gymnocystal), or a
protective mesh of fused spines (spinocystal), a calcified wall arched over the primary membranous frontal wall (umbonuloid), or
a calcified wall beneath the membranous frontal wall but above the ascus (lepralioid).
Funiculular system: branching semitubular to tubular system of mesenchimatous cords, connecting the gut of the polypide with
communication pores, facilitating transfer of nutrients throughout the colony. Present in gymnolaemates.
Funiculus: tubular muscular cord connecting the stomach and body wall (rarely, communication pore) in Bryozoa. See also:
Funicular system.
Gonozooid: a voluminous brood chamber present in cyclostomes, formed through the modification of a female autozooid (which
may involve extrazooidal structures, as in Lichenoporidae). Consists of an enlarged brood cavity (site of embryo incubation) and
an ooeciostome (aperture). See also: Ovicell.
Heterozooid: a specialized zooid with a non-feeding, vestigial or absent polypide.
Interzooidal avicularia: avicularia occurring between zooids, typically smaller than an autozooid and resting on the substratum.
Kenozooid: a heterozooid that lacks a polypide. Kenozooids fulfill many functions within the colony, including space-filling,
attachment, and colony support. See also: Rhizoid,Vicariozooid .
Lepralioid: see Frontal shield.
Membranous frontal wall: the frontal surface of a bryozoan cystid. In anascan bryozoans it is flexible enough to be displaced by the
parietal muscles, allowing the tentacle crown to be everted.
Mesozooid:inPaleozoic stenolaemates, a tapering chamber closed by a terminal diaphragm, potentially containing diaphragms or cysts.
May be extrazooidal or kenozooidal. See also: Alveoli,Cancelli,Dactylethrae,Kenozooid,Metapore,Nematopore,Tergopore.
Metapores: slender, deep cavities in the extrazooidal skeleton of stenolaemates that lack skeletal walls of their own. See also: Alveoli,
Cancelli, Dactylethrae, Kenozooid, Mesozooid, Nematopore, Tergopore.
Module: a repeated unit within an organism. Primary modules can be combined to form higher-level structures.
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
4Carolann R. Schack and others
Table 1. Continued
Nematopore: hollow structures with skeletal walls that have been incorporated into the extrazooidal skeleton of stenolaemate colonies. Also
called firmatopores. May have a thread-like or squat morphology and are typically interpreted as kenozooidal. See also: Alveoli, Cancelli,
Dactylethrae, Kenozooid, Mesozooid, Metapore, Tergopore.
Ooecium: two-layered calcified part of an ovicell, comprising an outer ectooecium (partly or mostly membranous in some taxa) and inner
endoooecium, with a coelomic lumen (sometimes obliterated) between the two layers that is connected to coelom of the originating
Ovicell: a specialized brood chamber found in cheilostomes, consisting of a calcifed ooecium and brood cavity, the entrance to which may
be closed by the ooecial vesicle and/or the operculum of the maternal zooid. A multi-zooidal complex. See also Ooecium,Gonozooid.
Polymorphism: variation in zooid morphology, including heterozooids and autozooidal polymorphs. See also Autozooidal polymorph,
Polypide: soft body of the bryozoan, principally the tentacle crown and its sheath, gut, and intrinsic musculature and nervous systems.
Pore plate: an interzooidal wall element with one to several communication pores.
Proximal: in the direction of the ancestrula (away from the growing edge, towards the origin of growth).
Rhizoid: an elongated kenozooid typically used for attachment to the substratum or branch support in erect colonies.
Stolons: structures used to support the colony off the substratum or used in overgrowth competition.
Spine: a polymorphic autozooidal module, either an appendage of an autozooid or a kenozooid. Typically slender, hollow, and with or
without a cuticular joint at its point of attachment. See also: Appendage.
Tergopore: hollow structures with similar walls to autozooids and interzooidal communication pores. Occur in stenolaemates. May be
kenozooidal. See also: Alveoli,Cancelli,Dactylethrae,Kenozooid,Mesozooid ,Metapore,Nematopore.
Umbonuloid:seeFrontal shield .
Vibraculum: an extreme form of an adventitious avicularium, in which the operculum is elongated as a thin setiform mandible.
Vicarious avicularia: avicularia that take the place of an autozooid in the budding sequence. Usually of a similar size to autozooids.
Vicariozooid: a kenozooid replaces an autozooid and is capable of budding. Used to increase colony rigidity and margin robustness.
The interzooidal walls of both gymnolaemate and
stenolaemate zooids are perforated by communication
pores (often grouped into pore plates). These pores allow
free communication between the exosaccal coeloms of
neighbouring stenolaemate zooids (although some may
become plugged by epithelial cells or covered by a calcareous
plate), while the communication pores of gymnolaemates are
each sealed by a cellular plug (or rosette; Mukai et al., 1997).
In all taxa, intrazooidal nutrient transfer may be
facilitated by the funiculus, a fluid-filled semitubular cord
that connects the stomach caecum to the body wall
(Ryland, 1979; Mukai et al., 1997). Since developing zooids
at the edge of a bryozoan colony lack fully formed
polypides, nutrients must be shared among individuals. In
stenolaemates, interzooidal nutrient transfer is achieved in
one of two ways: (i) diffusion or pressure changes through
communication pores; or (ii) exchange through confluent,
extra-skeletal spaces surrounded by a membranous wall
(Ryland, 1979; Boardman, 1998). Interzooidal nutrient
transport in gymnolaemates is facilitated by the funicular
system, a branched network associated with communication
pores that is secondarily evolved from the funiculus. In both
taxa, nutrients dissolved in coelomic fluid are transported
from actively feeding areas to the closest growing edge
(Miles et al., 1995 and references therein; Mukai et al., 1997).
Non-feeding modules are resourced in the same way.
In bryozoans, a distinction must be made between
sources of continuous variation (astogenetic, ontogenetic,
developmental noise) and polymorphism. The first zooid
in a colony (ancestrula) often has a different form from
later zooids, and intermediate forms exist between them.
Astogenetic variation occurs during maturation of the
colony, and ontogenetic variation occurs during maturation
of zooids in the same astogenetic zone. Autozooids may
also vary within the same astogenetic and ontogenetic zone
because of developmental noise or plasticity (Jebram, 1978).
Developmental noise is intrinsic variance in developmental
processes (Debat & David, 2001) and can result in slight
variation in autozooid morphology or ‘abnormal’ zooids
(e.g. double polypides; Jebram, 1978). Polymorphism is
typically defined as discontinuous variation in form (i.e.
no intermediate forms are present between polymorphic and
normal zooids). Therefore, slight gradations in autozooid
morphology are not considered polymorphic. Indeed, these
sources of variation are also present in polymorphic
Plasticity, defined as the ability to express different
phenotypes in response to environmental conditions, may
also produce continuous or discrete variation in zooid
morphology. Continuous plastic variation can occur in
both regular autozooids and polymorphic zooids. For
example, plasticity in zooids allows cheilostomes to form
abnormally shaped zooids where zooid growth is constrained
by some barrier (Jebram, 1978). Similarly, spinule length
in Membranipora membranacea increases with increasing
concentration of chemical cues from a nudibranch predator
(Harvell, 1998). However, it is only when plastic responses
produce discontinuous variation that they can be considered
polymorphic. Polymorphs may contribute to the division
of labour within a colony (Harvell, 1994; Lidgard et al.,
2012), which provides a further distinction between true
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 5
(A) (C)
Fig. 1. Simplified zooidal anatomy of (A) an anascan cheilostome, (B) a stoloniferan ctenostome, and (C), a fixed-walled cyclostome.
Reproductive organs and musculature not depicted. a, anus; bw, basal wall; co, collar; cp, communication pore; cw, cystid wall; f,
funicular cord; fm, membranous part of frontal wall; fw, frontal wall; gyc, gymnocyst (calcified part of frontal wall); izp, interzooidal
pores; ms, membranous sac; op, operculum; or, orifice; ph, pharynx; pp, pore plate (hosting a communication pore); re, rectum; se,
septum; st, stomach; t, tentacle; tm, terminal membrane; ts, tentacle sheath. A and C redrawn from Ryland (1970), B from Nielsen
& Pedersen (1979).
polymorphism and continuous plasticity (where differences
in function type are non-existent).
It is also important to consider when zooids become
polymorphic. Consistent polymorphs are zooids that
begin as polymorphs and remain polymorphic throughout
their lifespan (e.g. avicularia). Reversible polymorphs are
temporary modifications to autozooids, while irreversible
polymorphs begin as autozooids but are permanently
transformed into polymorphic zooids. Both may be tied
to ontogeny and the polypide degenerationregeneration
cycle. While irreversible and reversible polymorphs can be
discontinuous plastic responses, it is unclear if consistent
polymorphs are generated in response to environmental
conditions. Abiotic macroenvironmental conditions do not
influence the number and type of consistent polymorphs
produced (Hughes & Jackson, 1990; Simpson, Jackson
& Herrera-Cubilla, 2017). However, microenvironmental
variations within a colony and predatory epibionts (Lidgard,
2008; Lidgard et al., 2012) may induce the production of
consistent polymorphs.
(1) Polymorphism classification
Resource sharing within a colony (McKinney & Jackson,
1989) permits the existence of non-feeding polymorphs
(and developing zooid buds), although some polymorphic
zooids are capable of feeding. Polymorphic zooids are
distinguished from regular autozooids by modifications to
their cystid, polypide, or both (Fig. 2). The cystid may be
modified without change to the polypide and vice versa.
However, they are integrated structures that are not fully
independent: dwarf or irregularly shaped cystids possess
reduced or aborted polypides. In phylactolaemates and
cyclostome stenolaemates the polypide is formed before
the cystid, while in gymnolaemates the cystid is formed first
(Mukai et al., 1997). This suggests that the cystid exerts control
over polypide formation in the Gymnolaemata (the zooecium
size hypothesis; Harvell, 1994), while the polypide (and its
possible absence) exerts control on cystid formation in other
bryozoan taxa. The zooecium size hypothesis also operates
in the other direction: abnormally large polypides (conjoined
or double polypides) have been found only in enlarged
cystids (Jebram, 1978; Harvell, 1994), and ‘macrozooids’
in Trepostomata may have contained larger polypides
(Boardman & Buttler, 2005).
Despite high structural and physiological integration, the
cystid and polypide also exhibit a degree of modularity
in polymorphic zooids and can be considered ‘zooidal
semimodules’. Polymorphic zooids can therefore be classified
according to modifications of their semimodules (Table 2).
Extrazooidal structures, like polymorphs, are derived
phyletically from autozooids, so the distinction between
extrazooidal and zooidal may be unnecessary (P.D. Taylor,
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
6Carolann R. Schack and others
(A) (B)
200 µm 200 µm
Fig. 2. A monomorphic colony (A, Membranipora pura) and a polymorphic colony (B, Selenaria punctata). Av, avicularium; Az,
autozooid; F, female zooid with ooecium; M, male zooid. A by D.P.G., B by P.E. Bock.
personal communication). However, the distinction is useful
for determining structural homology and the evolutionary
pathways used to generate similar structures. Distinguishing
between different forms is difficult, so polymorphs will be
discussed in terms of their unifying similarities to facilitate
comparison among polymorph types and bryozoan taxa.
These unifying similarities group polymorph types based on
structure (although not implying homology) and probable
The resources required for the construction and support
of polymorphs must be offset by their function. As suspension
feeders, bryozoans maintain an energetically marginal exis-
tence. Therefore, any adaptation that provides even a small
increase in colony efficiency can significantly influence sur-
vival and competitive ability (Thorpe, 1979; Lidgard, 1985).
Polymorphic zooids may be arranged with regular
autozooids in a cormidium (a colony within a colony;
Beklemishev, 1969; Ryland, 1979; Haeckel, 1888; Section
VIII.2). The vital functions a cormidium can perform
are determined by its submodules, and the number of
each submodule type should influence the magnitude of
those functions (Schack et al., 2018). Submodules include
autozooids, autozooidal polymorphs (i.e. autozooids with
cystid modifications), heterozooids, multizooidal complexes
(Schack et al., 2018), or extrazooidal structures. Although
they can be part of a cormidium, non-feeding groups of
heterozooids and extrazooidal structures cannot constitute
a cormidium on their own (i.e. they cannot fulfil the
basic functions vital for life). These ‘paramodules’ can
occur at any level of modular organization and need
not be iterated within the colony (e.g. a kenozooidal
attachment stalk).
(1) Avicularia
Cheilostome avicularia are highly modified zooids: their
reduced polypide retains a sensory and potentially secretory
function, but can no longer feed (Mukai et al., 1997; Carter,
Gordon & Gardner, 2008). Instead, energy is likely obtained
from the funicular system (Lutaud, 1983; Carter et al.,
2008; Carter, Gordon & Gardner, 2010a). The operculum
is enlarged into a mandible while the cystid is distally
stretched and tapered (Fig. 3; Mukai et al., 1997). The main
types are: (i) vicarious avicularia, generally equal (but can
be larger) in size to autozooids and replacing them in the
budding sequence; (ii) interzooidal avicularia, smaller than
autozooids and fitted between them (i.e. their basal walls
rest on the substratum); (iii) adventitious avicularia, formed
on the frontal, basal, or lateral walls of the autozooid; and
(iv) vibracula, an extreme form of adventitious avicularium
with an elongated setiform mandible and unique hinge
structure (McKinney & Jackson, 1989; Carter et al., 2010a).
Sessile forms can be flush with the surface of the autozooid,
while pedunculate forms are elevated on a peduncle or stalk
(Fig. 4; McKinney & Jackson, 1989).
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 7
Table 2. Classification of polymorphs according to modification of their semimodules.
Polymorphic structure Definition Examples Note
Modified polypide Cystid identical to that of a regular autozooid. Tentacle (number, length, ciliation),
polypide size, and the presence/absence of other structures may be modified.
Male zooids in Odontoporella
bishopi (Cheilostomata;
Gordon, 1968; Carter &
Gordon, 2007).
Rare, perhaps owing to
difficulty of observation
(living material required).
May be due to polypide
cycles (Rogick, 1963;
Powell, 1967).
Modified cystid
Polypide identical to that of a regular autozooid. Retains feeding ability.
