ArticlePDF Available

Notes on Magnoliaceae

Authors:
  • Lab. for experimental plantsystematics, National herbarium of the Netherlands (Rijksherbarium)

Abstract

After the general chapters, mainly based on literature and dealing with wood anatomy, characters of the leaf epidermis and foliar sclereids, embryology and chromosome numbers, phytochemistry, and characters of the fruit, the generic delimitation of subfamily Magnolioideae is discussed. Paramichelia Hu and TsoongiodendronChun are reduced to Michelia Linné; Talauma Juss., Aromadendron Blume, Alcimandra Dandy, Dugandiodendron Lozano and Manglietiastrum Law are reduced to Magnolia Linné. Parakmeria omeiensis Cheng & Hu is renamed Magnolia omeiensis, Parakmeria yunnanensis Cheng & Hu is renamed Magnoliayunnanensis, Manglietiastrum sinicum Law is renamed Magnolia sinicum, and Alcimandra cathcartii Dandy is renamed Magnolia cathcartii Manglietia singalanensis Agostini is reduced to M. glauca var. sumatrana. Elmerrillia mollis Dandy and E. papuana Dandy are reduced to Elmerrillia tsiampacca (Linné) Dandy, the former is named subsp. mollis (Dandy) Noot. and of the latter var. glaberrimais renamed E. tsiampacca var. glaberrima (Dandy) Noot. Michelia arfakiana Agostini is reduced to Elmerrillia tsiampacca var. tsiampacca and Michelia sumatrae Dandy is renamed M. salicifolia Agostini. Newly described is Michelia koordersiana Noot. and keys are given to the genera, subgenera and
BLUMEA
31
(1985)
65-121
Notes
on
Magnoliaceae
with
a
revision
of
Pachylarnax
and
Elmerrillia
and
the
Malesian
species
of
Manglietia
and
Michelia
H.P.
Nooteboom
Rijksheibarium,
Leiden,
The
Netherlands
Contents
Summary
66
Introduction
66
1.
Characters
and
subdivision
of
the
family
67
Wood
anatomy
67
Leaf
epidermis
and
foliar
sclereids
69
Embryology
and
chromosome
numbers
71
Phytochemistry
72
Palynology
72
Morphology
74
Subdivision
of
Magnoliaceae
78
2.
Subfamily
Magnolioideae
78
Key
to
the
tribes
78
Generic
delimitation
in
Magnolieae
78
References
to
chapters
1&2
81
3.
Special
part
82
A.
Tribus
Magnolieae
82
Key
to
the
genera 82
1.
Magnolia
83
Key
to
the
subgenera
and
sections
84
a.
Subgenus
Magnolia
85
b.
Subgenus
Yulania
89
c.
Subgenus
Talauma
89
2.
Manglietia
91
Key
to
the
species
in
Malesia
92
3.
Pachylarnax
97
4.
Kmeria
,
98
Excluded
genera 99
Collections
of
Magnolieae
examined
99
B.
Tribus
Michelieae,
the
Malesian
species
99
Key
to
the
genera
100
5.
Elmerrillia
100
Key
to
the
species,
based
on
flower
and
fruit
characters
100
Key
to
the
species,
based
on
fruit
characters
101
6.
Michelia
108
Key
to
the
species
in
Malesia
109
66
BLUMEA
-
VOL.
31,
No.
1,1985
Summary
After
the
general
chapters,
mainly
based
on
literature
and
dealing
with
wood
anatomy,
char-
acters
of
the
leaf
epidermis
and
foliar
sclereids,
embryology
and
chromosome
numbers,
phyto-
chemistry,
and
characters
of
the
fruit,
the
generic
delimitation
of
subfamily
Magnolioideae
is
discussed.
Paramichelia
Hu
and
Tsoongiodendron
Chun
are
reduced
to
Michelia
Linné;
Talauma
Juss.,
Aromadendron
Blume,
Alcimandra
Dandy,
Dugandiodendron
Lozano
and
Manglietias-
trum
Law
are
reduced
to
Magnolia
Linné.
Parakmeria
omeiensis
Cheng
&
Hu
is
renamed
Magno-
lia
omeiensis,
Parakmeria
yunnanensis
Cheng
&
Hu
is
renamed
Magnolia
yunnanensis,
Manglieti-
astrum
sinicum
Law
is
renamed
Magnolia
sinicum,
and
Alcimandra
cathcartii
Dandy
is
renamed
Magnolia
cathcartii
Manglietia
singalanensis
Agostini
is
reduced
to
M.
glauca
var.
sumatrana.
Elmerrillia
mollis
Dandy
and
E.
papuana
Dandy
are
reduced
to
Elmerrillia
tsiampacca
(Linné)
Dandy,
the
former
is
named
subsp.
mollis
(Dandy)
Noot.
and
of
the
latter
var.
glaberrima
is
renamed
E.
tsiampacca
var.
glaberrima
(Dandy)
Noot.
Michelia
arfakiana
Agostini
is
reduced
to
Elmerrillia
tsiampacca
var.
tsiampacca
and
Michelia
sumatrae
Dandy
is
renamed
M.
salicifolia
Agostini.
Newly
described
is
Michelia koordersiana
Noot.
and
keys
are
given
to
the
genera,
sub-
genera and
sections.
Introduction
The
late
Dr.
J.
E.
Dandy,
Keeper
of
Botany
of
the
British
Museum
(Natural
His-
tory)
from
1956
to
1966,
started
his
work
on
Magnoliaceae
in
1925
working
for
Dr.
John
Hutchinson
at
Kew.
He
continued
working
at
Magnoliaceae
until
his
death
in
November
1976.
He
can
be
considered
the
greatest
authority
ever
on
the
family.
He
constantly
maintained
a
very
high
standard
of
scholarship
and
became
almost
mor-
bidly
reluctant
to
publish
his
continually
revised
works
which
were
never,
for
him,
quite
perfect
enough.
Consequently
the
monograph
on
the
Magnoliaceae,
upon
which
he
worked
for
nearly
50
years,
remained
unpublished.
Because
many
of
the
collections
were
together
at
the
time
he
died,
it
was
decided
that
I,
looking
for
a
new
family
to
revise
for
the
Flora
Malesiana,
should
start
with
the
Magnoliaceae.
During
that
work
I
had
to
look
outside
the
Malesian
area,
studying
the
genus
concept
within
the
family.
In
doing
that,
and
studying
the
species
within
the
area
as
well,
I
came
to
the
conclusion
that
the
tricks
of
nature
which
are
apparent
in
the
evolution
of
Magnoliaceae
have
prevented
Dr.
Dandy,
as
a
perfectionist,
from
the
final
circum-
scription
of
many
taxa.
Often
there
are
no
clear-cut
limits
between
the
taxa,
and
a
more
or
less
arbitrary
choice
has
to
be
made.
In
order
to
arrive
at
'sensible'
but
not
undisputable
taxa,
I
decided
to
reduce
several
genera,
even
one
described
by
Dandy
himself,
and
also
some
species.
After
finishing
the
manuscript
for
this
paper,
the
article
of
Law
Yuh-wu
on
the
taxonomy
of
the
family
Magnoliaceae
in
Acta
Phytotaxonomica
Sinica
22
(1984)
89—
108
was
published.
Most
of
it
is
written
in
Chinese
and
thus
the
argumentation
is
for
most
botanists
not
accessible.
His
division
of
the
family
in
subfamilies
is
used
here.
Law's
division
(I.e.
p.
105-106)
is
rather
confusing
as
he,
unintentionally,
puts
the
tribes
Magnolieae
and
Michelieae
under
the
subfamily
Liriodendroideae
and
all
the
subtribes
under
the
tribe
Michelieae.
