249
eggs (a hypothetical ancient variant of pattern II, resulting in
a non-feeding larva, see Sect. 3.4.1 ). Further evolution could
result in a transition from incipient to substantial placentotrophy with corresponding enlargement of both the embryo
and the brood chamber (pattern III). The transition to macrolecithal oogenesis accompanied (or not) by the enlargement
of oocytes could lead to pattern IV and ultimately to the suppression of matrotrophy, which could then fi nally disappear
through a shift to pattern II.
However, this alternative scenario is made doubtful by
two major arguments: the pattern of EEN distribution among
the taxa and the timing of their appearance in the fossil
record. First of all, though EEN has turned out to be commoner among Cheilostomata than previously thought, the
majority of them are still non-placental. It seems extremely
unlikely that matrotrophy evolved in the ancestral brooder,
became widespread and then was lost many times in different families. For instance, only one supposedly matrotrophic
Recent species ( Crassimarginatella falcata ) is known among
Calloporidae (Cook 1985 ), the earliest brooding family
known since the Albian (Late Cretaceous) and considered as
ancestral to a number of cheilostome lineages, including
microporids and cribrilinids. Yet the genus Crassimarginatella
itself is much younger, having evolved in the Danian (Early
Paleocene). There are similar additional examples in the
families Microporidae, Cribrilinidae, Poricellariidae and
Hippothoidae. Originating in the Late Cretaceous
(Cenomanian), the Cribrilinidae and Microporidae are two
other large “key” families supposedly ancestral to many of
the more advanced cheilostome lineages (Boardman et al.
1983 ; Gordon and Voigt 1996 ; Gordon 2000 ). In the
Microporidae, which probably evolved from a calloporid
ancestor, only two Recent species ( Micropora notialis and
Mollia multijuncta ) are known to possess EEN. Whereas
Micropora is known from the Cenomanian, Mollia is much
younger, having evolved in the Danian. In the largest
cheilostome family, Cribrilinidae, matrotrophy is suggested
in just one species ( Figularia fi gularis ), and the genus
does not appear until the Miocene. At the end of the
Cretaceous (Maastrichtian) two other genera – Poricellaria
(Poricellariidae) and Celleporella (Hippothoidae) – evolved
of which three Recent species are matrotrophic.
Secondly, although it is possible that placental analogues
are more widespread than is known for cheilostomes (sexual
reproduction has been studied anatomically in species from
less than 30% of all families), the paucity of matrotrophic
representatives among Recent genera of basal clades and the
large gaps between their time of origination suggest that placental analogues are unlikely to have evolved early and to
have achieved wide distribution in the Cretaceous.
Interestingly, the number of genera with proven or suggested
EEN increases considerably in the Tertiary. Eight genera
belonging to eight families are known since the Eocene
( Scrupocellaria , Beania , Cellaria , Figularia , Catenicella ,
Adeonellopsis , Hippopodina , Myriapora ), six other
genera from four families since the Miocene ( Costaticella ,
Adeonella , Laminopora , Synnotum , Watersipora , Urceolipora ), and one genus from the Oligocene ( Pterocella ). Nine
genera from fi ve families have no fossil record, i.e. they
either evolved relatively recently ( Retifl ustra , Isosecurifl ustra ,
Gregarinidra , Klugefl ustra , Bugula , Bicellariella , Epistomia ,
Cribricellina , Reciprocus ) or simply have not yet been discovered in the fossil record (in some cases because they are
lightly calcifi ed). It seems that matrotrophy was becoming
more and more common during cheilostome history, but again,
phylogenetic relationships between taxa including matrotrophic species, together with the distribution of reproductive
patterns, point to its independent origins (Ostrovsky 2013 ).
As a fi nal remark, in squamate reptiles, live-bearing (and,
subsequently, matrotrophy) was much more easily gained
than lost (Lee and Shine 1998 ), whereas in teleost fi shes
there is no evidence of such transitions (Goodwin et al. 2002 ;
Mank et al. 2005 ). Although hypotheses about reversals are
actively discussed, most authors tend to consider the acquisition of this novelty as a dominant trend in comparison to its
loss (Wourms and Lombardi 1992 ; Shine and Lee 1999 ;
Blackburn 1999c ; Reznick et al. 2007b ; Pollux et al. 2009 ;
see also Blackburn 2005a , b ). On the other hand, the maternal input can be highly labile. For instance, Dulvy and
Reynolds’s ( 1997 ) phylogenetic analysis showed 6-8 reversals from matrotrophic to lecithotrophic viviparity in elasmobranchs (see also Reynolds et al. 2002 ). Thus, estimates
of the number of independent origins of matrotrophy should
be combined with phylogenetic character mapping.
3.3.4 Origin of Viviparity in the Family
Epistomiidae
An exceptional case of independent evolution of EEN is represented by the viviparous family Epistomiidae, which possesses intraovarian matrotrophic incubation (reproductive
pattern V). In this group the origin of matrotrophy might
have been associated with the transition from incubation of
embryos in the brood chamber (for instance, in the internal
brood sac, as in Beaniidae) to embryonic development
directly in the ovary.
According to Dyrynda and King ( 1982 ), the “epistomiid” character state – intracoelomic incubation (larval
viviparity) and absence of polypide recycling – is the initial
variant, the basis for the subsequent evolution of extracoelomic brooding accompanied by degeneration-regeneration
of polypides. Their line of argument is clear – non-brooding ancestral forms with one polypide generation in the
zooid gave rise to the species with intracoelomic incubation
and no recycling and then extracoelomic brooding appeared.