Modifications may be to zooid size, shape, orificial structures, or to ‘appendages’
on the zooid surface
B-zooids in Steginoporella
Membranipora membranacea
autozooids with
predator-induced spinules
on the frontal wall
(Cheilostomata; Harvell,
Referred to as ‘autozooidal
polymorphs’ or
‘appendaged autozooids’
en, 1977; Schack et al.,
Modified cystid +polypide Cystid and polypide are
modified, and zooids may
have protrusible but
non-feeding polypides
Protrusible polypide Dwarf reproductive zooids in
Celleporella hyalina
nanozooids and
gonozooids in
en (1977) distinguished
‘autozooidal polymorphs’
from ‘heterozooids’ based
on their ability to feed (and
thus fulfil the primary
function of an autozooid).
However, there are several
variations in the
non-feeding zooids
(heterozooids) present.
Non-feeding polypides may
be capable of other
functions (sensory,
secretory, sperm gathering,
Avicularia with reduced
Absent Polypide Kenozooids
Multizooidal complex Formed from modifications to two or more zooids Some ovicells (Cheilostomata)
Extrazooidal structure
Structures external to zooidal boundaries at all stages of development (Boardman &
Buttler, 2005). Can be solid skeleton or space-enclosing skeletal structures
(McKinney & Jackson, 1989).
Alveoli, styles (Stenolaemata)
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
8Carolann R. Schack and others
(A) (B)
(C) M
Fig. 3. Structure of a vicarious avicularium, frontal view. (A) skeletal morphology, (B) living state, mandible closed, (C) living state,
mandible open (rotated 90). Cr, cryptocyst; Fm, frontal membrane; G, gymnocyst; M, mandible; Op, Opesia; Ot, opening to
tentacle sheath of reduced polypide (palatal foramen); P, palate; Pv, pivot structure (hinge); R. rostral area; Ri, rim; T, adductor
muscle tendon.
Mandibles vary from spatulate, triangular, or fan-shaped,
to highly irregular forms that bear spine-like projections
(Fig. 5; Schack et al., 2018) and the rostrum (part of the
cystid that the closed mandible fits into) is often modified to
accommodate particular mandible morphology. Avicularia
varying markedly in size and shape are often found on the
same colony. For example, Carter et al. (2008) described
avicularia with oval and acute mandibles in the same
colonies of Stephanollona scintillans, and in many other species
of Phidoloporidae.
There is also a high degree of variation present in
avicularian orifices and rudimentary polypide structure.
While most taxa only have a pore in the membranous
wall (palate) below the mandible, in some the tip of the
vestigial polypide can protrude when the mandible is open
(Carter et al., 2010a). In Rhynchozoon zealandicum, this polypide
has papillae covered in microvilli and cilia, while those of
Bugulina flabellata and Arachnopusia unicornis possess prominent
tufts of cilia. Nordgaardia cornucopioides,ontheotherhand,
has a tentacle sheath that is hypertrophied into a secretory
plug (a cluster of glandular cells containing secretion vesicles)
and has internal cilia (Carter et al., 2010a). Still other taxa
are equipped with tubular orifices or bulbous extensions,
providing evidence for the high degree of variation in
form (and potential function) present in avicularia (Carter
et al., 2008, 2010a). When the sensory cilia of avicularia
are triggered, the mandibles are snapped shut using
well-developed adductor muscles (Kaufmann, 1971; Mukai
et al., 1997). Like mechanosensory ciliary tufts, the secretory
plugs are associated with a cerebral ganglion, but the trigger
for glandular secretions and their function are unknown
(Carter et al., 2010a). Avicularia may also have chemosensory
capabilities: Microporella vibraculifera,Parasmittina collifera,and
Bugulina californica responded strongly to a mixture of amino
acids (known feeding stimulators, also indicative of prey
injury; Winston, 1991).
The function of avicularia is still debated, particularly
owing to the variation in avicularian morphology.
Adaptations in bird’s-head avicularia suggest a defensive
function: the shape and mineralization of these structures
reduces mechanical stress associated with grasping a
struggling organism, and nodding behaviour allows them
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 9
Fm Ad
Fig. 4. Comparison of the musculature and polypides of avicularia and an autozooid. (A) Cheilostome autozooid, (B) sessile
avicularium, (C) adventitious bird’s-head avicularium, (D) vibraculum. The cryptocyst is not shown in D. Ab, abductor muscle
(depresses frontal membrane and opens mandible); Ad, adductor muscle (closes mandible); Ci, ciliary tuft, part of the polypide; Cr,
cryptocyst; Fm, frontal membrane; Gy, gyrator muscle; M, mandible; Om, occlusor muscle of the operculum; Op, operculum; P,
palate; Pc, peduncle cushion, provided by autozooid; Pe, peduncle; Pm, transverse parietal muscle (depresses frontal membrane
and protrudes polypide); Po, polypide; R, rostrum; T, adductor muscle tendon. A and B modified from Ryland (1970), C from
Kaufmann (1971), D from Marcus (1962).
to increase capture rates by ‘patrolling’ (Kaufmann, 1971).
These avicularia can capture large organisms by seizing
terminal segments or hairs (e.g. a 4 mm amphipod caught by
a 0.2 mm avicularium; Kaufmann, 1971), although captured
epibionts typically escape (Kaufmann, 1971; Winston, 1986;
Lidgard et al., 2012). This suggests that mitigation of potential
damage, rather than removal or destruction of epibionts, is
the main function of such avicularia (Kaufmann, 1971).
If defence is the main function of avicularia, variation in
mandible morphology may simply reflect specializations for
catching different epibionts (Sil´
en, 1977). In addition to
mechanical defence, avicularia may provide a chemical
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
10 Carolann R. Schack and others
(B) (C)
Fig. 5. Diversity of avicularian mandibles. (A) Cellaria punctata,(B)Crassimarginatella harmeri,(C)Amastigia rudis,(D)Bryopesanser thricyng
(as Escharina pesanseris), (E) Licornia longispinosa (as Scrupocellaria), (F) Steginoporella dilatata (B-zooid), (G) Smittipora cordiformis, (H) Labioporella
thornelyae. A, C, and EH drawn from Harmer (1957), B drawn from Fransen (1986), D drawn from Gordon (1984).
deterrent which would help explain the presence of a
secretory plug in some forms of avicularia (Carter et al.,
2008, 2010a).
Avicularia may also direct waste-water currents to colony
margins and chimneys, aiding in the removal of excretory
waste. A waste-removal function may explain spathulate
mandible types and sporadic spacing of avicularia (wide
spacing may enhance waste-removal currents but would
decrease defensive capabilities; Kaufmann, 1971; Winston,
1984). It is also possible that avicularia catch and retain
epibionts to create bait for organisms small enough to be
consumed by the bryozoan [e.g. bacteria (Goldstein, 1880,
cited in Scholz & Krumbein, 1996)]. Several other functions
of avicularia have been proposed (nutrient storage, direct
food capture, aid in respiration, baffles), but there is little
evidence for these (Marcus, 1926, cited in Winston, 1984).
Unlike avicularian mandibles, which are hinged to swing
through one plane, the elongate mandibles of vibracula rotate
and sweep over the colony (McKinney & Jackson, 1989;
Carter et al., 2008). Their extreme morphology (Fig. 4), may
indicate a different purpose from other avicularia. Vibracula
respond to tactile stimulation and may sweep debris and
organisms from the colony surface (McKinney & Jackson,
1989; Winston, 1991). However, some free-living forms use
Fig. 6. A ‘walking’ colony of Selenaria maculata. C, colony; V,
vibracular seta. Drawn from Cook & Chimonides (1978).
vibraculum-like avicularia to raise their colony above the
substratum, unbury or overturn itself, or as a means of
locomotion (Fig. 6; Cook & Chimonides, 1978).
Some cheilostomes, mainly in Steginoporella (Steginoporell-
idae) and Macropora (Macroporidae), possess polymorphic
feeding zooids with enlarged mandibles. These ‘B-zooids’
have a reinforced orifice, a thick skeletonized operculum,
and larger opercular muscles (Banta, 1973; Cheetham et al.,
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 11
2006; Winston, Vieira & Woollacott, 2014). Steginoporella
mandibulata is the only species of Steginoporella where the
B-zooids have non-protrusible polypides and can be consid-
ered ‘true’ avicularia (Harmer, 1957; Banta, 1973). B-zooids
may represent an early stage in the evolution of avicularia
(Harmer, 1957; Banta, 1973).
The reduction of the polypide in avicularia allows the
avicularian cystid also to shrink in size, which in turn permits
more integrated forms (e.g. interzooidal and adventitious
avicularia), allowing defensive and feeding functions to be
fulfilled in the same space. The functional polypide of a
B-zooid allows it to feed, while its enlarged operculum
allows it to trap organisms (Winston, 2004). The loss of
a B-zooid’s feeding ability in the transition to a vicarious
avicularium would need to be offset by the increased
efficiency of another function. The energetic cost of this may
be the reason why only S. mandibulata has avicularia, while
other Steginoporella only have B-zooids (or lack mandibulate
polymorphs). However, B-zooids are far less prevalent than
avicularia in other cheilostome taxa. For example, the
genus Crassimarginatella has several species with B-zooids
(C. similis, potentially C. latens; Cook, 1968), while others
typically have vicarious avicularia. Although the transition
from B-zooid to vicarious avicularium may have initially
been more expensive, vicarious avicularia provide several
benefits: damage to a vicarious avicularium will not affect
colony feeding ability; muscle size is not limited by the
presence of a feeding polypide; and rostrum shape is not
restricted to accommodate the protruded tentacle crown
(e.g. B-zooids with narrow setose mandibles do not exist).
Separating feeding and defensive structures into discrete
modular units may be more efficient than a single
‘multitasking’ unit. Lineages that retain B-zooids either
cannot afford the reduction in feeding ability or do not
require the increased efficiency.
(b)Evolution of avicularia
The earliest avicularia appear in the Cretaceous
(AlbianCenomanian) genus Wilbertopora. These avicularia
range from barely modified B-zooids retaining a functional
tentacle crown and gut (W. mutabilis) to highly modified
(pointed and interzooidal avicularia inferred to have lacked
the ability to feed (W. acuminata and W. hoadleyae; Cheetham
et al., 2006). However, living genera also possess species with
a similar range of avicularian modification (Crassimarginatella;
Cook, 1968). There are several trends in the modification of
autozooids to avicularia: (i) distal elongation and decreased
opesial area; (ii) reduction of the polypide and its feeding
ability; (iii) possession of an elevated rim; (vi) production of
a resting surface for the mandible (e.g. an incomplete shelf
or hard, complete palate); (v) inclusion of a pivot structure
(condyles or a pivot bar) that separates the orificial and opesial
areas; and (vi) transition from a small, rounded operculum
to an enlarged, pointed mandible (Fig. 7; Cook, 1968;
Cheetham et al., 2006). However, all trends (particularly
46) may not necessarily be present in all avicularia. The
transition from rounded to pointed mandibles may follow a
(A) (B) (C) (D) (E)
Cr Cr
Fig. 7. Increasing modification of avicularian skeletal parts,
showing development of the cryptocystal shelf, reduction in
the post-mandibular area, elongation of the mandibular area,
and the development of hinge structures for the mandible.
Below the line is the post-mandibular area, above it is the
orificial/mandibular area. (A) autozooid, (B) B-zooid with
slight modification, as in Wilbertopora mutabilis, (C) B-zooid
with moderate modification, as in W. listokinae or Steginoporella
murachbanensis, (D) spathulate avicularium (strongly modified), as
in W. manubriformis or Crassimarginatella spatulata, (E) triangular
avicularium (strongly modified), as in W. sannerae or C. falcata.
C, condyles (hinge); Cr, cryptocystal shelf; Pb, pivot bar; Ri,
rostral rim.
latitudinal gradient but this has not been tested statistically
(Schopf, 1973).
Since Cretaceous Wilbertopora and PaleogeneRecent
Crassimarginatella both show a similar range of avicularian
modification, it is likely that avicularia arose independently
multiple times over the course of cheilostome evolution
(Cheetham et al., 2006). Although vicarious avicularia were
the first to evolve, adventitious avicularia occur more
frequently and at higher densities. This may reflect a
trend towards increased colonial integration and efficiency,
allowing colonies to possess highly modified avicularia
without losing the energy provided by fully functional
autozooids (Carter et al., 2010a; Lidgard et al., 2012). Even
if some avicularia no longer provide a solely defensive
function, repeated evolution of avicularia may be in
response to predation by epibionts (Lidgard et al., 2012).
Similarly, a trend for increasing colonial integration and
avicularian/vibracular capability in free-living lunulitiform
bryozoans is likely. In Otionellina (Otionellidae) vibracula
clean the colony surfaces, while vibracula in Selenaria
(Selenariidae; Fig. 6) use elongate mandibles to ‘walk’ across
the substratum, right themselves if overturned, and unbury
themselves if covered by sediment (Cook & Chimonides,
1978; Chimonides & Cook, 1981).
(2) Eleozooids and aviculomorphs
The clearest stenolaemate analogue to cheilostome avicularia
is found in the extinct Eleidae (Cyclostomata). Eleids are
unique among stenolaemates because they possessed a
calcareous operculum and avicularium-like heterozooids
called eleozooids (Taylor, 1986; Viskova, 2016). Like
avicularia, eleozooids may have lacked a feeding polypide.
Their preserved opercula are typically hypertrophied, but
may also be reduced in comparison with autozooids (Taylor,
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
12 Carolann R. Schack and others
(A) (B) (C)
Fig. 8. Eleid autozooid and eleozooids. (A) autozooid, (B)
demizooid, (C) trifoliozooid, (D) rostrozooid. Redrawn from
Fig. 1 of Taylor (1986).
1994). Rostrozooids had an aperture that is enlarged and
distally tapered into a rostrum, with similar variations to
avicularian mandibles. Trifoliozooids had small apertures
resembling an inverted ‘T’. Demizooids had D-shaped
apertures and opercula. Both trifoliozooids and demizooids
are smaller than autozooids but can replace them in a series
(Taylor, 1986, 1994) (Fig. 8).