67
H.P.
Nooteboom:
Notes
on
Magnoliaceae
1.
CHARACTERS
AND
SUBDIVISION
OF
THE
FAMILY
Wood
anatomy
(Largely
from
H.
Gottwald,
manuscript
of
a
lecture
given
for
the
International
Association
of
Wood
Anatomists
in
Amsterdam,
which
he
kindly
put
at
my
dis-
posal.)
The
structural
pattern
in
the
secondary
xylem,
typical
for
all
species,
shows
a
well
developed
fibrous
ground
tissue
of
thin
to
medium
thick-walled
fibres,
with
diameters
from
15
to
25
pm
depending
on
the
species.
The
vessels
show
a
slight
tendency
of
radial
grouping.
The
diameters
based
on
species
averages,
range
from
50
to
180
pm.
The
vessels
are
mostly
evenly
distributed
except
for
the
temperate
species
which
tend
to
possess
porous
zones.
The
horizontal
parenchyma
consists
of
unstoried
rays
of
mostly
3
cells
in
width
and
0.5—1
mm
high.
The
vertical
parenchyma
consists
of
continuous
parenchyma
rings,
3
to
6
cells
wide.
The
layers
between
these
rings
were
recognized
by
Chowdhury
(1964)
as
per-
fect
annual
rings,
also
for
the
tropical
Michelia
champaca.
Only
in
the
genus
Talauma
(
Magnolia
subg.
Talauma)
this
main
feature
shows
some
alteration
in
that
the
bands
tend
to
fork.
A
parenchymatic
sheath
surrounding
the
vessels
is
always
incomplete;
in
most
species
parenchyma
cells
are
even
difficult
to
detect.
With
reference
to
the
contents
of
parenchymatic
tissues
none
of
the
species
contain
any
kind
of
crystals,
nor
were
any
phenolic
substances
found
in
the
vessels.
Altogether
this
general
topo-
graphic
pattern
represents
a
structural
principle
which
can
also
be
found
in
very
dif-
ferent
families
as
e.g.
in
Swietenia
of
Meliaceae,
Beilschmiedia
of
Lauraceae,
in
Ver-
benaceae,
and
in
others,
widely
dispersed
in
lesser
or
higher
developed
groups
of
the
Angiosperms.
While
the
architecture
of
the
xylem
of
Magnoliaceae
has
no
individual
pattern,
there
are
some
striking
anatomical
details
occurring
only
in
groups
of
individual
taxa.
Firstly
the
extremely
primitive
vessel
pitting
varying
from
purely
scalariform
to
op-
posite.
Though
this
feature
represents
the
least
developed
type
of
pitting,
the
vessel
perforations
contain
mostly
multiple
perforations
of
only
1
to
15
bars.
In
the
ad-
vanced
taxa
Liriodendron
and
the
section
Yulania
of
Magnolia
also
simple
perforated
vessels
occur,
but
always
in
connection
with
opposite
pits.
Also
a
rare
feature
in
connection
with
the
vessels
is
the
occurrence
of
amorph-
ous
silica
as
a
coating
on
the
cell
walls
including
the
tyloses,
or
as
solid
occlusions
completely
closing
the
lumina.
According
to
Professor
Gottwald's
investigations,
the
complete
solid
plugs
of
silicium
dioxide
occur
in
all
species
of
section
Blumiana
of
Magnolia
subg.
Talauma
and
in
the
monotypic
section
Lirianthe
of
Magnolia
subg.
Magnolia.
The
estimated
volume
percentage
of
silica
is
up
to
8%,
based
on
air-dried
volume;
sometimes
almost
every
vessel
is
filled
with
silicium
dioxide.
This
means
that
these
taxa
belong
to
the
woody
species
with
the
highest
silica
content
known
in
the
vascular
plants.
Spiral
thickenings
on
the
vessel
walls
are
a
further
special
feature
of
limited
occurrence
in
Magnoliaceae.
They
are
only
weakly
pronounced
and
hard
to
detect.
68
BLUMEA
-
VOL.
31,
No.
1,
1985
The
problem
of
an
irregular
distribution,
frequency,
and
morphology
is
also
typi-
cal
for
the
rare
idioblastic
parenchyma
cells,
to
be
found
mainly
between
the
margin-
al
cells
of
the
rays.
They
are
thin-walled,
have
mostly
amorphous
brownish
contents,
and
are
often
referred
to
as
'oil
cells',
although
their
chemical
nature
is
still
un-
known.
In
some
genera
these
enlarged
cells
always
occur,
in
others
they
may
occur
or
not,
so
their
taxonomical
value
is
limited.
Yet
another
specific
feature
is
the
occurrence
of
silica
particles
in
the
ray
cells
of
a
few
species
in
the
sections
Maingola
(of
Magnolia
subg.
Magnolia)
and
Blumiana
(of
Magnolia
subg.
Talauma),
pointing
to
their
close
relationship.
The
only
crystalline
inclusion
is
found
in
fine
splits
of
wood
of
Magnolia
and
its
subg.
Talauma;
they
could
be
identified
as
consisting
of
calcium
carbonate
and
are
always
of
traumatic
origin.
The
last
special
item
concerns
the
occurrence
of
tyloses
in
the
fibres,
besides
those
in
the
vessels.
This
rare
feature
was
observed
in
24
species,
belonging
to
7
genera.
Concluding,
it
can
be
stated
that
the
family
Magnoliaceae
as
a
whole
can
be
iden-
tified
easily
on
its
xylem by
its
pronounced
homogeneity
and
its
special
features.
It
can,
therefore,
be
distinguished
from
all
other
families,
particularly
from
those
which
are
also
part
of
the
order
Magnoliales.
This
structural
separation
within
the
flowering
plants
is
very
remarkable
if
the
Magnoliaceae
are
considered
as
the
root
of
all
other
families.
The
homogeneity
of
characters
enhances
a
mixture
of
overlapping
structural
de-
tails
which
is
detrimental
to
the
internal
classification
of
the
family.
The
largest
genus,
Magnolia
(without
the
taxa
reduced
to
it
in
this
paper),
includes
almost
all
structural
items
which
are
otherwise
spread
over
the
remaining
genera.
There
are
only
few
taxa
that
can
be
distinguished
from
all
others
as
e.g.
the
genus
Liriodendron
which
has
neither
spirals
in
the
vessels
nor
tyloses
in
the
fibres.
Similar-
ly
section
Yulania
of
Magnolia
subg.
Yulania
is
characterized
by
exclusively
simple
perforations,
and
section
Blumiana
of
Magnolia
subg.
Talauma
by
its
silica-bearing
vessels
and
the
pronounced
heterogeneity
of
the
rays.
Canright
(1955)
comes
to
the
following
conclusions.
Aromadendron
(Magnolia
subg.
Talauma
sect.
Aromadendron)
differs
from
the
rest
of
subg.
Talauma
in
that
the
pores
are
almost
twice
the
size
of
the
investigated
Asiatic
species.
They
are
also
the
largest
in
the
whole
family.
Differences
are
also
apparent
in
the
wood
parenchy-
ma
distribution
and
fibre
tracheid
characteristics.
The
wood
of
Elmerrillia
falls
with-
in
the
range
of
the
woods
of
Michelia
whereas
the
wood
of
Kmeria
is
indistinguish-
able
from
that
of
many
tropical
Magnolia
species.
There
is
little
difference
in
the
wood
anatomy
of
Manglietia
and
many
temperate
Magnolia
species.