3.3 Evolution of Matrotrophic Incubation in Cheilostomata
eggs (a hypothetical ancient variant of pattern II, resulting in
a non-feeding larva, see Sect. 3.4.1 ). Further evolution could
result in a transition from incipient to substantial placentotrophy with corresponding enlargement of both the embryo
and the brood chamber (pattern III). The transition to macrolecithal oogenesis accompanied (or not) by the enlargement
of oocytes could lead to pattern IV and ultimately to the suppression of matrotrophy, which could then fi nally disappear
through a shift to pattern II.
However, this alternative scenario is made doubtful by
two major arguments: the pattern of EEN distribution among
the taxa and the timing of their appearance in the fossil
record. First of all, though EEN has turned out to be commoner among Cheilostomata than previously thought, the
majority of them are still non-placental. It seems extremely
unlikely that matrotrophy evolved in the ancestral brooder,
became widespread and then was lost many times in different families. For instance, only one supposedly matrotrophic
Recent species ( Crassimarginatella falcata ) is known among
Calloporidae (Cook 1985 ), the earliest brooding family
known since the Albian (Late Cretaceous) and considered as
ancestral to a number of cheilostome lineages, including
microporids and cribrilinids. Yet the genus Crassimarginatella
itself is much younger, having evolved in the Danian (Early
Paleocene). There are similar additional examples in the
families Microporidae, Cribrilinidae, Poricellariidae and
Hippothoidae. Originating in the Late Cretaceous
(Cenomanian), the Cribrilinidae and Microporidae are two
other large “key” families supposedly ancestral to many of
the more advanced cheilostome lineages (Boardman et al.
1983 ; Gordon and Voigt 1996 ; Gordon 2000 ). In the
Microporidae, which probably evolved from a calloporid
ancestor, only two Recent species ( Micropora notialis and
Mollia multijuncta ) are known to possess EEN. Whereas
Micropora is known from the Cenomanian, Mollia is much
younger, having evolved in the Danian. In the largest
cheilostome family, Cribrilinidae, matrotrophy is suggested
in just one species ( Figularia fi gularis ), and the genus
does not appear until the Miocene. At the end of the
Cretaceous (Maastrichtian) two other genera – Poricellaria
(Poricellariidae) and Celleporella (Hippothoidae) – evolved
of which three Recent species are matrotrophic.
Secondly, although it is possible that placental analogues
are more widespread than is known for cheilostomes (sexual
reproduction has been studied anatomically in species from
less than 30% of all families), the paucity of matrotrophic
representatives among Recent genera of basal clades and the
large gaps between their time of origination suggest that placental analogues are unlikely to have evolved early and to
have achieved wide distribution in the Cretaceous.
Interestingly, the number of genera with proven or suggested
EEN increases considerably in the Tertiary. Eight genera
belonging to eight families are known since the Eocene
( Scrupocellaria , Beania , Cellaria , Figularia , Catenicella ,
Adeonellopsis , Hippopodina , Myriapora ), six other
genera from four families since the Miocene ( Costaticella ,
Adeonella , Laminopora , Synnotum , Watersipora , Urceolipora ), and one genus from the Oligocene ( Pterocella ). Nine
genera from fi ve families have no fossil record, i.e. they
either evolved relatively recently ( Retifl ustra , Isosecurifl ustra ,
Gregarinidra , Klugefl ustra , Bugula , Bicellariella , Epistomia ,
Cribricellina , Reciprocus ) or simply have not yet been discovered in the fossil record (in some cases because they are
lightly calcifi ed). It seems that matrotrophy was becoming
more and more common during cheilostome history, but again,
phylogenetic relationships between taxa including matrotrophic species, together with the distribution of reproductive
patterns, point to its independent origins (Ostrovsky 2013 ).
As a fi nal remark, in squamate reptiles, live-bearing (and,
subsequently, matrotrophy) was much more easily gained
than lost (Lee and Shine 1998 ), whereas in teleost fi shes
there is no evidence of such transitions (Goodwin et al. 2002 ;
Mank et al. 2005 ). Although hypotheses about reversals are
actively discussed, most authors tend to consider the acquisition of this novelty as a dominant trend in comparison to its
loss (Wourms and Lombardi 1992 ; Shine and Lee 1999 ;
Blackburn 1999c ; Reznick et al. 2007b ; Pollux et al. 2009 ;
see also Blackburn 2005a , b ). On the other hand, the maternal input can be highly labile. For instance, Dulvy and
Reynolds’s ( 1997 ) phylogenetic analysis showed 6-8 reversals from matrotrophic to lecithotrophic viviparity in elasmobranchs (see also Reynolds et al. 2002 ). Thus, estimates
of the number of independent origins of matrotrophy should
be combined with phylogenetic character mapping.
3.3.4 Origin of Viviparity in the Family
Epistomiidae
An exceptional case of independent evolution of EEN is represented by the viviparous family Epistomiidae, which possesses intraovarian matrotrophic incubation (reproductive
pattern V). In this group the origin of matrotrophy might
have been associated with the transition from incubation of
embryos in the brood chamber (for instance, in the internal
brood sac, as in Beaniidae) to embryonic development
directly in the ovary.
According to Dyrynda and King ( 1982 ), the “epistomiid” character state – intracoelomic incubation (larval
viviparity) and absence of polypide recycling – is the initial
variant, the basis for the subsequent evolution of extracoelomic brooding accompanied by degeneration-regeneration
of polypides. Their line of argument is clear – non-brooding ancestral forms with one polypide generation in the
zooid gave rise to the species with intracoelomic incubation
and no recycling and then extracoelomic brooding appeared.
3.3 Evolution of Matrotrophic Incubation in Cheilostomata