There were no adventitious eleozooids (all are analogous
to vicarious or interzooidal avicularia), but there is still a
trend towards increasing colonial integration of polymorphs,
similar to that in cheilostomes (Lidgard et al., 2012). After
their initial appearance [100.5 113.0 million years ago
(Mya)], when they were randomly arranged within colonies,
a regular arrangement of eleozooids arose in free-walled
(93.9100.5 Mya) and fixed-walled (86.3 – 89.8 Mya) species
(Taylor, 1986). Although typically reserved for cheilostomes,
the term cormidium could be applied to these patterns of
autozooids and eleozooids (especially in the genus Atagma,
which has autozooids surrounded by smaller demizooids;
Taylor, 1986).
It is likely that eleozooids performed similar functions
to avicularia (Viskova, 2016). However, eleozooids have
no analogue in Recent cyclostomes (Taylor, 1986, 1994;
Viskova, 2016), so their exact function cannot be observed.
The mandibles of both avicularia and eleozooids
are derived from opercula, which are present only in
cheilostomes and eleids. In this way, eleids may be
convergent with cheilostomes (Taylor, 1986). It is clear
that the operculum is a key structure in the evolution of
avicularian-like polymorphs (Schopf, 1973; Taylor, 1986),
although the existence of an operculum does not automati-
cally ensure that these structures will arise, since eleids lacking
eleozooids persisted until the Maastrichtian (66.0– 72.1 Mya;
Taylor, 1986) and Recent non-polymorphic cheilostomes
[and one operculum-bearing genus of Ctenostomata
(Penetrantia); Pohowsky, 1978] still exist.
An operculum may not be necessary to evolve a
mandibulate polymorph. In Fenestropora, a Devonian genus
of stenolaemates, avicularian-like structures are present on
the underside of colony branches. These ‘aviculomorphs’
consist of a concave pit surrounded by a triangular outline
with an acute tip and may have had a cuticular mandible
(McKinney, 1998). Aviculomorphs may have arisen from
modifications to the laminar extrazooidal skeleton, like
cyclozooecia (McKinney, 1998). Cyclozooecia are a similar
size to aviculomorphs and occur primarily on the underside
of branches (McKinney, 1998). The difference between the
circular outline of cyclozooecia and the triangular outline of
aviculomorphs is reminiscent of the change from a circular
operculum to a pointed mandible in some avicularia.
Kenozooids lack a polypide. These zooids can either be
vicarious (budded between zooids) or adventitious (budded
on zooids). In gymnolaemates, kenozooids are not ‘empty’,
but are enervated by the funicular system and may
contain musculature (Cheetham & Cook, 1983; Lutaud,
1983; Gordon & Parker, 1991b). The funiculus is formed
from an epithelial layer of the cystid basal wall (at least
in Membranipora membrancea; Lutaud, 1983), not from the
developing polypide. Similarly, muscles external to the
polypide, such as apertural and parietodepressor muscles,
are formed from groups of myocytes on the walls of the cystid
buds (Lutaud, 1983). Since the cystid walls are formed before
the polypide bud itself, the construction of a gymnolaemate
kenozooid should not require a polypide bud during
In stenolaemates the polypide forms before the cystid
(Mukai et al., 1997), suggesting that initial development of
the polypide and subsequent degeneration are required
in kenozooid formation. Further evolution of kenozooidal
structures in stenolaemates may have resulted in genetically
‘preprogrammed’ kenozooids that skip polypide formation
entirely. However, doing so likely required more changes
in developmental timing (via further mutations) than in
The apparent ease by which gymnolaemates can
develop kenozooids may contribute to their high level of
polymorphism. Since the kenozooid is unconstrained by the
need to accommodate a polypide (both during and after
development), it can fulfil a wide range of morphologies.
In cheilostomes, these kenozooidal morphs were likely
responsible for two important taxonomic novelties: the
frontal shield and the ooecium (see Sections IV.5 & V.2; Dick
et al., 2009; Ostrovsky & Taylor, 2005a; Lidgard et al., 2012).
High polymorphism arising from kenozooids is apparent in a
cyclostome family with kenozooids: species of Crisiidae can
have branches of vicarious kenozooids (e.g. Bicrisia abyssicola),
hollow spines (Crisidia cornuta), and even kenozooidal rhizoids
(Crisia ramosa; Hayward & Ryland, 1985). While the presence
of kenozooids does not automatically ensure the evolution of
such structures, it provides an important module necessary
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 13
for their construction (similar to operculate zooids in the
development of avicularia).
Kenozooids are easily confused with extrazooidal
structures in stenolaemates. The main distinction between
kenozooids and extrazooidal chambers is that the space inside
a kenozooid cystid is homologous to that of an autozooid,
while the space in an extrazooidal chamber is created de novo
by delimiting extrazooidal walls and is not homologous
to that of an autozooid. Given the potential difficulty
in developing kenozooids in stenolaemates, developing
extrazooidal structures may have been an evolutionarily
easier solution, with the extrazooidal skeleton serving as
an ‘unconstrained module’. The evolution of extrazooidal
structures requires further investigation.
Kenozooids may also be confused with autozooidal
appendages. Autozooidal appendages are not polymorphic
zooids themselves but are modifications to the cystid of
an autozooid. These appendages vary in form (e.g. spines,
peristomes, tubercles) but make an autozooid polymorphic
only if they exhibit discontinuous variation with other zooids
in the same astogenetic and ontogenetic zone.
(1) Space-filling and strengthening structures
In cheilostomes, space-filling kenozooids are budded at
branch edges, colony stalks, and between zooids when
space is limited. In erect species, kenozooids often appear
in a consistent pattern (e.g. Spiralaria florea and Chelidozoum
ternarium) (Sil´
en, 1977; McKinney & Jackson, 1989; Gordon &
d’Hondt, 1991), although this can also occur less frequently in
encrusting species (Hesychoxenia praelonga; Gordon & Parker,
1991b). In Chelidozoum ternarium the main colony structure
is provided by slender kenozooidal chains (Gordon &
d’Hondt, 1991), while in Hesychoxenia praelonga kenozooids
are budded along the colony margin and alternate with pairs
of autozooids (Gordon & Parker, 1991b). These vicarious
kenozooids (‘vicariozooids’ of Sil´
en, 1977) can be irregular
or similar in shape to autozooids, but are typically smaller
than autozooids. Cheilostome kenozooids may bud spines
(as in Chorizopora ferocissima) or adventitious avicularia (e.g.
Chaperiopsis cristata), and may be the daughter zooid in an
ovicell complex (Gordon, 1984; Ostrovsky et al., 2009). In
general, the function of space-filling kenozooids may be to
increase colony rigidity and the robustness of colony margins
en, 1977; McKinney & Jackson, 1989). This function
explains the presence of elongated spicule bundles in the
kenozooids of Hesychoxenia praelonga, which encrusts flexible
seagrass stems (Gordon & Parker, 1991b).
In stenolaemates, space-filling structures may be arranged
between zooids to ensure sufficient distance between
apertures and tentacle crowns (as in the cyclostome suborders
Cancellata, Rectangulata, and Cerioporina; Taylor &
Weedon, 2000). Such structures may provide an alternative
colony-strengthening method to thickened autozooid walls
(Key, 1991) or an energetically expensive extrazooidal
skeleton. Space-filling polymorphs may also be formed
in response to occlusion by gonozooids (see Section
V.4; Sch¨
afer, 1991). Many types occur in stenolaemates:
mesozooids/mesopores, metapores, nematopores, cancelli,
alveoli, dactylethrae, and exilazooids/exilapores in addition
to kenozooids. Most of these structures are not clearly
zooidal, kenozooidal, or extrazooidal (Table 3), and the terms
are often ambiguous, and further revision of space-filling
polymorphs is necessary (e.g. Batson & Smith, 2018). Not
only will this clarify terminology, but it will also determine
the taxonomic extent of kenozooidal and extrazooidal
space-filling structures.
(2) Stolons
Stolons are elongated, cylindrical structures that are
typically kenozooidal (Sil´
en, 1977) but may be adventitious
appendages (e.g. Immergentia: Ctenostomata and Aetea:
Cheilostomata; Pohowsky, 1978); Hayward & Ryland, 1998).
In many ctenostomes stolons budded distally generate
the main architecture of the colony, creating branching
patterns, while autozooids and adventitious stolons are
budded laterally on a principal stolon (Fig. 9; Pohowsky,
1978); Cheetham & Cook, 1983; Souto, Fern´
& Reverter-Gil, 2010; Souto et al., 2011). In cheilostomes,
terminal kenozooidal stolons at colony margins and frontal
stolons form from an evagination of the frontal wall
(Gordon, 1972). Both can be produced within a single
colony (e.g. Schizoporella unicornis). Their morphology is
variable: frontal stolons in Celleporaria sp. are short and
bulbous (Osborne, 1984), but those in C. apiculata can
be 30 times the length of the autozooid (Tzioumis,
1994). Both types can be produced in response to
contact between encrusting bryozoan colonies (Fig. 9).
Stolon production can result in cessation or redirection
of growth in addition to competitive overgrowth (Osborne,
1984; Tzioumis, 1994). Large colonies of M. membranacea
use stolons to slow the growth of small intraspecific
competitors and allow the larger colony to surround it
(Padilla et al., 1996).
In addition to kenozooidal stolons, M. membranacea
produces vertical growths that may be analogous to
frontal stolons (Osborne, 1984). These ‘tower zooids’ are
kenozooids ontogenetically derived from autozooids (Cook
& Chimonides, 1980; Xing & Qian, 1999). Tower zooids
lack a polypide and orifice and, most notably, possess an
evaginated cuticular frontal wall that can grow up to 8.4 mm
high (Fig. 9; Cook & Chimonides, 1980; Hayward & Ryland,
1998). These structures may control water currents above the
colony surface or provide protection from abrasion (Cook &
Chimonides, 1980; Xing & Qian, 1999). This latter function
is particularly important for M. membranacea since it encrusts
red algae and may be frequently scraped across other fronds
or substrata (Cook & Chimonides, 1980; Xing & Qian,
1999). Frontal stolons are reversibly induced polymorphs,
only lasting for around 6 weeks (Osborne, 1984; Tzioumis,
1994), while tower zooids are irreversible polymorphs. When
comparing these structures, it is tempting to suggest that the
reversible frontal stolon may have given rise to the irreversible
tower zooid, or vice versa.
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
14 Carolann R. Schack and others
Table 3. Descriptions of space-filling polymorphs in stenolaemates.
Space-filling polymorph Description Taxa Author Condition
Metapore Deep cavities in the extrazooidal skeleton that lack
skeletal walls of their own.
Rhabdomesina, Cryptostomata; Petalopora
costata,Grammascosoecia dichotoma,Reteporidea
lichenoides (Petaloporidae: Cancellata)
Boardman (1983,
1998); Sch¨
Nematopores (syn.
Slender, thread-like kenozooids. Tubuliporina Brood (1972);
Boardman (1983)
Squat, unclosed kenozooids with strong skeletal walls. Discocytis eccentrica;Discocytis infundibibuliformis Sch¨
afer (1991) Kenozooidal
Tergopore Pores with walls that are structurally similar to
autozooids and are perforated with interzooidal
communication pores.
Crisina,Pleuronea Brood (1972) Kenozooidal
Alveoli Shallow polymorphs with walls perforated by
communication pores.
Disporella Boardman (1998) Extrazooidal
Morphological differentiation of the extrazooidal
Rectangulata Sch¨
afer (1991) Extrazooidal
Vacuoles with walls that have a similar
microstructure to autozooidal walls. Syn. cancelli
Disporella;Lichenopora: Rectangulata Brood (1972) Kenozooidal
Kenozooids often sealed by a calcified exterior wall.
Syn. cancelli
Cyclostomata Taylor & Weedon
Cancelli Vacuoles in the extrazooidal skeleton without skeletal
Hornerids Boardman (1998) Extrazooidal
Terminally occluded, short kenozooidal chambers. Crisina watersi (Petaloporidae: Cancellata);
Discocytis canadensis (Cytididae: Cancellata)
afer (1991) Kenozooidal
Alveoli occluded by calcaerous diaphragms. Lichenopora: Rectangulata Harmer (1896) Not given
Vacuoles with walls that have a similar
microstructure to autozooidal walls. Syn. alveoli
Lichenopora: Rectangulata Brood (1972) Kenozooidal
Kenozooids often sealed by a calcified exterior wall.
Syn. alveoli
Cyclostomata Taylor & Weedon
Dactylethrae Short kenozooids. Terebellaria, Cyclostomata Brood (1972);
Boardman (1983)
Small, club-like chambers, always closed by a
terminal diaphragm.
Cerioporina Sch¨
afer (1991) Extrazooidal
Mesozooid/ mesopore Tapering vacuoles, which are too small to house a
functional polypide and may be closed by a
diaphragm. Can possess internal diaphragms or
Palaeozoic Stenolaemata Boardman (1983,
Cyclostomata Sch¨
afer (1991) Extrazooidal
Trepostomata Boardman &
Buttler (2005)
Exilazooid/exilapore Vacuoles with few or no diaphragms, but possessing
internal flask-shaped structures.
Trepostomes Boardman &
Buttler (2005)
Small zooids with flask-shaped structures and
potentially reduced polypides.
Palaeozoic Stenolaemata Utgaard (1973);
Boardman (1983)
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 15
Fig. 9. Gymnolaemate stolons. (A) Kenozooidal stolons supporting autozooids in the ctenostome Amathia pustulosa, (B) tower zooids
in Membranipora membranacea (side view), (C) intercolonial competition between the cheilostomes Stylopoma duboisii (top right, generating
stolons) and Hippopodina feegeensis (bottom left). Az, autozooid; S, kenozooidal stolon; Ts, terminal stolon (kenozooidal); Tz, tower
zooid (autozooidal appendage). A drawn from Souto, Fern´
andez-Pulpeiro & Reverter-Gil (2011), B drawn from Hayward & Ryland
(1998), C drawn from Osborne (1984).