Although
the
primitive
nor
the
advanced
wood
anatomical
charac-
ters
are
confined
to
one
genus,
there
are
many
indications
that
the
woods
of
the
temperate
species
of
Magnolia
and
Liriodendron
are
the
most
specialized
while
most
tropical
species
of
Magnolia
subg.
Paramanglietia
Hu
&
Cheng
(1951:
255)
are
within
the
range
of
Manglietia
according
to
Chang Chetseng
(1984:
483).
Talauma
exhibits
the
largest
assemblage
of
primitive
characters.
The
tropical
Magnolia
paenetalauma
from
Hainan
exhibits
a
number
of
primitive
features
such
as
small
solitary
pores,
H.P.
Nooteboom:
Notes
on
Magnoliaceae
69
long
vessel
elements
with
many-barred
perforation
plates,
and
extremely
tall
hetero-
geneous
rays.
Also
in
the
external
morphology
the
species
of
section
Gwillimia
(of
Magnolia
subg.
Magnolia),
to
which
M.
paenetalauma
belongs,
resemble
the
species
of
subg.
Talauma
so
closely
that
they
cannot
be
distinguished
in
absence
of
fruits.
Leaf
epidermis
and
foliar
sclereids
(Mainly
after
Baranova,
1972,
and
Tucker,
1977.)
The
ordinary
epidermal
cells
of
leaves
of
Magnoliaceae
are
commonly
irregular
in
form,
or
less
often
polygonal,
with
sinuous
to
merely
curved
or
nearly
straight,
often
thickened,
walls.
Many
mem-
bers
of
the
family
have
a
subepidermal
layer
beneath
the
upper
epidermis,
some
have
such
a
layer
associated
with
both
upper
and
lower
epidermis,
and
few
have
it
only
with
the
lower
epidermis.
Many
species
lack
specialized
subepidermal
cells
entirely.
In
many
species
the
outer
walls
of
the
epidermal
cells
have
pores,
fewer
or
more
numerous
according
to
the
species.
Similar
pores
are
seen
in
Cycas
and
some
other
genera
of
the
Cycadales.
Hairs
may be
present
on
both
upper
and
lower
epidermis,
or
only
on
the
lower.
In
some
species
the
leaves
lack
hairs
altogether.
They
are
uniseriate,
consisting
of
one
to
several
or
many
cells.
There
are
two
types
of
hair
base.
In
the
first
type
the
hair
rests
on
a
normal
or
modified
epidermal
cell
or
group
of
cells.
The
position
of
the
detached
hair
may
readily
be
observed
on
the
cuticular
membrane
because
of
the
persistent
hair
base.
This
type
of
hair
base
may
have
only
2—3
cells
or
many
highly
modified
cells.
In
the
second
type
the
hair
base
replaces
a
normal
epidermal
cell,
so
that
the
loss
of
a
hair
leaves
a
pore
in
the
cuticular
membrane.
This
type
of
hair
base
is
only
seen
in
Manglietia.
Baranova
(1972)
regards
this
hair
base
as
primitive.
The
hair
bases
of
the
other
genera
are
more
or
less
advanced.
An
early
stage
of
specializa-
tion
is
seen
in
several
genera:
Magnolia,
incl.
Aromadendron
and
Talauma,
and
Michelia
incl.
Paramichelia.
In
this
type
the
hair
base
consists
of
2—4
slightly
modi-
fied
epidermal
cells.
Intermediates
between
this
type
of
hair
and
the
Manglietia
type
occur
in
Manglietia
and
in
Magnolia
maingayi.
The
most
complex
hair
base
types
occur
in
Elmerrillia,
Tsoongiodendron,
and
scattered
species
of
Magnolia.
Here
the
hair
base
consists
of
many
highly
modified
epidermal
cells.
Baranova
distinguishes,
in
order
of
advancement,
the
following
hair
base
types:
1
Manglietia
type,
2.
Magnolia
maingayi
type,
3.
Talauma
type,
4.
Elmerrillia
type.
Manglietia
has
type
1
and
2.
Magnolia
s.s.
has
type
2,
3,
and
4;
the
other
pubescent
genera
possess
only
one
type
each.
The
different
types
of
hair
base
in
Magnolia
are
concentrated
in
sect.
Maingola,
whereas
all
the
other
species
possess
only
the
Talauma
type.
The
stomata
are
paracytic
in
Magnolioideae
(but
in
rare
cases
Nong
Van
Tiep,
1980:
519,
found
also
anomocytic
stomata
in
Manglietia),
paracytic
as
well
as
anomocytic
in
Liriodendroideae.
They
are
confined
to
the
lower
surface.
Significant
thickenings
on
the
walls
of
the
epidermal
and
subsidiary
cells,
as
well
as
strongly
developed
cuticular
thickenings
on
the
outer
walls
of
the
guard
cell,
occur
mainly
in
the
tropical
sections
of
Magnolia
s.s.
as
well
as
in
many
Asiatic
species
of
Talauma.
They
are
lacking
in
all
temperate
species
of
Magnolia
and
in
Elmerrillia,
Pachylarnax
and
Liriodendron.
BLUMEA
-
VOL.
31,
No.
1,
1985
70
Foliar
sclereids
are
present
in
certain
taxa.
They
vary
widely
in
form,
size,
wall
thickness,
pitting,
and
degree
of
ramification.
They
may be
found
in
four
tissues
or
cell
assemblages:
idioblasts,
mesophyll,
dermal
layers,
and
in
the
vein
sheath
system
including
veinlet
endings.
Idioblastic
sclereids
occur
commonly
in
the
petiole
and
midrib.
They
are
concen-
trated
around
the
midrib
bundles,
but
always
intermittent
rather
than
continuously
distributed.
This
distributional
pattern
along
the
midrib
is
found
in
the
majority
of
taxa
sampled.
Sclerification
of
the
midrib,
however,
appears
less
well
developed
in
short-lived
deciduous
magnoliaceous
leaves.
This
pattern
of
sclereids
along
the
mid-
rib
is
so
common
among
evergreen
taxa,
that
with
few
exceptions
it
is
of
little
use
diagnostically.
Less
common
are
the
following
patterns
of
distribution
of
sclereids:
1)
Throughout
the
leaf
('diffuse'),
these
sclereids
occur
in
the
mesophyll
at
the
same
paradermal
level
as
the
veinlet
terminations.
2)
Bordering
the
larger
veins
as
well
as
the
midrib.
3)
Scattered
along
the
margin
of
the
leaf,
as
well
as
along
the
midrib.
Sclerified
spongy
mesophyll
is
found
in
all
examined
species
of
Manglietia
(this
is
not
mentioned
by
Nong
Van
Tiep,
1980),
in
eight
species
of
Magnolia,
in
five
species
of
Michelia,
and
in
four
species
of
Talauma.
Adaxial
or
abaxial
epidermal
and/or
hypodermal
layers
may be
sclerified.
Some-
times
only
scattered
cells
of
a
layer,
such
as
those
over
a
vein,
are
sclerified.
Usually
the
wall
thickenings
are
not
massive
and
the
cells
probably
remain
alive.
Sclerified
epidermis
is
found
in
species
of
Alcimandra,
Elmerrillia,
Magnolia
s.s.,
Michelia,
Talauma,
and
in
all
the
Manglietia
species
examined.
Sclerification
takes
the
form
of
a
sclerified
hypodermis
in
13
taxa
examined,
including
10
species
of
Manglietia.
Except
for
the
types
discussed
above,
most
sclereids
in
magnoliaceous
leaves
are
associated
with
the
venation
system.
