250
The development of the embryo outside the maternal zooid
allowed polypide regeneration, while spatial separation of
gametogenesis and brooding allowed multiple use of the
fertile zooid.
To begin with, this hypothesis is not supported by paleontological data – brooding in ovicells dates back to the Middle
Cretaceous whereas the fi rst epistomiids ( Synnotum ) are
known from Miocene deposits. Epistomiids have avicularia
similar to those of bugulids so their ancestor probably
belonged to the same superfamily (Buguloidea) possessing
extrazooidal brooding. Sexual dimorphism characteristic of
the Epistomiidae is also a derived character. Further, it is diffi cult to imagine a reason for the transition from viviparity
(generally considered as derived and an evolutionarily expedient form of parental care) to external brooding involving
egg transfer from the zooid. Any kind of internal brooding
ensures good protection of the embryo and the possibility of
forming a large larva. So even if epistomiids did originate
from a non-brooding ancestor, this was a cul-de-sac branch
in the evolution of brooding.
This family is much more likely to have evolved from
bryozoans with extrazooidal brooding (pattern II) by acquiring intrazooidal/intraovarian embryonic incubation. As with
the above-discussed transition from pattern IV to pattern III,
the mode of oogenesis shifted from macrolecithal to
microlecithal.
3.3.5 Adaptive Importance of Placental
Analogues in Cheilostomata
If, as argued above, placental analogues indeed evolved
numerous times (at least, 22) in bryozoans, the question
arises about the selective importance of such a feature.
Existing hypotheses reasonably consider placentation as a
byproduct of the evolution of parental care in Cheilostomata.
Santagata and Banta ( 1996 , р. 178) proposed a hypothesis,
according to which the earliest form of embryo incubation
was “vestibular brooding,” which resulted in the acquisition
of placental nutrition via the vestibular wall (see above).
Another hypothesis was suggested by Hughes ( 1987 ),
who thought that skeletal brood chambers initially were protective structures, later assuming the function of extraembryonic nutrition in some species. The structure of different
types of brood chambers, their distribution among cheilostomes as well as fossil evidence all point in favour of this
hypothesis.
Dyrynda and Ryland ( 1982 ), who described polypide
recycling in the maternal zooid during matrotrophic brooding in Bugula fl abellata , suggested that the “evolution of
embryonic placentation” (p. 255) provided an uninterrupted
nutrient supply to the embryo during periods when the feeding apparatus and gut (polypide) degenerated, thus supporting maximum larval production in bryozoans with ephemeral
colonies. However, in matrotrophic Beania bilaminata and
Watersipora subtorquata , and in all catenicellids and urceoliporids studied so far, the polypide never regenerates during
placentotrophic incubation. The same is true of species in the
family Epistomiidae (see above), in which uninterrupted
EEN is supported by intracolonial transport of nutrients via
funicular cords (Marcus 1941b ; Dyrynda 1981 ; Dyrynda and
King 1982 ). Thus, as with oogenesis, matrotrophic nutrition
occurs independently of the presence or absence of a functioning polypide (see Dyrynda and Ryland 1982 ; Dyrynda
and King 1983 ; Ostrovsky 1998c , 2009 ). This suggests a
high degree of colonial integration, enabling interzooidal
distribution of nutrients to non-feeding (including incubating) zooids. Thus, any connection between the evolution of
placentotrophy and polypide recycling is unlikely.
The role of EEN in accelerating embryogenesis, though
possible, is unknown. For instance, in matrotrophic
Celleporella hyalina the larva is incubated from 12–14 days
(Hughes 1987 ) to 3–4 weeks (Cancino and Hughes 1988 ).
Similarly larval development requires from 10–14 to 30 days
in non-matrotrophic cheilostomes studied (see Silén 1945 ;
Gordon 1977 ; Nielsen 1981 ). So, at present the data are too
few to draw fi rm conclusions.
It was suggested earlier that EEN affords simultaneous
embryonic development and growth and thus may accelerate the rate of reproduction in the early part of the process
(Ostrovsky et al. 2009a ). The fi rst small microlecithal oocyte
in the ovary of species with pattern III should theoretically
mature faster than the large macrolecithal egg in nonplacental brooders with pattern II. While the macrolecithal
oocyte is maturing, the microlecithal oocyte will be transferred to the ovicell, and the new egg will begin its formation in the ovary immediately after oviposition. In this
situation, the speed of embryogenesis would not be important. Of more importance is that the fi rst larvae will be
released earlier in matrotrophs since their oogenesis is
shorter. For instance, it takes about 4 weeks from the beginning of egg formation in the ovary until larval release in
Callopora dumerilii (Silén 1945 ) and 6 weeks in Chartella
papyracea (both non- matrotrophic cheilostomes) and just
3 weeks in matrotrophic Bugula fl abellata (Dyrynda and
Ryland 1982 ; Dyrynda and King 1983 ) and B . simplex
(Grave 1930 ; Ryland 1974 ). Mawatari ( 1951 ) found that B .
neritina released its fi rst larvae just 1 week after the fi rst few
ovicells were observed in the colony. Such a strategy (simultaneous embryonic growth and development) would benefi t
species with ephemeral colonies that live in seasonal waters,
allowing them to occupy free biotopes/niches because of
more rapid production of the fi rst generation of larvae.
Indeed, matrotrophy has been recorded in the families
Bugulidae, Beaniidae, Candidae, Flustridae, Cellariidae,
Poricellariidae, Catenicellidae, Urceoliporidae and
Epistomiidae, the species of which all possess erect weakly
calcifi ed colonies and many evidently live just a few months.