(3) Rhizoids
Rhizoids (rhizozooids, rootlets, and radical fibres) are thin
tubes that can anchor erect colonies or raise planar
colonies above soft substrata (e.g. Beaniidae, Petraliidae,
Retiflustra; Harmer, 1926; Cook et al., 2018). These structures
possess a variety of terminal ends to anchor colonies
(single tip, fimbriate, holdfasts; Fig. 10; Vieira et al., 2014;
Schack et al., 2018), which can be adhesive (Cook &
Chimonides, 1981, 1985; Cook et al., 2018). Ctenostome
rhizoids may bore through organic substrata (e.g. echinoid
spines and sipunculan cuticles by Triticella minini and T.
maiorovae repectively; Grischenko & Chernyshev, 2015, 2017),
perhaps via biochemical erosion (e.g. some shell-boring taxa;
Pohowsky, 1978). Some small free-living colonies may use
turgor pressure in rhizoids to elevate their colonies above the
substratum (Fig. 11; Hirose, 2011).
In cheilostomes and cyclostomes, rhizoids are typically
modified kenozooids that bud adventitiously from autozooids
or vibracular chambers (Taylor & Weedon, 2000). In
ctenostomes, rhizoids develop as a series of septa-separated
tubes, the first of which is an evagination of a stolon or
autozooid wall (Souto et al., 2010, 2011). This is similar to the
jointed rhizoids in the cyclostome family Crisiidae (Hayward
& Ryland, 1985). While most rhizoids are kenozooidal,
they can also be appendages formed from evaginations of
the body wall [‘props’ as in Chaperiopsis uttleyi (Gordon,
1992) and Favosipora (Gordon & Taylor, 2010); Fig. 10]
or the frontal cuticle [e.g. Steginoporella neozelanica (Gordon,
Voje & Taylor, 2017)]. Extrazooidal rhizoids also occur:
sand-dwelling colonies of Lanceopora and Sphaeropora fossa use
a single rhizoid to elevate their colonies (Fig. 11). The rhizoid
is formed from the cuticular body wall and is able to repair
damage and maintain tugor pressure (provided by coelomic
fluid) (Cook & Chimonides, 1981, 1985; Cook et al., 2018).
(a)Articulated colonies
Erect bryozoan colonies may possess articulated joints. The
joints may be indeterminate (provided by dorsal rhizoid
bundles; Hageman et al., 1998) or determinate (autozooidal
appendages; Sch¨
afer, Bader & Blaschek, 2006). Determinate
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
16 Carolann R. Schack and others
2 mm
2 mm
(A) (B)
Fig. 10. Rhizoids (zooids) versus props (autozooidal appendages). (A) Camptoplites sp. (Cheilostomata) with rhizoids (Rh), (B) Pandanipora
helix (Cyclostomata) with props (P). Photographs by A.V. Grischenko.
articulation is most common in cheilostomes (particularly
Cellaria) but is also found in Crisia (Cyclostomata) and in
Paleozoic stenolaemates (Wyse Jackson, Ernst & Andr´
2017). In Cellaria sinuosa the joints are chitinous, while in Crisia
eburnea they appear to be collagenous (Sch¨
afer et al., 2006).
Articulation allows erect colonies to resist mechanical stress
through flexibility (McKinney & Jackson, 1989; Poluzzi,
Masini & Capozzi, 1991) rather than investing in skeletal
strength, extrazooidal calcification, or branch thickening
(Cheetham, 1986). This may be an adaptive strategy for
high-energy environments (McKinney & Jackson, 1989;
Poluzzi et al., 1991; Sch¨
afer et al., 2006). If mechanical stress
results in joint breakage, the internodes (zooid clusters) can
generate new colonies and function as asexual propagules (as
in C. sinuosa; Bader, 2000).
(4) Spines
‘Spine’ refers to a variety of skeletal and cuticular
structures. These are typically hollow with an inner lining of
epithelial cells, and possess varying degrees of calcification
(unmineralized to fully calcified) (Sil´
en, 1977; Ostrovsky,
1998; Lidgard et al., 2012). Spines can have a cuticular basal
articulation or can be unjointed costae. Spines exhibit a wide
range of variation (tubercles, branching spines, shield-like
scuta) (Sil´
en, 1977; Gordon, 1984, 1986; Vieira et al., 2014)
that generally bud adventitiously on zooids and are incapable
of budding zooids.
(a)Gymnolaemate spines
Most cheilostome spines are autozooidal appendages
(Ryland, 1979; Cheetham & Cook, 1983). True kenozooidal
spines (‘spinozooids’ of Sil´
en, 1977) around the opesia have
been found only in Bellulopora (Ostrovsky & Taylor, 2005a),
where they form a costate frontal shield. While kenozooidal
spines are separated from the autozooid by communication
pores, spines formed as appendages of the autozooid are
not (Sil´
en, 1977; Cheetham & Cook, 1983; Ostrovsky &
Taylor, 2005a; A.N. Ostrovsky, personal communication).
Other spine-like kenozooids exist in a variety of taxa
[e.g. terminal kenozooids in Chelidozoum ternarium (Gordon
& d’Hondt, 1991); and the ctenostome Amathia wilsoni
(Chimonides, 1987)], although these structures are not
budded adventitiously as true kenozooidal spines. In some
cheilostomes spines project internally from cystid walls; they
may be needle-like, branching, or form a robust scoop-like
‘plectriform apparatus’ to guide the polypide as it everts
(Gordon & Parker, 1991a, 1991c).
Although they may not possess structural homology, most
spines are functionally homologous. It is likely that spines
first evolved in response to predation (Dick et al., 2009). The
earliest known cheilostome with spines is Charixa burdonaria,
which had a pair of latero-oral spine bases (Taylor, Lazo
& Aguirre-Urreta, 2009). This suggests that oral spines
evolved first 135.0135.8 Mya, before spreading to the
periopesial margin to protect the membranous frontal wall
as in Spinicharixia 124.5112.0 Mya (Taylor et al., 2009).
The clearest evidence for a defensive role is in
M. membranacea, which produces cuticular spinules (short
spine-like appendages, non-homologous to most spines)
on the membranous frontal wall in response to grazing
by nudibranchs (Harvell, 1984, 1986). Interestingly, three
genotypes are present in M. membranacea: colonies with
inducible spinules (irreversible polymorphs), and colonies
that produce and do not produce spinules regardless of
nudibranch-related cues (consistent polymorphs) (Harvell,
1998). When nudibranchs are removed, colonies with the
inducible genotype cease spinule production within 1 day.
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 17
Fig. 11. Colonies supported and anchored by turgid rhizoids (not to scale). (A) Lanceopora sp., (B) Sphaeropora fossa,(C)Conescharellina
catella (rhizoids below the sediment not shown), (D) relative sizes of colonies (arrow points to C. catella). Er, extrazooidal rhizoid; F,
fibril of extrazooidal rhizoid (an appendage, not a new rhizoid); H, head of the colony, containing autozooids; P, everted tentacle
crown of polypide; Rh, rhizoid (zooidal). The top of the sediment is represented by a horizontal black line. A drawn from Cook &
Chimonides (1985), B drawn from Cook & Chimonides (1981), C drawn from Hirose (2011).
Although spinule production decreases colony growth rate
by 14% (Harvell, 1986), the cost is likely balanced by
the reduction in damage: spinulated colonies were largely
undamaged and experienced lower predation rates than
unspinulated colonies (Harvell, 1984). Spines in Electra spp.
deter nudibranchs by ‘spearing’ them (Cook, 1985). Defence
against nudibranchs is important since many nudibranchs
feed only on bryozoans (Lidgard, 2008).
Although marginal spines on the edges of colonies appear
important in preventing overgrowth by other organisms
(Stebbing, 1973; Scholz, 1995), spines alone provide no
significant advantage in interspecific competition between
bryozoans (Barnes & Rothery, 1996).
While defence is likely their main function, spines may
also provide protection against abrasion. This is particularly
useful for species that often experience tumbling and
those that live on kelp fronds [like Electra pilosa and the
ctenostome Flustrellidra hispida (Stebbing, 1973; Whitehead,
Seed & Hughes, 1996)]. Protection from abrasion and
predation may have different structural requirements (e.g.
pointed tip versus rounded tubercle). Oral spines may
also increase local turbulence, trapping food particles
near the tentacle crown or by increasing boundary layer
thickness and allowing tentacle crown protrusion into
higher-velocity waters (Riedl & Forstner, 1968, cited in
Whitehead et al., 1996).
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
18 Carolann R. Schack and others
(C) (D)
(E) (F)
Fig. 12. Frontal shields of cheilostomes. (A) Basic anascan condition (no shield), frontal wall wholly or principally membranous
(cuticular), (B) membranous frontal wall protected by a shield of overarching spines (spinocyst), (C) ascophoran with gymnocystal
shield, i.e. calcified frontal body wall (cuticle overlies the skeletal layer), (D) ascophoran with an umbonuloid shield, (E) ascophoran
with a lepralioid shield, (F) ascophoran with a lepralioid shield with pseudopores. Note that the frontal shield in DF has an
overlying hypostegal coelom and cuticular epitheca. Asc, ascus; Crs, cryptocystal shelf; Cu, cuticle; Efm, homologous extension of
the membranous frontal wall, the epithelial cells of which peel back from the cryptocystal shelf in lepralioid shields (Cheetham &
Cook, 1983); Esw, external (gymnocystal) skeletal wall; Hc, hypostegal coelom; Pfm, primary cuticular frontal wall (reinforced in C
by calcifying beneath it to form a gymnocyst); PP, pseudopores plugged with tissue; Sp, spines; Us, umbonuloid shield. Drawn from
Mukai et al. (1997).
(b)Stenolaemate spines
In stenolaemates, kenozooids may be modified to form
‘spinose kenozooids’ (Sil´
en, 1977; Taylor & Weedon, 2000).
These heterozooids are analogous to the spinozooids of Sil´
(1977). These spines may be single kenozooids or slender
chains separated by cuticular joints (e.g. Crisidia cornuta;
Hayward & Ryland, 1985). Apertural spines can surround
the aperture of a zooid, although typically only one spine
is present. These spines are projections of the inner body
wall and only occur in free-walled autozooids that lack a
peristome (Taylor & Weedon, 2000). Like the plectiform
apparatus, mural spines (spinules) and pustules extend into
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 19
the zooidal chamber from the skeletal wall. These spines
may be attachment structures for the polypide [e.g. some
Disporella species (Boardman, 1998)].
Extrazooidal spine-like ‘stylets’ may also be present in
stenolaemates, which unlike spines, possess a solid core
enclosed by laminating sheaths (Blake, 1983; Boardman &
Buttler, 2005). A few rare forms arise from zooidal boundaries
[zooecial styles, exozonal styles (Boardman & Buttler, 2005)],
and should be considered autozooidal appendages instead of
extrazooidal structures. Stylets may have supported tissues of
confluent coeloms, facilitating interzooidal communication
(Blake, 1983; Boardman, 1983). The loss of stylets in
fixed-walled stenolaemates, which possessed communication
pores, supports this theory (Boardman, 1983). However,
prominent stylets may have served a defensive purpose
(Blake, 1983).
(5) Frontal walls
Kenozooids and spines both contributed to the evolution
of ascophoran frontal shields with a hypostegal coelom.
Periopesial spines (around the frontal membrane) became
rigid costae through the loss or calcification of their
cuticular joint base. Fusions between costae led to the
creation of the cribrimorph costal shield during the
Late Cretaceous [although costal shields have evolved
independently many times (Dick et al., 2009; Lidgard et al.,
2012)]. The cribrimorph costal shield then acted as a surface
for overlying kenozooids, allowing thicker shields to be
created (Gordon & Voigt, 1996; Lidgard et al., 2012). This
can be observed in the Cretaceous Ascancestor bretoni, which
possesses broad interzooidal kenozooids that overgrow the
lateral and frontal walls of autozooids (Gordon & Voigt,
1996). Subsequent reductions in the underlying spines (costal
field) and the frontal membrane produced umbonuloid and
lepralioid shields (Fig. 12). This restricted the hydrostatic
mechanism used to protrude the tentacle crown and required
the development of an ascus (a compensation sac that evolved
independently many times; Gordon, 2000). Development of
the costal shield and ascus likely increased survival and
contributed to the ascophoran radiation (Gordon & Voigt,
1996; Dick et al., 2009; Lidgard et al., 2012).
Reproductive polymorphism includes morphological varia-
tion in both zooids and brood chambers. Excellent reviews
exist for Gymnolaemata (Ostrovsky, 2013) and Stenolaemata
afer, 1991), summarized here (see also Reed, 1991).
(1) Gonochoristic zooids
Autozooids can be sterile, hermaphroditic (testes and ovaries
developing together), male, or female. Hermaphroditic
zooids typically undergo reversible changes to the polypide
(e.g. development of intertentacular organ) (Reed, 1991). In
colonies with gonochoristic zooids (most gymnolaemates and
stenolaemates), male and female zooids can be differentiated
based on modifications to the polypide, cystid, or both (Reed,
1991). Cystid modification can include changes in size (dwarf
females of Haplopoma sciaphilum;Sil
en & Harmelin, 1976),
shape (Reptadeonella violacea; Hayward & Ryland, 1999), or in
characteristics of the orifice (some Calyptotheca; Ryland, 1976).
For example, Celleporella hyalina and Antarctothoa bougainvillei
(both Hippothoidae) possess autozooids, ovicelled female
zooids, and male zooids that are all distinct in terms of their
skeleton and polypide structure. Male and female zooids are
dwarfed in comparison to sterile autozooids, with reduced
polypides, the female’s being more rudimentary (Marcus,
1938, and Hughes, 1987, cited in Ostrovsky, 1998). The most
extreme gonochoristic changes result from the formation of
brooding structures.
In Odontoporella bishopi (Fig. 13), autozooids and male zooids
have identical cystids but different polypides (Carter &
Gordon, 2007). The egg-producing autozooids have 1516
tentacles, while males have only eight unciliated tentacles
(four short, four long) and a reduced polypide filled with
spermatozoa [Gordon, 1968 as Hippopodinella adpressa]. The
identical skeletal morphology of autozooids and males may
be a consequence of cyclic polypide degeneration and
regeneration, since in A. bougainvillei and A. tongima autozooids
may regenerate as males (Rogick, 1963; Powell, 1967 as
Hippothoa bougainvillei).
Although the vast majority of bryozoans are colonial
hermaphrodites, there may be dioecious cyclostome colonies
composed of only male autozooids or sterile autozooids,
female autozooids (future gonozooids) and gonozooids,
which are incapable of self-fertilization [e.g. Filicrisia geniculata
(Jenkins et al., 2017)]. The sex ratio within colonies can also
be influenced by environmental conditions: when exposed to
stress the cheilostome Celleporella hyalina increased production
of males and suppressed investment in brooding females
(Hughes et al., 2003).