For
instance
with
the
vein
sheath
in
taxa
with
prominent
lateral
veins,
with
the
bundle
sheath
extensions
along
leaf
margins,
and
sclerified
transitional
elements
at
the
veinlet
terminations.
Differentiation
of
the
sclerified
vein
sheath
differs
among
species.
Sclerified
leaf
margins
occur
in
many
taxa.
They
may
be
discontinuous
or
scat-
tered,
or
they
may
form
a
continuous
massive
marginal
ridge.
This
ridge
may be
entirely
fibrous
or
contain
a
vascular
bundle,
or
it
may
contain
essentially
continu-
ous
sclerified
epidermal
and
subepidermal
cells.
At
the
veinlet
terminations
of
Magnoliaceae
a
variety
of
tracheids,
sclereids,
and
intermediate
or
transitional
cells
are
found.
Taxonomically
the
mentioned
characters
are
of
different
value.
Each
genus
shows
a
particular
range
of
features,
some
of
which
can
be used
diagnostically
to
separate
genera.
The
coriaceous
texture
of
tropical
leaves
is
achieved
in
various
ways
arising
from
the
variable
form
and
diverse
distribution
of
the
sclereids.
Talauma
species
characteristically
have
the
entire
vein
system
of
the
leaf
encased
in
sclerenchymatous
sheaths,
and
the
veinlets
terminate
in
sclerified
elements.
Manglietia
species
in
contrast
have
parenchymatous
vein
sheaths
near
the
termina-
tions
and
lack
sclerified
terminal
cells,
but
most
of
the
mesophyll
and
epidermal
cells
are
sclerified.
Manglietia
leaves
are
characterized
by
sclerified
epidermal
and
hypoder-
mal
layers,
sclerified
spongy
parenchyma
in
the
mesophyll,
unlignified
lobate
vein-
H.P.
Nooteboom:
Notes
on
Magnoliaceae
71
sheath
cells,
and
an
absence
of
either
sclerified
veinlet
terminal
cells
or
a
sclerified
leaf
margin.
No
other
genus
combines
all
those
features,
although
others
do
show
one
or
more.
Sterile
specimens
of
Manglietia
can
easily
be
recognized
on
the
basis
of
cleared
leaf
segments.
Magnolia
s.s.,
Michelia,
and
Talauma
exhibit
in
many
of
their
members
sclerified
veinlet
terminal
cells,
thick
sclerified
leaf
margins,
and
stellate
sclereids
along
the
midrib.
Rarely
those features
may be
diagnostic
for
species,
but
in
most
cases
they
are
too
common
to
be
useful.
The
foliar
characteristics
of
Talauma
include
veinlets
terminating
in
thick-walled
stellate
sclereids
and
thinner
walled
tracheary
elements.
These
cell
types
are
typically
absent
from
temperate
Magnolia
sections
and
therefore
obviously
an
adaptation
to
climatic
conditions.
In
all
the
Asiatic
species
of
Talauma
examined
there
is
a
thick
sclerified
margin,
usually
including
a
vein.
In
the
American
species
this
is
absent
but
in
few
of
them
a
weak
approach
to
this
condition
is
seen.
The
massice
marginal
ridge
helps
to
identify
certain
species
of
Talauma
and
Mag-
nolia,
Manglietia (3
species),
Michelia
(3
species)
and
Aromadendron
elegans.
In
Alcimandra,
Liriodendron,
Paramichelia,
and
most
Elmerrillia
species
foliar
sclereids
are
rare.
This may
indicate
that
these
taxa
have
not
developed
vegetative
modifications
different
from
those
of
the
related
large
genera.
It
seems
to
me
that
on
leaf
anatomical
characters
Manglietia
is
a
specialized
off-
shoot
of
Magnolia.
Regarding
the
other
genera
there
are
no
leaf
anatomical
evidences
pointing
to
them
being
separate
genera.
Embryology
and
chromosome
numbers
(Hayashi,
1964,1966;
Bhandari,
1971.)
The
anther
wall
at
the
microspore
mother
cell
stage
comprises
epidermis,
endothecium,
3
or
4
middle
layers,
and
a
bilateral
glandular
tapetum.
By
the
time
the
cytokinesis
is
completed
in
the
microspore
mother
cells,
a
large
number
of
Ubisch
granules
line
the
inner
walls
of
the
tapetum.
In
a
fully
mature
anther
the
papillate
epidermis
and
endothecium
along
with
2
or
3
middle
layers
persist.
After
meiosis
II
in
the
microspore
mother
cell
the
cytokinesis
takes
place
by
furrowing,
resulting
in
tetrahedral
or
isobilateral
tetrads.
The
mature
pollen
is
shed
at
the
2-celled
stage.
The
generative
cell
is
surrounded
by
a
thin
sheath
of
finely
granu-
lar
cytoplasm
and
a
delicate
membrane.
The
ovules
are
anatropous,
bitegmic,
and
crassinucellate;
the
outer
integument
is
vascularized.
The
hypodermal
archesporium
is
multicellular
and
of
hypodermal
origin
although
ultimately
only
one
cell
functions.
The
primary
parietal
cell
divides
repeat-
edly
to
form
the
parietal
tissue
so
that
the
megaspore
mother
cell
is
buried
deep
in
the
nucellus.
At
the
end
of
meiosis
II
linear
or
T-shaped
megaspore
tetrads
are
form-
ed.
The
chalazal
megaspore
functions,
giving
rise
to
the
Polygonum
type
of
embryo
sac.
The
synergids
and
antipodal
cells
are
ephemeral.
The
endosperm
is
cellular
from
the
beginning,
and
embryogeny
conforms
to
the
Myosurus
variation
of
the
Onagrad
type
or
is
irregular.
The
seed
coat
is
organized
chiefly
from
the
outer
integument
while
the
inner
in-
tegument
is
represented
by
a
layer
of
crushed
cells.
In
a
ripe
seed
the
testa
consists
of
an
outer
fleshy
region
comprising
the
epidermis
of
the
outer
integument,
2
or
3
72
BLUMEA
-
VOL.
31,
No.
1,
1985
layers
of
tangentially
elongated
cells,
a
10—12-layered
fleshy
zone,
2
or
3
layers
of
tangentially
compressed
cells,
and
an
inner
stony
region
of
3
or
4
layers
of
lignified
cells.
The
basic
chromosome
number
is
x
=
19.
Polyploidy
has
been
demonstrated
in
Magnolia.
Treseder
(1978:
208)
reports
for
cultivated
Magnolias
besides
diploid,
also
triploid,
tetraploid,
pentaploid,
hexaploid,
heptaploid,
and
octoploid
numbers,
while
aneuploid
numbers
also
are
recorded.
Phytochemistry
(After
Hegnauer,
1969.)
Like
other
families
of
woody
Policarpicae
the
Magnolia-
ceae
show
a
combination
of
chemical
features
highly
characteristic
for
this
alliance.
They
are
characterized
by
the
following
metabolic
peculiarities:
1.
Accumulation
of
silica
in
many
species.
2.
Synthesis
of
isochinolin
alkaloids.
3.
Production
of
essential
oils
and
depositing
the
latter
in
idioblasts.
4.
The
absence
of
triterpene
accumulation.
5.
In
the
phenolic
constituents
vicinal
trihydroxylation
appears
to
be absent
in
anthocyanins,
leucoanthocyanins,
flavonols,
and
C
6
-C
t
acids.
Syringyl
substi-
tution
occurs
in
cinnamic
acids
(sinapic
acid),
cinnamic
alcohols
(syringin)
and
lignans
(lirioresinol).
6.
Cyanogenesis
is
at
the
moment
only
known
in
the
genus
Liriodendron.