3 Evolution of Reproductive Patterns in Cheilostomata
The development of the embryo outside the maternal zooid
allowed polypide regeneration, while spatial separation of
gametogenesis and brooding allowed multiple use of the
fertile zooid.
To begin with, this hypothesis is not supported by paleontological data – brooding in ovicells dates back to the Middle
Cretaceous whereas the fi rst epistomiids ( Synnotum ) are
known from Miocene deposits. Epistomiids have avicularia
similar to those of bugulids so their ancestor probably
belonged to the same superfamily (Buguloidea) possessing
extrazooidal brooding. Sexual dimorphism characteristic of
the Epistomiidae is also a derived character. Further, it is diffi cult to imagine a reason for the transition from viviparity
(generally considered as derived and an evolutionarily expedient form of parental care) to external brooding involving
egg transfer from the zooid. Any kind of internal brooding
ensures good protection of the embryo and the possibility of
forming a large larva. So even if epistomiids did originate
from a non-brooding ancestor, this was a cul-de-sac branch
in the evolution of brooding.
This family is much more likely to have evolved from
bryozoans with extrazooidal brooding (pattern II) by acquiring intrazooidal/intraovarian embryonic incubation. As with
the above-discussed transition from pattern IV to pattern III,
the mode of oogenesis shifted from macrolecithal to
microlecithal.
3.3.5 Adaptive Importance of Placental
Analogues in Cheilostomata
If, as argued above, placental analogues indeed evolved
numerous times (at least, 22) in bryozoans, the question
arises about the selective importance of such a feature.
Existing hypotheses reasonably consider placentation as a
byproduct of the evolution of parental care in Cheilostomata.
Santagata and Banta ( 1996 , р. 178) proposed a hypothesis,
according to which the earliest form of embryo incubation
was “vestibular brooding,” which resulted in the acquisition
of placental nutrition via the vestibular wall (see above).
Another hypothesis was suggested by Hughes ( 1987 ),
who thought that skeletal brood chambers initially were protective structures, later assuming the function of extraembryonic nutrition in some species. The structure of different
types of brood chambers, their distribution among cheilostomes as well as fossil evidence all point in favour of this
hypothesis.
Dyrynda and Ryland ( 1982 ), who described polypide
recycling in the maternal zooid during matrotrophic brooding in Bugula fl abellata , suggested that the “evolution of
embryonic placentation” (p. 255) provided an uninterrupted
nutrient supply to the embryo during periods when the feeding apparatus and gut (polypide) degenerated, thus supporting maximum larval production in bryozoans with ephemeral
colonies. However, in matrotrophic Beania bilaminata and
Watersipora subtorquata , and in all catenicellids and urceoliporids studied so far, the polypide never regenerates during
placentotrophic incubation. The same is true of species in the
family Epistomiidae (see above), in which uninterrupted
EEN is supported by intracolonial transport of nutrients via
funicular cords (Marcus 1941b ; Dyrynda 1981 ; Dyrynda and
King 1982 ). Thus, as with oogenesis, matrotrophic nutrition
occurs independently of the presence or absence of a functioning polypide (see Dyrynda and Ryland 1982 ; Dyrynda
and King 1983 ; Ostrovsky 1998c , 2009 ). This suggests a
high degree of colonial integration, enabling interzooidal
distribution of nutrients to non-feeding (including incubating) zooids. Thus, any connection between the evolution of
placentotrophy and polypide recycling is unlikely.
The role of EEN in accelerating embryogenesis, though
possible, is unknown. For instance, in matrotrophic
Celleporella hyalina the larva is incubated from 12–14 days
(Hughes 1987 ) to 3–4 weeks (Cancino and Hughes 1988 ).
Similarly larval development requires from 10–14 to 30 days
in non-matrotrophic cheilostomes studied (see Silén 1945 ;
Gordon 1977 ; Nielsen 1981 ). So, at present the data are too
few to draw fi rm conclusions.
It was suggested earlier that EEN affords simultaneous
embryonic development and growth and thus may accelerate the rate of reproduction in the early part of the process
(Ostrovsky et al. 2009a ). The fi rst small microlecithal oocyte
in the ovary of species with pattern III should theoretically
mature faster than the large macrolecithal egg in nonplacental brooders with pattern II. While the macrolecithal
oocyte is maturing, the microlecithal oocyte will be transferred to the ovicell, and the new egg will begin its formation in the ovary immediately after oviposition. In this
situation, the speed of embryogenesis would not be important. Of more importance is that the fi rst larvae will be
released earlier in matrotrophs since their oogenesis is
shorter. For instance, it takes about 4 weeks from the beginning of egg formation in the ovary until larval release in
Callopora dumerilii (Silén 1945 ) and 6 weeks in Chartella
papyracea (both non- matrotrophic cheilostomes) and just
3 weeks in matrotrophic Bugula fl abellata (Dyrynda and
Ryland 1982 ; Dyrynda and King 1983 ) and B . simplex
(Grave 1930 ; Ryland 1974 ). Mawatari ( 1951 ) found that B .
neritina released its fi rst larvae just 1 week after the fi rst few
ovicells were observed in the colony. Such a strategy (simultaneous embryonic growth and development) would benefi t
species with ephemeral colonies that live in seasonal waters,
allowing them to occupy free biotopes/niches because of
more rapid production of the fi rst generation of larvae.
Indeed, matrotrophy has been recorded in the families
Bugulidae, Beaniidae, Candidae, Flustridae, Cellariidae,
Poricellariidae, Catenicellidae, Urceoliporidae and
Epistomiidae, the species of which all possess erect weakly
calcifi ed colonies and many evidently live just a few months.
3 Evolution of Reproductive Patterns in Cheilostomata