(2) Cheilostome brood chambers
Cheilostome bryozoans may either spawn embryos (as
planktotrophic cyphonautes larvae) or incubate embryos
before release (lecithotrophic coronate larvae) (Ostrovsky,
2013). Cheilostome brooding methods include: external
membranous sacs (ovisacs), calcified chambers (ovicells),
uncalcified internal brood sacs, or the intracoelomic space
(Ostrovsky, 2013). The origin of ovisacs is unclear (likely a
sticky fertilization envelope; Str¨
om, 1977; Ostrovsky, 2013),
but they are present only during the reproductive period
(e.g. Aetea species; Str¨
om, 1977). Since the ovisac does not
involve modification of the maternal zooid it should not be
considered a polymorphic appendage. External ovisacs also
appear in ctenostomes and may represent the most primitive
brooding method (Ostrovsky, 2013).
Most cheilostomes brood their embryos in external calci-
fied chambers called ovicells. These consist of the ooecium
(protective hood), the brood cavity (incubation chamber),
and a closing mechanism (Ostrovsky, 2013). Eggs pass into
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
20 Carolann R. Schack and others
(A) (B)
Fig. 13. Odontoporella bishopi, showing differences between the
polypides of (A) an autozooid and (B) male zooid. Redrawn
from Gordon (1968). Note that the four short tentacles of the
male zooid (B) are drawn offset for clarity.
the ovicell through a gonopore of the maternal zooid, and the
ovicell may be closed by an ooecial vesicle, the operculum,
or both. The vesicle is a membranous evagination of the
maternal zooid’s distal wall, the interior of which is confluent
with the coelom of the maternal zooid, and is operated by
special musculature (Ostrovsky, 2013). In some taxa the ooe-
cial vesicle aids in the delivery of extraembryonic nutrition
(Moosbrugger et al., 2012). The maternal zooids can be con-
sidered polymorphic on the basis of the ooecial vesicle alone.
The double-walled ooecium is not formed by the maternal
(egg-producing) zooid, but by a proximal-facing outgrowth of
a distal zooid (Fig. 14). This distal zooid can be an autozooid,
an avicularium, or a kenozooid (Ostrovsky, Dick & Mawatari,
2008; Ostrovsky et al., 2009; Ostrovsky, 2013). In many
cases, the distal zooid forms the floor of the brood chamber
in addition to the ooecium [‘type 1’ ooecia of Ostrovsky,
2013]. ‘Type 2’ ooecia are formed by a kenozooid that
is budded interzooidally or adventitiously by the maternal
zooid. In these cases, the whole (or nearly the whole) of
the kenozooid constitutes the ooecium, and the floor of the
brood chamber is provided by the maternal zooid (Fig. 14;
Ostrovsky, 2013). Here, a ‘type 1’ ooecium is an adventitious
appendage of the distal daughter. This makes ‘type 1’ ooecia
similar to adventitious spines, which have been documented
on autozooids as well as kenozooids (e.g. Petalostegus bicornis,
Chorizopora ferocissima; see Section IV.4; Gordon, 1984) and
avicularia (e.g. Valdemunitella fraudatrix,Corbulella translucens;
Gordon, 1984, 1986). By contrast, a ‘type 2’ ooecium is a
complete kenozooid and is a full polymorphic zooid similar
to adventitious avicularia or kenozooidal spines in Bellulopora
(Ostrovsky & Taylor, 2005a). Regardless of ooecium type,
the ovicell represents a multizooidal complex in which the
ooecium originates from a distal zooid while the eggs and
closing structure (when present) originate from the maternal
zooid (Lidgard et al., 2012; Ostrovsky, 2013).
Other brood chambers are formed entirely by overarching
spines (acanthostegal brood chambers), or by ooecia
constructed from multiple lobes or plates. Acanthostegal
brood chambers are multizooidal complexes, but represent
an alternate path in brood-chamber evolution rather than an
early stage of modern ovicells (Ostrovsky, 2013). Multilobed
ooecia can be formed from kenozooids budded by the
maternal zooid (‘type 2’ ooecia, as in Catenicula)orby
projections of the maternal frontal wall (as in Thalamoporella)
(Ostrovsky, 2013). This makes the ovicells of Thalamoporella
unique: they are not multizooidal complexes but integrated,
single-zooid structures.
Internal brood chambers (endozooidal) are contained
entirely within the maternal zooid. The walls and floor of the
brood cavity are formed from an uncalcified invagination
of the distal wall (Ostrovsky, 2013). Brood cavities open
to the environment are closed by an ooecial vesicle (likely
homologous to that in ovicells), while those opening to the
vestibulum may have a flap to isolate the brood cavity
(Ostrovsky, 2013). The ability to brood internally relies
partly on the ability to ‘make room’ for the developing
embryo within the cystid of the maternal zooid. This can
result in maternal zooids with enlarged cystids (e.g. Chlidonia
pyriformis), but this does not necessarily need to occur (e.g.
Watersipora subtorquata) (Ostrovsky, 2013). In both cases the
maternal zooid may be considered polymorphic owing to the
presence of a brood cavity (internal cystid modification).
(a)Evolution of brood chambers
The ooecium likely evolved from spines on the proximal
opesial rim of the daughter zooid (Fig. 15; Ostrovsky &
Taylor, 2005a, 2005b), instead of spines from the maternal
zooid as postulated by Sil´
en (1977). In addition to protecting
the frontal membrane, spines of a distal zooid may have also
protected membranous ovisacs produced by a proximal zooid
(Ostrovsky, 2013). Increased embryo protection could arise
from curving spines towards the proximal ovisac/maternal
zooid. This was likely the first step to a complete ooecium
and ovicell (Ostrovsky & Taylor, 2005a). Following that,
fusion (or reduction in number) and flattening of spines, loss
of articulation and relocation of spine bases, filling openings
in the ooecium, and forming a concave ovicell floor are all
trends toward the evolution of non-spinose ovicells (which
have evolved at least twice; Ostrovsky & Taylor, 2005a).
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 21
bf MZ
en ov
bf MZ
bf MZ
Fig. 14. Type 1 (appendage) and type 2 (zooidal) ooecia. (A) Type 1 ooecium of Callopora lineata, provided by a distal autozooid,
(B) type 1 ooecium of Tegella armifera, provided by a distal kenozooid, (C) type 2 ooecium of Cribrilina annulata, (D) type 2 ooecium
of Cauloramphus spinifer. In A and B the ooecium is an appendage of the daughter zooid, the brood chamber floor is provided by the
daughter zooid, and the ovicell is closed by an ooecial vesicle (acleithral). In C and D the ooecium is a kenozooid budded by the
maternal zooid, the brood chamber floor is provided by the maternal zooid, and the immersed ovicells are closed by the operculum
and ooecial vesicle (cleithral). bc, brood chamber; bf, brood chamber floor; DZ distal zooid; ec, ectooecium; en, endooecium; MZ
maternal zooid; op, operculum; ov, ooecial vesicle; ts, tentacle sheath. Calcified walls in black, membranous walls in grey. Drawn
from Ostrovsky (2013).
(A) (B) (C) (D)
Fig. 15. Steps in the evolution of the ooecium from periopesial spines. (A) Distelopora bipilata,(B)D. spinifera,(C)Gilbertopora larwoodi,
(D) Wilbertopora mutabilis. Redrawn from Ostrovsky & Taylor (2005b).
In addition to the formation of non-spinose ovicells, there
is trend towards the immersion of the brood chamber in
the maternal zooid (Ostrovsky et al., 2008; Ostrovsky, 2013).
Ovicells can be hyperstomial (prominent: more than half of
the brood cavity is above the frontal surface), subimmersed,
or immersed (entirely below the surface; Ostrovsky et al.,
2008). As the brood cavity becomes immersed, the ooecium
is reduced to a caplike, vestigial structure and the walls and
floor of the brood cavity become uncalcified (Ostrovsky,
2013). It is highly probable that uncalcified internal brood
sacs were derived (multiple times) from immersed ovicells
(Ostrovsky, 2013). Internalization of the brood chamber also
results in the increased integration of the cormidium through
the reduction of the component zooid: the brood chamber
is no longer a multizooidal complex (ovicell) but a structure
formed by a single maternal autozooid.
Lecithotrophic larvae typically have higher survival rates
than planktotrophic larvae (Mercier, Doncaster & Hamel,
2013), but brooding lengthens their exposure to motile
benthic predators. In the absence of protective structures,
predation on embryos and larvae is stronger in the benthos
than in the plankton (Allen & McAlister, 2007). Benefits of
yolky larvae are realized only if they survive the brooding
stage. Therefore, the construction of a calcified ovicell and the
immersion of the brood cavity were likely driven by predation
from small epibionts (Lidgard et al., 2012; Ostrovsky, 2013).
The development of ovicells and lecithotrophic larvae likely
coincided, and poorly dispersing larvae may have resulted
in population fragmentation and increased speciation, thus
triggering the cheilostome radiation (Ostrovsky & Taylor,
2005a; Lidgard et al., 2012).
(3) Ctenostome brood chambers
Ctenostomes may brood embryos externally or internally.
Most external brood chambers in ctenostomes are simply the
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
22 Carolann R. Schack and others
fertilization envelope of the embryo and therefore are not
polymorphic structures of the maternal zooid. This envelope
may attach directly to the surface of the maternal zooid [e.g.
Triticella koreni and Paludicella articulata (Str ¨
om, 1977)] or may
be attached by flexible mucus [Pottsiella erecta (Smith, Werle &
Klekowski, 2003)]. Most ctenostomes house the developing
embryo in a vestibule (e.g. Bulbella abscondita), tentacle sheath
(e.g. Amathia verticillata), or the diaphragm between the two
(e.g. Alcyonidium duplex) (Str¨
om, 1977; Micael et al., 2018), and
as such are not considered polymorphic.
Specialized, temporary structures may be developed in
maternal zooids after polypide degeneration. Maternal
zooids in Amathia gracilis develop ‘ciliated gutters’ to move
the oocyte to the tentacle sheath (Reed, 1988), while other
species develop ‘brood pouches’ between the degenerated
tentacle sheath and vestibule [e.g. Alcyonidium polyoum (Str¨
1977)]. After adhering to the maternal zooid, embryos
may also be brooded in epithelium-closed invaginations
of the vestibule body wall (as in Victorella muelleri and
Nolella dilatata;Str
om, 1977). Since these modifications are
temporary, maternal zooids producing specialized structures
are reversible polymorphs.
While most ctenostomes exhibit brooding methods that
do not require polymorphic zooids, some shell-boring
species possess ovicells or gonozooids (Soule, 1950a, 1950b;
Pohowsky, 1978). These structures exist in the genus
Penetrantia, the only operculate ctenostome genus, and the
fossil species Spathipora cheethami (Pohowsky, 1978). The brood
chambers are globular, cuticular structures (Sil´
en, 1946; plate
12 in Pohowsky, 1978). In P. concharum and P. sileni the brood
chamber opens into the tentacle sheath, while in P. densa it
opens directly to the environment via a widened aperture
en, 1946; Soule, 1950b; plate 11 in Pohowsky, 1978).
Like other internally brooding ctenostomes, the polypide
degenerates (Sil´
en, 1946; Soule, 1950b), potentially to supply
extraembryonic nutrition. However, maternal zooids hosting
these brood chambers are consistent polymorphs owing to
their modified polypides: while there are 12 tentacles in
normal autozooids, maternal zooids only possess eight (Sil´
1946). The brood chambers themselves, unlike cheilostome
ovicells, are probably non-kenozooidal appendages of the
maternal zooid. The evolutionary drivers for brood chambers
in Penetrantia are unclear, especially since internal brooding
is possible in other boring bryozoans (Soule, 1950a).
(4) Cyclostome gonozooids
Gonozooids are the external brood chambers of cyclostomes.
Unlike the multizooidal brood chambers of cheilostomes,
gonozooids are formed through the modification of a
single zooid and are less numerous within the colony. The
gonozooid consists of a proximal undifferentiated section,
an enlarged area forming the brood cavity where the
embryos are incubated, and the brood chamber aperture
(ooeciostome) homologous to the autozooidal peristome
om, 1977; Sch¨
afer, 1991). Typical cyclostome gonozooids
are lacking in only two families: Cinctiporidae, which has
very large autozooids (Boardman, McKinney & Taylor,
Fig. 16. Club-shaped gonozooid of Filicrisia geniculata. Az,
autozooidal peristomes; Gz, gonozooid; Op, ooeciopore, or
gonozooidal aperture; Os, ooeciostome, which is homologous
to the zooidal peristome. Drawn from Jenkins et al. (2017).
1992) that might function reproductively (unconfirmed)
and Anyutidae, which has autozooids possessing modified
peristomes with inferred reproductive function (Grischenko,
Gordon & Melnik, 2018).
Spatial competition between gonozooids and autozooids
results in a variety of gonozooid forms. The simplest
of these have pyriform (pear-shaped) brood cavities
(Fig. 16). In Tubuliporina, the expansion of the pyriform
brood cavity results in the dwarfism or elongation of
neighbouring autozooids. Enlargement of elongate/pyriform
brood cavities in Cerioporina and Eleidae results in the
occlusion of both autozooids and mesopores (Sch¨
afer, 1991).
Since feeding is vital to colony survival, it is advantageous
to limit obstruction of autozooids by brood-cavity growth
(Fig. 17). Loosely arranged zooid apertures (e.g. Articulata),
dorsal budding (Tervia: Tubuliporina), and peristomial
budding (only in Stomatopora gingrina: Tubuliporina) all limit
the interference of pyriform brood cavities and allow them to
expand freely. Grouping autozooidal apertures into fascicles
(tight bundles) can also promote this (e.g. Fasciculipora ramosa),
but zooids may still be occluded (Sch¨
afer, 1991). To avoid
autozooid occlusion, gonozooids may expand laterally by
forming lobes between aperture rows or may become
perforated by autozooids (which form their peristomes on
the brood cavity roof). Since lobate and perforated forms
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 23
Fig. 17. Irregular gonozooid of Tubulipora plumosa. Az,
autozooid; Gz, gonozooid; Op, ooeciopore. Drawn from Jenkins
et al. (2017).
are common in colonies with alternating zooid rows and
quincunxially arranged zooids respectively, the shape of
a brood cavity may be controlled by the organization of
neighbouring zooids (Sch¨
afer, 1991).