The
cyanogenetic
compounds
appear
to
belong
to
the
same
kinds
as
those
found
in
other
Polycarpicae.
7.
Oil
accumulation
in
seeds.
8.
Synthesis
of
cyclolitols
(pinitol,
liriodendritol,
and
quercitol).
Much
research
is
still
needed,
especially
for
the
cardiac
active
compounds
found
in
Magnolia-
ceae,
of
which
the
chemistry
is
still
entirely
unknown.
Palynology
(After
J.
Praglowski,
1974.)
The
pollen
grains
of
Magnoliaceae
are
l-(ana-)colpate,
bilateral,
heteropolar.
Aperture
simple,
with
markedly
thin,
frequently
slightly
undu-
lated
margins.
Colpus usually
longer,
occasionally
as
long
as,
or
rarely
shorter
than
the
longest
axis.
Colpus
invagination
usually
absent
or
insignificant.
Exine
structure
in
tectate
grains
(about
95%)
consists
of
a
continuous
tectum
perforatum
supported
by
bacula
which
have
no
geometrical
connection
with
the
tectine
of
the
exine
sculp-
ture.
In
rarely
occurring
semitectate
grains
the
exine
structure
is
microreticulate,
without
sculpturing.
Sculpturing
rugulose,
or
less
frequently
pollen
grains
have
a
comparatively
smooth
tectum.
Sexine
at
the
proximal
face
thicker
than
nexine.
Bacula
minute,
usually
indistinct.
Nexine
thicker
than
sexine,
usually
half
of
the
sexine
thickness
or
less.
Praglowski
distinguished
between
the
following
pollen
types
(E
t
=
longest
axis):
73
H.P.
Nooteboom:
Notes
on
Magnoliaceae
1.
Pollen
grains
tectate,
exine
structure
consisting
of
tectum
perforatum
and
bacula
arranged
without
geometrical
connection
to
the
tectum.
2.
Tectum
surface
smooth.
3.
Pollen
grains
large,
Ei
45
pm
or
larger
Type
1
3.
Pollen
grains
small,
Ej
<
45
pm
Type
2
2.
Tectum
surface
with
rugulose
sculpturing.
4.
Sculpturing
slightly
rugulose.
5.
Pollen
grains
large,
Ej
45
pun
or
larger
Type
3
5.
Pollen
grains
small,
Ei
<
45
pm
Type
4
4.
Sculpturing
coarsely
rugulose
Type
5
1.
Pollen
grains
semitectate,
exine
structure
microreticulate.
6.
Pollen
grains
large,
E,
45
pun
or
larger
Type
6
6.
Pollen
grains
small,
E,
<
45
pun
Type
7
In
the
genera
of
Magnoliaceae
the
following
pollen
types
occur:
Elmerrillia:
type
7
(E.
tsiampacca
subsp.
mollis)
or
type
2
(the
other
investigated
taxa,
E.
tsiampacca
subsp.
tsiampacca
and
E.
ovalis).
Michelia
(incl.
Paramichelia
and
Tsoongiodendron
):
type
2,
but
in
M.
floribunda
(<Garrett
649,
BM)
as
well
type
2
as
type
4
are
found.
Pachylarnax:
type
7.
Alcimandra:
type
2.
Kmeria:
no
pollen
was
investigated.
Magnolia:
the
types
1,2,3,4,
and
6.
Talauma:
type
1
(and
3:
T.
rabianae).
Manglietia:
type
1
and
3.
Aromadendron:
type
2
and
4.
Liriodendron:
type
3,5,
and
6.
Elmerrillia
pollen
grains
show
partly
similarity
with
pollen
grains
of
Michelia,
but
the
microreticulate
structure
of
the
grains
of
E.
tsiampacca
subsp.
mollis
is
rather
unique
showing
no
resemblance
to
Michelia
pollen.
The
reticulum
is
considerably
coarser
than
that
of
Pachylarnax
pollen
grains.
Michelia,
Paramichelia,
and
Tsoongiodendron
possess
pollen
grains
that
are
very
similar.
Pachylarnax
possesses
pollen
with
microreticulate
exine
structure
consisting
of
thin
muri
encompassing
very
minute
lumina.
This
rather
peculiar
exine
structure
is
more
delicate
than
that
in
other
Magnoliaceae.
Similarly,
the
rather
symmetric
ellipsoidal
shape
of
the
pollen
makes
it
quite
difficult
to
include
them
among
typical
magnoliaceous
pollen.
Alcimandra
pollen
possess
an
exine
without
sculpturing
which
show
resemblance
with
some
Talauma
pollen.
Magnolia
pollen
grains
show
a
rather
large
morphological
variation,
only
types
5
and
7
do
not
occur
in
this
genus.
As
type
5
is
only
met
with
in
the
genus
Liriodendron,
nearly
all
the
pollen
types
of
subfamily
Magnolioideae
are
74
BLUMEA
-
VOL.
31,
No.
1,1985
found
in
Magnolia.
However,
pollen
grains
of
Elmerrillia
and
Michelia
show
slight
dif-
ferences
to
that
of
Magnolia.
Manglietia
pollen
grains
show
a
high
morphological
similarity
to
those
of
Magnolia.
Talauma
pollen
grains
show
features
similar
to
those
encountered
in
Magnolia
subg.
Magnolia.
Aromadendron
pollen
grains
show
similarity
to
Alcimandra
pollen
grains,
but
also
with
those
of
Magnolia
sect.
Maingola.
On
palynological
evidence
the
joining
of
Michelia,
Tsoongiodendron,
and
Para-
michelia
is
strongly
supported.
The
same
holds
for
Manglietia,
Magnolia,
and
Talau-
ma,
while
the
joining
of
Aromadendron
with
Magnolia
is
not
contradicted.
Morphology
All
Magnoliaceae
are
trees
or
shrubs
with
caducous
or
(always
in
Malesia)
persis-
tent
leaves.
The
leaves
possess
a
pair
of
stipules
which
are
connate
in
bud,
forming
a
cap
enclosing
the
young
leaves
in
such
a
way
that
the
stipules
of
the
last
developed
leaf
enclose
the
terminal
bud
and
then
rupture
longitudinally
(the
two
original
sti-
pules
become
separate)
and
fall,
leaving
a
circular
scar
around
the
twig.
They
can
be
free
from
the
petiole
or
(partly)
adnate
to
it,
leaving
a
conspicuous
scar
when
falling.
Growth
of
the
internodes
often
is
faster
in
the
beginning
of
the
season,
thus
forming
longer
internodes
than
later.
In
tropical
species,
however,
there
may
be
no
conspicu-
ous
difference
in
length
of
internodes
throughout
the
twig.
Flowers
The
flowers
are
borne
on
a
pedicle
which
mostly
is
the
ultimate
internode
of
a
brachyblast
(always
in
Elmerrillia
and
Michelia,
where
the
brachyblast
is
axillar;
in
the
other
genera
the
brachyblast
or
the
pedicle
with
the
flower
is
termi-
nal).
The
flower
bud
is
often
protected
by
the
stipules
of
the
uppermost
leaf
especial-
ly
in
the
genera
with
terminal
flowers.
In
temperate
species
these
stipules
often
are
leathery
and/or
hairy.
The
brachyblast,
if
present,
bears
one
to
several
spathaceous
bracts
which
also
surround
and
protect
the
flower
bud.
Essentially
these
bracts
con-
sist
of
a
petiole
with
its
stipules
(here
always
adnate
to
it),
together
forming
a
cap
like
the
one
surrounding
the
vegetative
buds.
Usually
the
leafblade
is
absent,
but
sometimes
a
reduced
blade
can
be
observed.
The
uppermost
spathaceous
bract,
some-
times
called
bracteole,
surrounds
the
base
of
the
pedicle.
This
pedicle
can
be
very
short
or
(nearly)
absent
to
very
long.
Perianth
The
perianth
consists
of
6
to
many
parts
which,
at
least
when
present
in
low
numbers,
are
inserted
in
whorls.
Tucker
(1960)
demonstrated
that
the
6
peri-
anth
parts
of
Michelia
fuscata
are
initiated
successively,
although
from
two
tiers.
There
is
little
difference
in
the
fundamental
plan
of
vascular
systems
in
the
differ-
ent
genera
(Ueda,
1984).
The
outer
whorl
of
the
perianth
often
is
not
recognized
as
a
calyx.
Erbar
&
Leins
(1982)
found
in
Magnolia
stellata,
in
which
the
constituent
flower
appendages
are
many
and
in
indefinite
numbers,
that
the
tepals
arise
in
a
spiral,
rather
fast
after
each
other.
The
first
primordia
possess
a
relatively
broad
plain
of
insertion.
After
developing
of
six
primordia
there
is
a
more
or
less
alternation
of
3
x
3,
on
a
triangular
apex
of
the
flower
axis.
Higher
on
this
apex
becomes
5-angular
because
the
primordia
7
and
8
stay
not
much
higher
than
4, 5,
and
6.
The
primordia
1, 2,
and
3
often
do
not
differentiate,
or
only
one
of
them
differentiates.
The
spiral
75
H.P.
Nooteboom:
Notes
on
Magnoliaceae
in
which
the
tepals
origin
is
maintained
when,
successively,
the
stamens
and
carpels
arise
and
can
be
turning
right
or
left.
In
Magnolia
denudata
the
number
of
tepals
is
fixed.
The
tepals
of
the
outer
whorl
arise
after
each
other,
with
a
clear
time
delay
between
the
first
and
the
other
two
who
arise
rather
fast
after
each
other.
As
in
Magnolia
stellata
the
plain
of
insertion
becomes
gradually
narrower
in
the
later
arising tepals.
The
nine
tepals
are
arranged
in
three
whorls
of
3,
but
are
not
aequidistant
within
their
whorl.
I
observed
the
same
phenomenon
in
Michelia
nilagirica
from
Ceylon.
Erbar
&
Leins
(1982,1983)
explain
the
existence
of
tepals
in
alternating
whorls
of
three
because
in
a
spiral
3
tepals
are
inserted
in
one
revolution
and
because
there
is
a
time
delay
after
the
formation
of
each
whorl.
After
the
arising
of
three
primordia
there
is
a
long
pause.
This
forces
the
next
three
primordia
into
the
gaps
between
the
former.
According
to
Ueda
(1984)
Magnoliaceae
possess
a
true
calyx.
There
are
several
differences
between
the
outer
whorl
and
the
others
like
texture,
shape
and
size,
thickness,
and
width
at
the
base.
Except
for
the
texture
those
differences
can
be
ex-
plained
by
Erbar
&
Leins'
conclusions
too.
Another
difference
is
in
the
vasculariza-
tion.
Ueda
also
argues
that
the
direction
of
the
helix
is
commonly
different
in
the
sepals
and
the
petals,
which
is
in
my
opinion
contradictory
to
the
observations
of
Erbar
and
Leins.
Vascular
system
The
floral
axis
of
the
representatives
of
Magnoliaceae
is
characterized
by
two
vascular
systems
(Skipworth,
1970),
a
stelar
and
a
cortical
system
of
vascular
bundles.
The
brachyblast,
at
its
base,
contains
a
ring
of
normal
collateral
vascular
bundles.
The
cortical
system
arises
below
the
flower,
either
direct-
ly
from
the
stele
or
as
branches
of
leaf
and
stipule
traces.
In
many
cases
it
finally
comprises
a
lattice
in
gynoecial
and
androecial
regions,
while
its
strands
are
frequent-
ly
interconnected
by
apparent
rings
of
vascular
tissue
in
the
region
of
the
perianth.
Generally
the
cortical
system
provides
the
lateral
traces
of
the
flower
appendages,
the
stelar
system
the
median
traces.
This
strongly
suggests
the
essential
similarity
of
bracts,
perianth
members,
stamens,
and
carpels.
According
to
Skipworth
it
is
highly
probable
that
all
floral
parts
in
Magnoliaceae
are
of
leaf
rank.
According
to
Ueda
(1984)
sepals
are
supplied
by
five
to
fifteen
strands
which
are
arranged
in
two
levels
at
the
base
of
the
sepals:
a
midvein
and
two
strong
lateral
veins,
sometimes
an
addi-
tional
two
strong
lateral
veins,
in
the
lower
level,
and
the
other,
weak,
veins
in
the
upper
level.
Petals
are
vascularized
by
eleven
to
seventeen
traces
in
three
levels
at
their
base.
A
midvein
and
two
strong
lateral
veins
in
the
lower
level,
three
to
five
lateral
veins
in
the
middle
level,
and
the
other
five,
weak veins
in
the
upper
level.
Stamens
The
stamens
are
spirally
arranged
and
develop
either
directly
after
the
tepals
and
in
the
same
8
(7—10)
parastichies
like
in
Magnolia
stellata
(Erbar
&
Leins,
1982),
or
after
a
relatively
long
interval
like
in
Magnolia
denudata
(I.e.)
and
Michelia
fuscata
(Tucker,
1960).
In
Magnolia
denudata
the
spiral
of
the
stamens
jumps
in
relation
to
the
basal
spiral.
Several
primordia
originate
too
early
compared
with
the
other
ones.
So
stamen
9
arises
before
stamen
6.
This
jump
is
conveyed
to
a
higher
level
where
the
same
phenomenon
occurs
with
stamen
14
and
17
and
with
stamen
22
and
25.
76
BLUMEA
-
VOL.
31,
No.
1,
1985
A
number
of
tropical
species
has
stamens
closely
comparable
to
those
of
Degene-
riaceae
and
Himantandraceae,
viz.
broad
3-veined
microsporophylls
with
two
pairs
of
linear
sporangia
deeply
embedded
in
their
surface.
Trends
of
specialization
found
in
the
stamens
of
the
Magnoliaceae
are:
1.
elongation
of
the
apices;
2.
differentiation
of
a
filament;
3.
reduction
in
number
of
veins
from
three
to
one;
4.
an
increase
in
the
relative
size
of
the
sporangia
and
a
concomitant
increase
in
their
amount
of
pro-
tuberance;
5.
transition
from
a
laminal
to
a
marginal
position
of
the
sporangia;
and
6.
development
of
an
enveloping
fibrous
layer
in
the
'connective'
(Canright,
1952).
The
stamens
mostly
are
innervated
by
three
veins.
The
median
trace
normally
de-
parts
from
the
stelar
system
of
the
floral
axis
while
the
two
lateral
depart
from
the
cortical
system.
In
several
species
(most
in
Michelia)
only
one
trace
is
present.
In
other
genera
also
5-7
traces
can
be
found,
often
in
the
same
flower
as
stamens
with
three
traces.
Teratological
stamens
often
occur,
from
broad
petaloid
stamens
to
narrow
filamented
ones.
They
were
found
in
the
same
flower
of
Michelia
champaca.
Also
carpels
with
varying
degrees
of
fertility
were
found
among
the
stamens.
Carpels
The
carpels
are
arranged
spirally
on
the
lengthened
receptacle.