Space-filling polymorphic structures are typically occluded
by the expanding brood cavity. However, some gonozooids
incorporate these structures into their brood cavities. In
Petaloporidae (Cancellata) the roof and walls of brood
cavities are covered in mesopores, while in Discocytis
canadensis (Cytididae: Cancellata) the brood cavity walls
contain cancelli. Most striking are the brood cavities of
Patinella (Lichenoporidae: Rectangulata): the fertile zooid
grows alongside extrazooidal skeletal structures (alveoli), and
eventually the walls separating the fertile zooid and the
alveoli are reabsorbed to create a confluent brood cavity
(Fig. 18; Sch¨
afer, 1991). The gonozooids of Patinella are
therefore ‘multimodular’ structures, similar to the ovicells of
Cheilostomata (although the potential extrazooidal character
of the alveoli means that they may not be multizooidal
Gonozooids are irreversible polymorphs. Some
egg-producing autozooids are transformed to gonozooids
initially possessing a fully formed or non-feeding polypide
(with short tentacles and a reduced digestive tract) that is
everted from the orifice (Borg, 1926; Reed, 1991). It is likely
that a tentacle crown is necessary to collect sperm (Sil´
1972), potentially through the supraneural pore as in gym-
nolaemates (Ryland, 1976). Once an egg in a gonozooid is
fertilized, the polypide degenerates and the embryos receive
nutrition from a mass of cells generated by the surrounding
membranous sac (Borg, 1926; Str¨
om, 1977; Reed, 1991).
Regardless of shape, gonozooid walls are highly punctate,
potentially ensuring adequate respiration for the embryos
Fig. 18. Brood chamber development of the free-walled
cyclostome Patinella radiata. Interzooidal communication pores
are not shown. (A) The female zooid is separated from adjacent
alveoli by skeletal walls, (B) a large brood chamber is constructed
through the reabsorption of skeletal walls of the female zooid and
adjacent alveoli. The brood chamber possesses a calcified roof
with additional alveoli developing on the surface. A, ancestrula;
Al, alveolus; Az, autozooid; Bc, brood chamber; F, female zooid;
H, hypostegal coelom; Oe, ooeciostome; R, sites of reabsorption.
Drawn from Sch¨
afer (1991).
within (Ryland, 1976). Cyclostomes have polyembryonic
development, so the primary embryo undergoes cloning by
multiple fission (Borg, 1926). Lecithotrophic larvae (Str¨
1977) are released from the gonozooid via the ooeciostome,
which is curved to direct larvae away from the feeding
currents of autozooids (Sch¨
afer, 1991). Larvae are released
over an extended period (69 days in Filicrisia geniculata),
suggesting prolonged budding of the primary embryo
(Jenkins et al., 2017).
Gonozooids are expensive to maintain. It is unlikely that a
single degenerated polypide could provide sufficient nutrients
for the development of over 150 larvae (Kluge, 1962, in
Jenkins et al., 2017) during an extended period. Therefore,
the gonozooid and embryonic multiplication and growth
must be supported via resource transfer from autozooids
(Jenkins, Bishop & Hughes, 2015). The high cost of brooding
may limit the number of gonozooids present in a colony
(Hughes et al., 2005; Jenkins et al., 2015), and species of
cyclostomes reduce female investment in the absence of
alien sperm. When reproductively isolated, Tubulipora plumosa
reduced both gonozooid production and the number of
larvae (produced through intraclonal self-fertilization), while
female Filicrisia geniculata formed incomplete gonozooids
(Jenkins et al., 2015).
Polyembryony is a paradoxical strategy because it
combines the disadvantages of both cloning and sexual
reproduction (Hughes et al., 2005). However, polyembryony
may be advantageous for cyclostomes. Bryozoans broadcast
their sperm in the water column, resulting in severe sperm
dilution and rare fertilization events. Polyembryony may
allow colonies to capitalize on infrequent fertilization events
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
24 Carolann R. Schack and others
(Ryland, 1976), potentially aided by retention of allosperm
in the feeding current (Hughes et al., 2005). The extended
release of cloned larvae would allow a genotype to be tested
over a variety of temporal environments, some of which may
be more favourable for larval survival (Jenkins et al., 2017).
(5) Brood chambers in other stenolaemates
Structures resembling brood chambers exist in some
Paleozoic stenolaemates. Zooids with ovicell-like swellings
have been found in fenestrate colonies. These brood
chambers are associated with a single zooid and may
have retained a feeding polypide. Since these structures
are small and numerous they probably each contained a
single embryo, rather than being sites of polyembryony
afer, 1991). Cystoporate brood chambers are more
similar to cyclostome gonozooids with enlarged ‘brood
cavity’ sections likely lacking a feeding polypide, but
were unlikely to be polyembryonic (Sch¨
afer, 1991). While
it is difficult to determine when polyembryony evolved,
the presence of brood chambers in Paleozoic taxa
indicates that lecithotrophic larvae were already present
afer, 1991).
Bryozoan colonies may reproduce asexually. This is best
known in the Phylactolaemata, which produce statoblasts
(Wood, 1983). Statoblasts develop from epithelial cells in the
funiculus (Wood, 1983; Mukai et al., 1997). Once mature,
statoblasts detach from the funiculus and either remain in
the coelom until the colony dies or are released through a
temporary pore (Wood, 1983). Their chitinous outer layer
allows statoblasts to survive harsh conditions [desiccation,
ingestion (Wood & Marsh, 1996; Figuerola et al., 2004)]
thus helping to maintain the population. Statoblast variants
include dispersing floatoblasts and non-dispersing sessoblasts,
and both typically go through a dormant phase before
germinating (Wood, 1983).
While statoblasts are produced inside a zooid, asexual
propagules in gymnolaemates are budded from other zooids.
Asexual propagules in ctenostomes are called hibernacula
(Fig. 19). These structures are budded like zooids but lack a
polypide. Instead, they are filled with granular reserves that
are coloured like embryos and may consist of similar material
(Jebram, 1975). These granular reserves may also be similar
to the ‘nutrient storage cells’ associated with the embryos
and funicular system of some cheilostomes (Ostrovsky, 2013).
Since hibernacula lack a polypide, and secondary formation
of hibernacula after polypide degradation has not been
reported, nutrient reserves must be supplied by neighbouring
Like sessoblasts, hibernacula have strengthened outer
walls, remain cemented to the substratum after the death of
the colony, and are only dispersed through colony breakage
(Jebram, 1975; W¨
oss, 1996). Germination of hibernacula
may be triggered by changing environmental conditions
(e.g. increasing temperature; Carter et al., 2010b). During
their germination, the granular reserves become transparent,
suggesting that they are metabolized autozooids or stolons
(Jebram, 1975). Despite evidence for seasonal changes in
hibernacula production (Carter et al., 2010b), temperature
and salinity do not have a significant influence (Jebram,
1975). Instead, insufficient or low-quality food can trigger
the formation of hibernacula (forming after only a day of
starvation). This suggests that seasonal variations in food
supply, rather than the direct influence of abiotic factors,
drive colonies to produce hibernacula (Jebram, 1975).
‘Sac-zooids’ are found in shell-boring ctenostome genera
(Ropalonaria,Spathipora,andTerebripora). These kenozooidal
structures are filled with granules (Pohowsky, 1978),
suggesting a nutrient-storage function and possible homology
with hibernacula. A nutrient-storage function would allow
a colony to survive during periods of food shortage, but
whether a sac-zooid could bud a new zooid after colony
death is unclear. Such an ability would allow a colony to
remain in a favourable substratum and would make the
distinction between hibernacula and sac-zooids irrelevant.
In the ctenostome Timwoodiellina natans, small buds break
off from the colony and swim using their tentacle crown.
However, nautizooids are probably produced through sexual
reproduction and are not polymorphic zooids (Wood &
Okamura, 2017).
In Cheilostomata, dispersing asexual propagules (‘sacculi’
occur only in Aetea. Species in this genus are often epiphytes
on seagrass (Balduzzi et al., 1991). These are laterally budded
from the autozooid. Sacculi have a small stalk that widens
into a punctate body that is either narrowly ovate with a
single adhesive tip (Type A) or obtriangular and forked into
thin appendages, each of which possess an adhesive tip (Type
B) (Simma-Kreig, 1969; Balduzzi et al., 1991). Their small
size suggests that they have a reduced polypide or lack one
entirely. After breaking off from their parent zooid, sacculi
attach to the substratum and generate a new autozooid
(Simma-Kreig, 1969; Balduzzi et al., 1991).
While hibernacula and sessoblasts typically ensure
recolonization of an old substratum, dispersing sacculi and
floatoblasts enable the colonization of new substrata. This
suits the ephemeral nature of the seagrass fronds on which
sacculiferous Aetea typically grow (Pergent et al., 2008) and is
a more viable strategy than over-wintering.
The swollen attachment kenozooids of Caulibugula tuberosa
(Cheilostomata) may function similarly to hibernacula by
storing nutrients and allowing the colony to regrow after
annual dieback (Hastings, 1939).
Although some phylactolaemates can produce both
floatoblasts and sessoblasts (e.g. Plumatella;W
oss, 2008), the
ability to generate both dispersing and non-dispersing asexual
propagules has not been found in gymnolaemates. This may
be related to the freshwater habitat of phylactolaemates: lakes
and rivers undergo strong seasonal variation (maintenance
challenge) and may pose difficulties for colonization (dispersal
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 25
Az Az
Fig. 19. Asexual propagules. (A) Type A sacculus of Aetea lepadiformis, (B) type B sacculus of A. lepadiformis,(C)Paludicella articulata
zooid budding a hibernaculum, (D) base of a Caulibugula tuberosa colony with over-wintering stolons. Az, autozooid; H, hibernaculum;
Ns, nutrient-storing stolon; S, structural stolon. A and B drawn from Balduzzi, Barbieri & Gristina (1991), C from Rogick & van der
Schalie (1950), D from Hastings (1939).
(1) Evolution of asexual propagules
Asexual propagation is important for bryozoans in unstable
conditions (Okamura & Hatton-Ellis, 1995; Carter et al.,
2010b). The statoblasts of phylactolaemates are not
polymorphic zooids, since they are budded within a zooid. By
contrast, hibernacula and sacculi both bud from zooids and
are considered zooidal. In some species of Amathia,stolons
can also become filled with yolky reserves, and function in
a similar manner to hibernacula [as in Bowerbankia (Jebram,
1975)]. This suggests that hibernacula may have evolved
from the kenozooidal stolon.
Kenozooids per se have not been described in Aetea
(Hayward & Ryland, 1998), so sacculi may have arisen
directly from autozooids. The homology of sacculi with
autozooids (Sil´
en, 1977) is supported by the existence of ‘free
autozooids’ in Aetea. Free autozooids do not cement to the
substratum, but are capable of budding normal encrusting
zooids (Simma-Kreig, 1969). It is unclear whether these free
autozooids possess a feeding polypide. Like sacculi, the free
autozooids are punctate, exhibit a different colour to the rest
of the colony, attach to the substratum via tubular projections
instead of the main body, and may be laterally budded
(Simma-Kreig, 1969). Perhaps free autozooids represent an
intermediate step between sacculi and regular autozooids.
(1) Nanozooids
Nanozooids are dwarf zooids unique to cyclostomes and
possess a reduced, but protrusible, polypide. Primary
nanozooids are restricted to the cyclostome genus Diplosolen
en & Harmelin, 1974; Taylor & Weedon, 2000). They
possess a single, long unciliated tentacle (Sil´
en & Harmelin,
1974; Fig. 20) that is responsive to touch (indicating a
potentially sensory function) and is able to remove detritus
from the colony surface. A cleaning function would make
the primary nanozooid a functional analog of vibracula
en & Harmelin, 1974). In Diplosolen obelium, the primary
nanozooids are regularly budded with autozooids (Sil´
& Harmelin, 1974) and together may be considered a
cormidium composed of two submodules.
Unlike primary nanozooids and other heterozooids,
secondary nanozooids are originally budded as autozooids
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
26 Carolann R. Schack and others
300 um
Fig. 20. Cyclostome nanozooids. (A) Diplosolen sp. with primary nanozooids, (B) Plagioecia sp. with secondary nanozooids, (C)
Favosipora candida with secondary nanozooids (top-down view). Alv, alveolus; Az, autozooid; AzL, living autozooids; AzD, autozooids
with degenerated polypide and orifices sealed by a calcified terminal diaphragm; PNz, primary nanozooids; SNz, secondary
nanozooids. A and B drawn from Figs 11 and 16 in Sil´
en & Harmelin (1974), photograph in C taken by D.P.G.
en & Harmelin, 1974) and are an example of ontogenetic
polymorphism (Taylor & Weedon, 2000). When the polypide
in these autozooids degenerates, the peristome falls off and
the original orifice is sealed by a terminal diaphragm that
is typically calcified. In secondary nanozooids, this terminal
membrane hosts a secondary orifice through which a severely
shortened unciliated tentacle is protruded (Fig. 20; Sil´
en &
Harmelin, 1974). The function of secondary nanozooids is
particularly obscure: they cannot feed (tentacle unciliated)
or clean (tentacle too short), they are not reproductive
(no male sex cells observed), and no secretory glands or
special organs have been observed (Sil´
en & Harmelin, 1974).
Instead, secondary nanozooids may have a sensory function.
Although the tentacle is short, it remains protruded for
long periods, even when autozooids’ tentacle crowns have
been retracted (Sil´
en & Harmelin, 1974). This behaviour
would enable the colony to gather sensory information
without danger to the autozooids. Either way, secondary
nanozooids may have given rise to primary nanozooids
en & Harmelin, 1974), and similar structures are found in
other cyclostomes and stenolaemates (e.g. Bancroft, 1986).