As
said
above,
in
Magnolia
stellata
at
least
they
arise
in
the
same
spirals
as
stamens
and
tepals,
i.e.
they
arise
in
the
same
8
(sometimes
7
or
9)
parastichies.
Tucker
(1961)
found
the
same
for
the
carpels
in
Michelia
fuscata
of
which
she
investigated
the
phyl-
lotaxis
and
the
vascular
organization.
The
phyllotaxis
and
the
vascular
organization
are
closely
related.
The
carpels
in
this
species
are
initiated
along
each
of
7,
8,
or
10
helical
parastichies
according
to
a
complex
repetitive
sequence.
The
pattern
of
the
dorsal
carpellary
trace
fusions
is
orderly
for
each
of
the
flowers
investigated.
Among
flowers
one
finds
differing
numbers
of
parastichies,
different
angles
of
divergence,
and
varying
sequences
of
parastichies
which
reflect
the
order
of
carpel
initiation.
The
angle
of
divergence,
although
consistent
for
any
one
parastichy,
can
vary
greatly
be-
tween
parastichies.
The
carpels
of
the
Magnoliaceae
(and
the
related
Himantandraceae,
see
Canright,
1960)
exhibit
drastic
trends
of
modification
from
the
primitive
ranalian
megasporo-
phylls.
The
most
significant
specializations
are:
1)
closure
of
the
fertile
part
of
the
carpel
(but
see
fig.
2
which
shows
that
the
carpels
in
Magnolia
nitida
possess
free
margins);
2)
restriction
of
the
external
stigmatic
surfaces
to
the
margins
of
the
con-
duplicate
style;
and
3)
conspicuous
changes
in
the
vascularization
of
the
carpels.
There
is
also
a
tendency
towards
a
reduction
in
the
number
of
ovules
(to
two
in
many
Magnoliaceae)
and
to
lateral
adherence
or
actual
concrescence
of
crowded
car-
pels.
The
carpels
of
Magnoliaceae
exhibit
more
or
less
advanced
stages
of
phylogene-
tic
modification.
Thus
the
gynoecium
is
in
contrast
to
the
androecium
where
relative-
ly
primitive
forms
of
microsporophylls
are
retained
by
a
number
of
species.
The
carpels,
too,
are
supplied
by
three
veins.
A
dorsal
trace
runs
upwards,
enters
the
carpel,
and
then
suddenly
turns
downwards.
At
this
point
the
so-called
ascending
strand
is
diverged
from
the
dorsal
strand
and
gradually
bifurcates
to
branch
off
the
ovular
strands
and
to
finally
unite
with
the
two
ventral
strands.
The
dorsal
traces
commonly
originate
from
the
stelar
system
while
the
two
ventral
traces
originate
from
the
cortical
system.
Canright
(1960),
however,
found
in
cleared
material
from
77
H.P.
Nooteboom:
Notes
on
Magnoliaceae
Michelia
champaca
that
'not
uncommonly,
the
basal
carpels
of
a
gynoecium
are
vas-
cularized
entirely
by
the
cortical
system,
and
the
apical
carpels
entirely
by
the
stelar
system.'
Fruits
The
fruits
of
Magnoliaceae
principally
consist
of
one
(rarely,
in
Michelia
montana)
to
several
or
many
free
carpels
spirally
arranged
around
the
recep-
tacle.
The
fruiting
carpels
are
woody
and
entirely
free
in
some
taxa
while
they
are,
secondarily,
more
or
less
connate
in
others.
The
young
carpels
are
always
free
when
initiated,
and
only
relatively
late
in
ontogeny
they
may
become
concrescent.
Van
Heel
(1981,
1983)
suggests
that
the
carpels
are
basically
cupuliform.
He
criti-
cizes
conclusions
entirely
based
on
investigation
of
mature
carpels.
After
investiga-
tion
of
ontogenetical
stages
with
a
scanning
electron
microscope
he
comes
to
the
conclusion
that
carpels
arise
as
cup-shaped
(ascidiate)
or
partly
cup-shaped
organs
lateral
or
terminal
on
the
flower
apex.
In
Magnoliaceae
the
carpels
arise
as
hemi-
spherical
primordia
according
to
Van
Heel.
The
development
begins
with
the
widen-
ing
of
the
lower
half
of
the
hemispherical
primordium,
after
which
a
thick
margin
grows
outwards
and
also
upwards.
The
upper
(adaxial)
region
of
the
initial
primor-
dium
develops
into
the
inside
margin
of
the
cup.
The
adaxial
inside
margin
originally
is
distinct,
as
well
as
the
obliquely
cup-shaped
lateral
development
of
the
carpel
as
a
whole.
Later
on,
however,
this
adaxial
region
cannot
be
distinguished
any
more
from
the
floral
axis.
According
to
Canright
the
margins
of
the
carpels
become
adnate
to
the
floral
axis.
According
to
Van
Heel's
observations
there
is
no
later
fusion,
but
al-
ready
in
an
early
state
there
is
no
abaxial
margin
present.
On
the
side
of
the
floral
axis
the
cup
is
deepened
secondarily,
and
the
ovules
originate
in
it
laterally
on
the
inside
margin
close
to
the
floral
axis
when
an
ovary
and
a
style
have
not
yet
differentiated.
Dehiscence
of
the
mature
carpels
is
mostly
along
the
dorsal
suture,
often
also
along
the
ventral
suture.
Sometimes
the
carpels
become
bivalved,
the
two
valves
only
adnate
to
the
central
axis,
shedding
their
seeds.
Sometimes
the
carpels
form
a
pseudo-
syncarp,
becoming
more
or
less
concrescent.
But
only
the
outer
layers
of
the
exocar-
pium
are
really
concrescent,
giving
the
fruit
the
appearance
of
a
syncarp.
In
the
latter
case
the
dorsal
(abaxial)
parts
of
the
carpels
finally
fall
away,
leaving
their
base,
which
is
inbedded
in
the
receptacle,
exposed
with
the
seeds
mostly
hanging
from
their
placenta
on
the
elongate
funicle
which
in
that
stage
consists
of
spiral
vessels.
In
some
taxa
abaxial
parts
of
the
carpels
when
falling
also
dehisce
along
the
dorsal
suture
(Tsoongiodendron,
but
also
in
some
Talauma
species
and
in
Manglietiastrum,
fig.
3).
In
Magnolia
nitida
and
M.
kachiraricharai
the
carpels
are
essentially
connate.
When
maturing
they
tear
apart
and
dehisce
along
the
dorsal
suture
thus
giving
the
appearance
of
a
common
Magnolia
fruit
(fig.
1,2).
This
condition
probably
exists
in
many
species
with
crowded
carpels.
The
concrescence
of
the
carpels
has
apparent-
ly
developed
independently
in
different
lineages
of
Magnoliaceae.
Species
with
free
and
concrescent
carpels
sometimes
are
very
closely
allied,
judging
from
the
other
characters.
Therefore
I
do
not
accept
the
concrescence
of
the
carpels
alone
as
a
good
character
for
delimitation
of
genera
and
I
consider
e.g.
Paramichelia
and
Tsoongio-
dendron
as
congeneric
with
Michelia.
No
more
do
I
see
reason
to
split
Elmerrillia
on
the
base
of
this
character.
78
BLUMEA
-
VOL.
31,
No.
1,
1985
Subdivision
of
Magnoliaceae
The
family
can
easily
be
divided
into
two
subfamilies:
Magnolioideae.
Leaves
entire
or
occasionally
2-lobed
at
the
apex;
stipules
free
from
the
petiole
or
adnate
to
it.
Anthers
introrse
or
latrorse.