Nanozooids may also be generated in response to
disturbances in growth, whether external (e.g. substratum
irregularities), or because of reduced space from colony
structuring (e.g. swelling of gonozooids, see Section V.4)
en & Harmelin, 1974). These irregularly formed
structures can resemble primary nanozooids or secondary
nanozooids (e.g. Plagioecia sarniensis and P. dorsalis,
respectively), although a protrusible polypide has not been
observed (Sil´
en & Harmelin, 1974).
(2) Zooeciules
Zooeciules are dwarf zooids in the cheilostome superfamily
Hippothooidea (particularly in Hippothoa,Plesiothoa,and
Trypostega). They possess an orifice (sometimes with condyles)
and an operculum, suggesting that they at least possess
opercular muscles and a polypide to create the orificial
aperture (Lutaud, 1983) (Fig. 21). Zooeciules can replace
autozooids (budding distally or laterally), bud laterally on
cauda, or adventitiously on the frontal wall (D.P.G, personal
observations). They may also bud autozooids, female zooids
or other zooeciules (Hastings, 1979). Although zooeciule size
and shape varies, the orifice remains minute (Osburn, 1952)
so the structure is unlikely to contain a protrusible polypide.
Zooeciules may be male zooids, but definitive evidence
for this (e.g. sperm, protrusible tentacles) is lacking (Hastings,
1979), despite the presence of dimorphic females (Hastings,
1979; Hayward & Ryland, 1999). Instead, zooeciules may be
analogous to avicularia (Osburn, 1952). Vicarious avicularia
have been found in species with zooeciules (Hippothoa distans;
Hastings, 1979). These avicularia are larger than autozooids
and have extensive musculature, but their mandibles are
relatively unmodified. If vicarious avicularia were developed
first, then zooeciules may be more derived forms of avicularia
(Osburn, 1952).
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 27
500 µm
Fig. 21. Zooeciules of Plesiothoa trigemma. Az, autozooid; Zo,
(1) Origin and evolution of polymorphism
Modularity sets the stage for differentiation and polymor-
phism. Modularity increases adaptive potential by allowing
modules to experience selection separately and by limit-
ing the damage from deleterious mutations (Kirschner &
Gerhart, 1998; Carroll, 2001). Adaptive potential is further
enhanced through function-preserving redundant modules
(Kirschner & Gerhart, 1998; Budd, 2006).
Novel phenotypes are generated from new genotypes
or the plastic (not necessarily adaptive) response of the
original genotype to a new environment (West-Eberhard,
2003). Since zooids in a bryozoan colony are genetically
identical, a genotypic change will essentially affect all colony
members. While new genotypes create novel ‘colony-wide’
phenotypes, the development of zooidal polymorphism
requires phenotypic variance among colony modules. This
variation is provided by the plastic response of zooids to
microenvironmental changes within colonies (e.g. differences
in flow, substratum irregularities, crowding). In contrast
to ‘random’ variation resulting from the imprecision of
developmental machinery, plasticity provides predictable
variation: if the magnitude of each environmental effect and
the rules governing developmental responses are known, one
could predict zooid form.
Both new genotypes and plastic responses are important
to the origin of polymorphism: new genotypes can
result in novel structures or pathways, while plasticity
provides phenotypic variation within a genotype (i.e. a
single colony). Plastic phenotypes (including polymorphisms)
are generated through the modification of existing
structures (Sil´
en, 1977). This modification can involve
deletion, duplication, reorganization, fusion/segregation,
or heterochrony (changes in developmental timing)
(West-Eberhard, 2003). For example, the production of
kenozooids involves deletion of the polypide, the formation of
an avicularium likely involved heterochrony (enlargement of
operculum, reduction of polypide) (Harvell, 1994; Cheetham
et al., 2006), and the development of some ooecia and frontal
shields involved the fusion of spines (Dick et al., 2009; Lidgard
et al., 2012; Ostrovsky, 2013).
The changes in gene expression that control bryozoan
polymorphism are currently unknown. Hox genes control
oralaboral regionalization of body sections and are ideal
for examining differences in body-plan formation (Halanych
& Passamaneck, 2001). In hydroids, the development and
maintenance of specialized zooids may be controlled by
differential expression of the Hox gene Cnox-2 (Cartwright,
Bowsher & Buss, 1999). In gastrozooids (and solitary forms)
there is a low expression of Cnox-2 in the hypostome and
tentacular region, and a high expression in the body column.
While dactylozooids and gonozooids (expanded body
columns) have high Cnox-2 expression, tentaculozooids
(expended hypostome and tentacles) have low Cnox-2
expression (Cartwright et al., 1999). By analogy, this type
of differential Hox-gene expression may also play a role
in the development of bryozoan polymorphs. In addition
to Hox genes, differential expression of genes related to
polymorph-specific processes or structures [e.g. nematocysts
in siphonophore gastrozooids (Siebert et al., 2011); calcium
deposition by axial polyps in Acropora corals (Hemond,
Kaluziak & Vollmer, 2014); mandibular glands in termite
soldiers (Miura, 2005)] should also be investigated for
bryozoan polymorphs.
Once zooidal plasticity modifies colony fitness, selection
(or stochastic processes) can maintain plasticity or result in
the canalization of a response. Selection for intracolonial
plasticity can only occur if the benefit of possessing plastic
zooids is greater than the benefit of non-plastic zooids
exhibiting one level of the plastic response (Van Kleunen
& Fischer, 2005). For example: there are two non-plastic
variants in M. membranacea (always unspined or always spined)
and a plastic genotype where peripheral autozooids can
be induced to form spines (Harvell, 1998). In this case,
the benefit of only producing spines when needed should
be greater than the costs of plasticity: the lag (2days;
Harvell, 1984) between unspined and spined states leaves
the colony vulnerable to predation from nudibranchs, while
misinterpreting a cue can result in accidental expenditure of
resources [although accidental spine production may be rare
because of cue thresholds (Harvell, 1984)] (Van Kleunen
& Fischer, 2005). Additional costs of plasticity may include
increased developmental instability and the maintenance of
sensory structures to detect cues (Van Kleunen & Fischer,
2005). If plasticity is too costly, the responses of zooids may
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
28 Carolann R. Schack and others
become invariant to environmental cues (Pigliucci, Murren
& Schlichting, 2006). Unfortunately, the mechanisms of
canalization (i.e. genetic assimilation) are currently unknown
(Harvell, 1994; West-Eberhard, 2003; Pigliucci et al., 2006).
Complete canalization is impossible when the plastic
response of zooids produces non-feeding forms, as is the case
with kenozooids and avicularia. Instead, selection can change
the sensitivity of the response and remove intermediate
forms. As a hypothetical example, a colony produces a
continuous range of zooid types (X0–X10). Differing levels
of the cue across the colony produce different zooids
(none =X0, intermediate =X1–X9, high =X10), resulting
in a patchwork of responses within the colony. However, if
X1–X9zooids perform worse than X0and X10 zooids in
their respective tasks, selection may change the sensitivity
threshold so that only X0and X10 zooids are produced. This
would result in discrete zooidal plasticity, i.e. polymorphism.
While X0and X10 forms could become relatively invariant to
the external environment, internal rules would be necessary
to govern when and where each zooid type is produced
(e.g. to prevent an imbalance of feeding: non-feeding forms).
These internal rules govern the formation of cormidia and
the zooidal composition of colonies.
While complete canalization of non-feeding zooidal
polymorphs is not possible, canalization can be achieved
through the transformation of a zooidal polymorph to an
autozooidal appendage. Such a transformation is seen in
the diverse evolution of non-cribrimorph frontal shields via
overgrowth of the frontal surface by kenozooids, which
eventually lose their zooidal nature (Gordon & Voigt,
1996; Taylor, Casadío & Gordon, 2008). Similarly, spines
may have been derived from spine-like kenozooids. The
mechanisms underlying the change from zooid to appendage
are unclear, but likely involve strict cormidial assembly
rules, adventitious budding, and fusion of zooids to create
a confluent coelom. Regardless of the method, it provides
an ‘evolutionary workflow’ for developing novel structures:
a kenozooid is derived from an autozooid and modified
into a novel structure that is then incorporated into
an autozooid as an appendage. This transition increases
colonial integration, a trend also observed in the transition
from vicarious to adventitious avicularia. However, it is
unlikely that adventitious avicularia could ever become true
appendages (possessing a confluent coelom with the host
autozooid) since they require their own sensory mechanisms
and musculature [although autozooids’ musculature can
provide nodding behaviour in birds-head avicularia
(Kaufmann, 1971)].
Thus, the origin and continued evolution of zooidal
polymorphism in bryozoans requires: (i) modular units
with modifiable structures; (ii) plastic responses to cues
that are heterogenous across the colony; (iii) removal of
intermediate forms by changes to cue sensitivity thresholds;
and (iv) development of cormidial and colony-level
assembly rules to control polymorph budding. Whether
these rules hold true for other colonial taxa is worth
(2) The cormidium: zooid assembly rules
Colonies are built module by module, rather than through
‘top-down’ control exerted by the colony as a whole. This
means that bryozoan zooids must react individually to
stimuli (including interzooidal communication). However,
each zooid follows the same ‘instructions’, which results
the formation of secondary and tertiary structures through
module iteration (Hageman, 2003). Few reactions are
possible for zooids in monomorphic colonies although
diverse colony forms are still possible, but zooids in
polymorphic colonies require additional reactions (to various
cue thresholds) to assemble cormidia and organize different
cormidial types. These cormidial assembly rules reflect both
selection at the zooid level (microenvironmental) and at
the colony level (macroenvironmental). Ultimately, cormidia
and colony form are emergent properties of zooid modularity
and plasticity.
Cormidial assembly requires interzooidal communication
and may be loose or rigid. For example, two simple rules
govern zooid patterns in Hesychoxenia praelonga:(i) marginal
zooids must be kenozooidal; and (ii) only 2 –3 autozooids may
bud next to each other (Gordon & Parker, 1991b). Similarly,
cormidial assembly rules can ensure adequate water flow via
the production of maculae (Banta, McKinney & Zimmer,
1974; Anstey, Pachut & Prezbindowski, 1976). These
structures, found in some living and fossil stenolaemates
and cheilostomes, consist of autozooids, polymorphs, and
an extrazooidal skeleton (Banta et al., 1974; McKinney &
Jackson, 1989; Taylor & Weedon, 2000).
Cormidial assembly rules may also vary with environmen-
tal conditions (reflecting the plasticity inherent in polymor-
phism). For example, in Celleporella hyalina (Cheilostomata)
environmental stress induces formation of male zooids and
suppression of female zooids (Hughes et al., 2003). Allocat-
ing energy to male zooids may be a ‘reproductive bailout’
in conditions where the parent colony (and any brooded
offspring) have a high mortality risk (Hughes et al., 2003).
Aside from ‘reproductive bailouts’, cormidial assembly rules
in bryozoans may be invariant along macroenvironmental
gradients (Hughes & Jackson, 1990; Simpson et al., 2017).
In addition, colonies with multiple types of cormidia
require rules that dictate the organization of different
cormidia types. The most striking example is the cheilostome
Corbulipora tubulifera, which was thought to be three separate
species owing to its unique cormidial morphologies (Fig. 22).
The base of the colony is formed of cribrimorph autozooids
and rhizoids. From this, a kenozooidal stalk is produced
that eventually buds ovicelled autozooids with membranous
frontal walls and vicarious avicularia. These membranous
autozooids alternate with cribrimorph autozooids as the
colony grows (Bock & Cook, 1994). This complicated colony
structure must rely on a set of assembly rules to dictate the
organization of each cormidium type, and the location and
duration of cormidial phases (which could be invariant or
Cormidia are found in a variety of taxa besides
bryozoans. In most siphonophores, cormidial rules are so
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 29
(A) (B)
Fig. 22. Corbulipora tubulifera. (A) Colony structure, (B) bilaminate phase, (C) flustrine phase, (D) encrusting phase. Avi, avicularium;
Az, autozooid; Kz, kenozooidal stalk; Ovi, ovicell; Rh, rhizoid. Drawn from Bock & Cook (1994).
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
30 Carolann R. Schack and others
strict that they may be considered semi-canalized. For
example, the siphonophore Bargmannia elongata produces
polymorphic polyps through the division of a probud, which
always differentiates into a gastrozooid, a tentaculozooid,
a gonozooid, five bracts, and several small buds in
that order (Dunn & Wagner, 2006). By contrast, some
colonial hydrozoans have dynamic cormidial rules. The
hydroid Thecocodium quadratum has two polymorphs: a large
gastrozooid with a mouth, digestive ability, and no tentacles;
and a mouthless dactylozooid with four tentacles to capture
prey. Gastrozooids typically stimulate the production of
nearby dactylozooids and suppress the budding of nearby
gastrozooids. However, when gastrozooids are fed dead
material (capture by dactylozooids not required) then
dactylozooid production is suppressed and more gastrozooids
are budded, illustrating a dynamic response to environmental
conditions (Pfeifer & Berking, 2002). While eusocial insect
colonies do not have cormidia per se, they do regulate
the relative proportions of polymorphic castes. In stingless
bees (Melipona beecheii) workers execute females that develop
into extra queens (Wenseleers et al., 2004). However, insect
colonies are composed of modules that are not genetically
identical: this means that some taxa have caste ratios that are
more dependent on genotypes of colony individuals, rather
than cue-mediated regulation (Schwander et al., 2010).
(3) Incidence of polymorphism and degrees of
Many different degrees of modular organization exist. The
ability to evolve polymorphism in bryozoans may be tied
to the level of module dissociation, which may explain why
polymorphism is not present in all bryozoan classes (Ryland,
1979). Phylactolaemates lack polymorphism and have low
module dissociation, while gymnolaemates have the highest
level of polymorphism (Mukai et al., 1997). Stenolaemates
have a moderate degree of polymorphism, but varying levels
of module dissociation (lower in free-walled than fixed-walled
stenolaemates) (Ryland, 1979; Boardman, 1998). In addition,
bryozoan taxa have different levels of developmental
dissociation: the polypide is formed first in stenolaemates
and phylactolaemates, while in gymnolaemates the cystid
is formed first (Lutaud, 1983; Mukai et al., 1997). It
appears that module dissociation is tied to polymorphism in
bryozoans. However, a close examination of polymorphism
in different genera and quantitative variation in interzooidal
communication is warranted [see Bone & Keough, 2010 and
Miles et al., 1995].
Across colonial marine taxa, increased module com-
partmentalization may be related to the incidence of
polymorphism (Venit, 2007). For example, colonial ento-
procts have weakly dissociated modules [common vascular
system (Borisanova & Malakhov, 2011)] and lack polymor-
phism (Brusca & Brusca, 2003). Conversely, the strongly
compartmentalized zooids in doliolids (Thaliacea) can be
polymorphic: in the asexual pelagic phase of their life cycle,
doliolid colonies have feeding gastrozooids and reproductive
phorozooids that detach and bud gonozooids (Alldredge &
Madin, 1982). This polymorphism may result from nutri-
ent transduction across a ‘placenta-like’ membrane between
adjacent zooids (Venit, 2007).
However, weak compartmentalization of modules does not
prevent the evolution of polymorphism. Colonial hydroids,
siphonophores, and sea pens all have common gastrovascular
cavities, and therefore have a level of module dissociation
similar to phylactolaemates. Despite this, these taxa can
be highly polymorphic, possessing polyps specialized for
reproduction, defence, feeding, colony structure, and asexual
propagation (Table 4) (Cartwright et al., 1999; Cartwright,
2003; Dunn, 2005; Dunn &Wagner, 2006; Williams
et al., 2012; Sanders et al., 2014). By contrast, scleractinian
corals have a wide range of modular dissociation: polyps
may be completely separated by the skeleton, share an
overlying tissue (cenosarc), or may possess gastrovascular
connections (coelenteric canals) through voids in the
skeleton (Swain et al., 2018). Despite this range in zooid
compartmentalization, polymorphism is rare and poorly
developed when present: apical polyps in Acropora colonies
are sterile, possess fewer tentacles and have lower symbiont
levels than gamete-producing radial polyps (Hemond et al.,
2014; Swain et al., 2018).
While strong dissociation of modules may increase the
ability to generate polymorphs, some module integration
is necessary to evolve non-feeding polymorphs: without
nutrient transfer at some stage, non-feeding forms are
impossible. Pelagic salps (Thaliacea), for example, form
chains of zooids that are attached together by special plaques
that can propagate nerve impulses (Mackie & Bone, 1977
in Mackie, 1986). However, the attachment plaques do
not transmit nutrients and colonies can easily break apart
in response to a stimulus (Mackie, 1986), indicating high
module dissociation. Pyrosome colonies (Thaliacea) also lack
vascular, muscular, and nervous connections between zooids.
As such, no polymorphic zooids have been found in salp or
pyrosome colonies (Venit, 2007).
Eusocial insect colonies are an exception to the rule
that colonies with fully compartmentalized modules lack
polymorphism. These colonies have modules that are
physically compartmentalized but behaviourally integrated.
This allows eusocial insects to produce non-feeding forms:
for example, in termites the queen may be unable to move
through the subterranean colony and workers must provide
food (e.g. Serritermes serrifer; Barbosa & Constantino, 2017).
Despite high compartmentalization, eusocial insects exhibit
only a moderate level of polymorphism (polymorphic workers
found in only 15% of ant genera and one bee species;
Wheeler, 1991; Gr¨uter et al., 2012).
While strongly dissociated modules may make it
easier physically to develop polymorphs, it is neither a
guarantee nor a requirement for their evolution. However,
complete compartmentalization of modules will prevent
polymorphism unless colony members are behaviourally
integrated. Trends in dissociation of modules and the
incidence of polymorphism in colonial invertebrates may
be hampered by differences in morphology across taxa.
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 31
Table 4. Polymorphism in colonial invertebrate taxa, showing level of module compartmentalization, level of polymorphism, and the functions the colony can perform. Functions fulfilled by structures that are not
equivalent to an individual module (e.g. non-zooidal structures) are indicated by an asterisk.
Phylum Taxa Compartmentalization Polymorphism Feeding Reproduction Structure Defence
propagules Other Source
Bryozoa Ctenostomata High Communication
pores plugged by
specialized cells,
associated with
funicular system
Moderate Autozooids Gonochoristic
zooids, ovicells
Spines Hibernacula Sil´
en (1977);
(1983); Wood
(1991); Taylor
& Weedon
Cheilostomata High high avicularia,
spines, stolons
sacculi Vibracula
Moderate Interzooidal
communication via
Moderate Gonochoristic
zooids, gonozooids
Spines, styles,
Low Confluent coelom Low– moderate Kenozooids,
Spines, styles
Phylactolaemata Low None Autozooids *Statoblasts
Entoprocta Coloniales Low Fluid circulates
through common
vascular system
None Zooids Brusca & Brusca
Borisanova &
Cnidaria Colonial hydroids
Low Common
cavity connected
via stolon network
Moderate Gastrozooids,
Gonozooid +
Medusa Cartwright et al.
tova &
Low Common
High Gastrozooids,
Gonozooid +
*Primary buds in
break off
Dunn & Wagner
Sea pens
Low Common
cavity, autozooids
linked by solenia
(lateral channels)
Moderate Autozooids Autozooids Oozooid Autozooids Acrozooids Siphonozooid,
Hoeksema &
Acropora hard corals
Low Common
cavity through
perforate skeleton
Low Radial zooids Apical zooids Hemond et al.
(2014); Swain
et al. (2018)
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
32 Carolann R. Schack and others
Table 4. Continued
Phylum Taxa Compartmentalization Polymorphism Feeding Reproduction Structure Defence
propagules Other Source
Echinophyllia aspea,
Low Confluent
cavity through
aligned costosepta
Zooids with polymorphic corallite
morphology (flat versus
protruded; polymorphic central
Arrigoni et al.
(2016); Swain
et al. (2018)
Other scleractinian
High– complete Zooids separated by
skeleton, may
share overlying
tissue (cenosarc)
None Zooids Swain et al.
Tunicata Doliolids (Thaliacea) High Zooids separated by
a permeable
Low– moderate Gastrozooids Gonozooids Oozooids Phorozooid Alldredge &
Madin (1982)
Salps (Thaliacea) Complete No vascular,
muscular, or
between zooids
None Zooids
Venit (2007)
Colonial ascidians
High– complete Zooids contiguous,
may share a
common tunic
None Blastozooids Sabbadin (1979);
Brown &
Swalla (2012)
Low Zooids share a
common tunic and
blood vasculature
Hemichordata Rhabdopleura
Complete Zooids thecate,
interzooidal stolon
restricted by
?None Zooids Gonochoristic zooids ?Hibernacula
(may be
Urbanek & Dilly
(2000); Maletz
?High Zooids thecate,
connected by a
stolon running
through a
common canal
?Unclear Autotheca ?Bitheca Maletz (2017)
Arthropoda Eusocial aphids
Complete Individuals not
Low Workers Stem mothers,
Soldiers Alate clones,
Wheeler (1991);
Tyerman &
(2004); Miura
(2005); Gr¨uter
et al. (2012)
Eusocial termites
alate (winged) or
ergatoid (wingless)
Eusocial ants and
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
Modularity is the mother of invention 33
To address this, a quantitative approach should be used
to determine module dissociation, e.g. by measuring
nutrient transduction (Mackie, 1986; Miles et al., 1995),
and to quantify morphological and genetic polymorph
differentiation (e.g. Hemond et al., 2014). Such a study may
reveal that other aspects of modularity are more important
than module dissociation. Although module dissociation may
be influential for Bryozoa (Ryland, 1979), it may have little
bearing on the level of polymorphism in other taxa.
(4) Types of polymorphs
While all zooidal polymorphism must arise from
morphological plasticity in autozooids (or other polymorphs),
the types of polymorphs are dependent on the modifiable
structures present within a zooid, and the functions required
by a colony. For example, definite mandibulate structures
(avicularia, eleozooids) exist only in bryozoan taxa that have
an operculum. The operculum and its associated musculature
are preadaptations for a grasping function (Winston, 2004)
and thus make mandibulate polymorphs more likely to
evolve in operculate taxa (Cook, 1968; Cheetham et al.,
2006). Similarly, the nematocyst-laden tentacles of zooids in
colonial cnidarians easily lend themselves to defence (e.g. the
sea pansy Renilla kollikeri; Kastendiek, 1976) and few ‘steps’
should be required to derive defensive polymorphs from them
(e.g. tentaculozooids and dactylozooids in colonial hydroids
and siphonophores). Once derived, polymorphs can be
further modified to perform different functions (e.g. cleaning
vibracula from defensive avicularia). Kenozooids and the
extrazooidal skeleton in particular provide modifiable
structures since they are no longer constrained by the
presence of a polypide. The ability to form a ‘free module’
may explain the high level of polymorphism in bryozoans,
even compared to polymorphic cnidarians.
The two functions every colonial organism must fulfil
are feeding and reproduction, and these are the most
common polymorphs (Table 4) particularly because the
‘machinery’ required to perform these functions is present
in monomorphic zooids. Specialized reproductive modules
have developed in most polymorphic colonial taxa except sea
pens [which are gonochoristic at the colony level (Edwards &
Moore, 2008; Williams et al., 2012)] and Acropora stony corals
[polyps are simultaneous hermaphrodites and spawn rather
than brood (Harrison, 2011)].
While polymorphs probably originate under heterogenous
microenvironmental conditions, their continued evolution
can be shaped by both micro- and macroenvironmental
stressors (in addition to stochastic processes). Defensive
polymorphs may arise in environments with many predators,
competitors, or abrasive substrata. Changes in morphology
may reflect environmental pressures for different functions
or increased efficiency [e.g. the potential latitudinal gradient
in avicularian mandible shape (Schopf, 1973)]. Similarly,
cleaning polymorphs could arise in areas with high
fouling and sedimentation rates, asexual propagules may
develop in response to harsh seasonal conditions, and
attachment structures should be developed based on
colony form and substratum type. However, changes along
macroenvironmental gradients may not affect the level
of polymorphism in bryozoans (Hughes & Jackson, 1990;
Simpson et al., 2017), even though high levels of stress can
trigger a polymorphic response [e.g. increased production of
male zooids (Hughes et al., 2003)].
It is assumed that division of labour increases efficiency
and is therefore favourable under selection (Nyhart &
Lidgard, 2011; Lidgard et al., 2012; West & Cooper, 2016).
Increased efficiency may stem from the elimination of
task-switching costs (Goldsby et al., 2012) in addition to
increased performance from physiological specialization.
Since many polymorphs are non-feeding, an increase in
efficiency would offset the cost of such structures, particularly
in the energetically marginal lifestyle of bryozoans [Thorpe,
1979; Lidgard (1981) in Gordon, Clark & Harper, 1987;
Lidgard et al., 2012]. However, the magnitudes of these
efficiency gains, if any, have not been determined. It is
possible that polymorphisms may provide neutral efficiency,
indicating a lack of directional selection. Interestingly,
polymorphic bryozoan colonies typically succeed ephemeral
non-polymorphic ones, suggesting that polymorphism
represents increased investment in colony survival and
competitive ability (Simpson et al., 2017).
This review has described the forms and functions of
bryozoan polymorphisms. Limitations on the realized
morphospace of bryozoan polymorphisms have yet to be
determined (Schack et al., 2018). Polymorphisms do not arise
de novo but are formed from the modification of fundamental
autozooid modules (Sil´
en, 1977). Modifications can be
made to the polypide (in reproductive polymorphs), the
cystid (appendaged autozooids) or both (heterozooids). The
realized morphospace of bryozoan polymorphism and the
extent to which modules can be structurally differentiated
may be limited by the structure of the base autozooid.
For example, the limited bryozoan nervous system may
prevent avicularia, which have intrinsic musculature, from
becoming ‘true’ limbs. The grasping long-stalked avicularia
in Camptoplites (Bugulidae) look like protolimbs, and can
nod using the musculature of the autozooid-provided
peduncle (Kaufmann, 1971) but the stalks themselves have
no independent power of mobility. While high module
dissociation may be related to the level of polymorphism in
bryozoans, this is not always the case for other colonial taxa
[e.g. colonial cnidarians (cf . Venit, 2007)]. The development
of polymorphism does rely on ‘true’ coloniality: modules
must be behaviourally or physically connected, otherwise
non-feeding polymorphs cannot be supported.
Most polymorphisms have unknown or highly debated
functions. It is important to note that a polymorph may have a
variety of sub-functions in addition to (although less efficiently
than) its main function (e.g. a knife can be used to drink
soup, albeit very inefficiently). Determining these functions,
Biological Reviews (2018) 000– 000 ©2018 Cambridge Philosophical Society
34 Carolann R. Schack and others
along with their scope and effectiveness, will require in
vivo observations and empirical testing. Understanding form
and function will provide information on polymorphism
diversity and their division of labour (e.g. potential
overlaps in function between polymorphisms). Combined
with evolutionary phylogenies, knowledge of function will
allow the construction of ‘differentiation sequences’ showing
whether certain polymorphs always evolve first or are more
likely to evolve (e.g. reproductive polymorphs are common
in most polymorphic colonial organisms).
Three interrelated aspects of polymorphism urgently
require further study: developmental mechanisms, ener-
getics, and traitenvironment relationships. Understanding
these, in addition to form and function, would contribute
greatly to our understanding of the evolution of modu-
larity, module differentiation, and the division of labour;
concepts which, in turn, are key to understanding the evo-
lution of complexity (Kirschner & Gerhart, 1998; Carroll,
2001). One hopes that interest in bryozoans and their
polymorphisms – will continue to increase, for it is clear
that modularity and polymorphism have been key to their
evolutionary success.
(1) Modularity is vital to the evolution of polymorphism:
the dissociation between modules allows heterogenous
selective pressures to act independently and facilitates the
division of labour.
(2) Module dissociation between zooids and dissociation
between the cystid and polypide may be related to the level
of polymorphism in bryozoans. However, this may not be
the case for other colonial organisms.
(3) Cormidia, the organization of cormidia types, and
colony form are emergent properties arising from the
modularity and plasticity of bryozoan zooids.
(4) Superficially similar polymorphs (e.g. spines, rhizoids,
space-filling polymorphs) can be kenozooidal, autozooidal
appendages, or extrazooidal structures. Careful study of their
development and internal anatomy is required to separate
the different types.
We thank Victoria University of Wellington for supporting
this work (VUW grant 80837, and doctoral scholarship to
C.R.S.) and the National Institute of Water and Atmosphere
Research for use of research facilities.