Fruiting
carpels
longitudinally
dehiscent
or
circumscissile,
at
least
the
base
remaining
adnate
to
the
torus,
free
or
concrescent
into
a
syncarp,
never
samaroid.
Testa
free
from
the
en-
docarp,
externally
arilloid.
-
Liriodendroideae
(Bark.)
Law
Yuh-wu.
Leaves
2-10-lobed,
the
apex
truncate
or
widely
emarginate;
stipules
always
free
from
the
petiole.
Anthers
extrorse.
Fruit-
ing
carpels
indehiscent,
samaroid,
produced
at
the
apex
into
a
long
wing-like
beak,
caducous.
Testa
adherent
to
the
endocarp.
Only
one
genus,
Liriodendron,
which
will
not
be
treated
here.
2.
SUBFAMILY
MAGNOLIOIDEAE
In
subf.
Magnolioideae
two
tribes
clearly
can
be
recognized.
KEY
TO
THE
TRIBES
1
a.
Growth
sympodial.
Flower
buds
arising
terminal
on
the
twigs
A.
Tribus
Magnolieae
b.
Growth
monopodial.
Flower
buds
arising
on
brachyblasts
in
the
axis
of
the
leaves
B.
Tribus
Michelieae
GENERIC
DELIMITATION
IN
MAGNOLIEAE
The
problem
of
Dugandiodendron.
The
genus
Dugandiodendron
Lozano-Con-
treras
(1975)
differs,
according
to
Lozano,
from
Talauma
in
the
position
of
the
flower
and
in
the
prefoliation
of
the
vegetative
buds.
The
morphology
of
those
char-
acters
is
extensively
discussed.
Lozano
recognizes
in
the
family
Magnoliaceae
two
types
of
anthotaxis:
1)
a
terminal
flower
('flor
terminal')
and
2)
a
pseudolateral
flower
('flor
pseudolateral').
He
states
that
the
so-called
pseudolateral
flower
is
found
in
the
genera
Alcimandra,
Elmerrillia
and
Michelia
and
his
newly
described
Dugandioden-
dron.
As
a
matter
of
fact,
the
flowers
in
Alcimandra
are
terminal
as
they
are
in
Mag-
nolia
and
Talauma.
Alcimandra
was
described
because
of
its
stipitate
gynoecium.
In
Elmerrillia
and
Michelia
the
flowers
(or
flowershoots)
are
axillary
and
the
growth
monopodial.
From
Lozano's
description
of
Dugandiodendron
'Flores
solitarii
pseu-
dolaterali
(ob
ramificationem
sympodialem
pro
maxima
parte
accrescentum
acro-
tomicam
in
vertice
ramulorum)',
as
well
as
from
the
drawings
and
the
collections
that
I
have
seen
(the
lectotypes
of
Magnolia
ptaritepuiana
Steyermark
and
M.
roraimae
Steyermark,
both
from
FI)
it
is
clear
that
the
flowers
in
his
genus
are
terminal
as
they
are
in
Magnolia
and
Talauma.
From
Lozano's
commentary
in
Taxon
(1984)
I
understand
that
he
distinguishes
between
a
terminal
flowerbud
giving
rise
to
one
pair
79
H.P.
Nooteboom:
Notes
on
Magnoliaceae
of
bracts
and
one
flower
only
in
Dugandiodendron
and
a
terminal
flowerbud
giving
rise
to
a
short
shoot
with
more
than
one
pair
of
bracts,
the
latter
morphologically
being
stipules
often
possessing
a
petiole
and sometimes
a
leaf
as
well,
in
Magnolia
and
Talauma.
Both
states
are
met
elsewhere
in
Talauma
and
Magnolia
and
do
not
constitute
a
differential
character
at
genus
niveau.
The
other
differential
character
mentioned
by
Lozano
is
the
prefoliation
in
the
vegetative
buds.
He
distinguishes
between
'prefoliacion
erecta
convoluta'
and
'prefo-
liacion
erecta
heliciconvoluta'.
The
latter
is,
according
to
Lozano,
only
found
in
Du-
gandiodendron
and
probably
in
Elmerrillia
and
is
accompanied
by
obvious
longitudi-
nal
marks
on
the
lamina.
However,
those
marks
are
often
hardly
or
not
at
all
visible
according
to
both
Lozano's
and
my
own
observations.
In
my
opinion,
and
according
to
Van
Heel
(pers.
comm.,
1983)
who
examined
preserved
buds
of
Elmerrillia
ovalis,
collected
by
me,
it
is
only
a
matter
of
degree
whether
the
prefoliation
is
convolute
or
heliciconvolute.
In
Elmerrillia
the
younger
buds
are
often
the
convolute
kind,
while
the
older
ones
sometimes
become
heliciconvolute.
As
a
matter
of
fact,
in
a
Ta-
lauma
species
from
Mt
Kinabalu,
Sabah,
North
Borneo,
which
is
not
yet
described
(Carr's
Talauma,
W.
Meijer,
1968;
P.
Cockburn,
1980),
the
same
marks
occur,
de-
scribed
by
Meijer
as
'2
furrow-like
lines
of
depression
at
both
sides
of
the
midrib
or
less
parallel
with
the
leaf
margin.'
These
lines
originate
from
the
same
kind
of
pre-
foliation
as
in
Lozano's
Dugandiodendron.
The
only
conclusion
I
can
reach is
that
Dugandiodendron
was
described
on
differential
characters
that
I
fail
to
observe.
Dr.
Praglowski
in
Stockholm
(author
of
Magnoliaceae
in
World
Pollen
and
Spore
Flora
3,1974)
has
been
so
kind
as
to
look
at
the
pollen
of
Magnolia
(Dugandioden-
dron)
ptaritepuiana.
His
conclusion
is
that
the
pollen
'can
perfectly
well
be
con-
sidered
a
Magnolia
species
....
The
exine
surface
with
rugulose
exine
structure
and
tectal
perforations
....
both
features
which
are
very
typical
for
Magnolia
pollen.'
From
the
above
I
conclude
that
the
species
ascribed
to
Dugandiodendron
by
Lozano
belong
to
Magnolia
or
Talauma.
As
I
consider Talauma
to
be
congeneric
with
Magno-
lia,
all
species
of
Dugandiodendron
belong
to
Magnolia.
Delimitation
of
the
other
genera
of
Magnolieae.
After
coming
to
the
conclu-
sion
that
in
Michelia
and
Elmerrillia
concrescence
of
the
carpels
is
of
no
importance
as
a
character
on
generic
level,
but
can
be
on
a
species
level
(example:
it
is
the
only
character
that
always
holds
good
between
Elmerrillia
ovalis
and
E.
tsiampacca),
I
came
to
study
the
species
of
Magnolia
sect.
Gynopodium
Dandy.
The
species
are
characterized
by
a
usually
shortly
stalked
gynoecium,
a
complete
lack
of
hairs,
and
comparatively
small
glossy
green
leaves
with
free
stipules.
Because
of
the
gynophore
Keng
(1955)
based
his
genus
Micheliopsis
on
Magnolia
kachirachirai,
which
had
been
described
as
Michelia
kachirachirai
by
Kanehira
&
Yamamoto
because
of
the
same
character
and
reduced
to
Magnolia
by
Dandy
(1927).
Keng
later
(1976)
reduced
the
genus
to
Magnolia.
An
interesting
feature,
already
described
by
Keng
(1955),
is
that
the
carpels
are
connate
when
young
(Keng,
1978,
joined
already
the
genera
Magnolia,
Manglietia,
Talauma,
and
Aromadendron).
When
they
become
ripe
they
tear
apart
BLUMEA
-
VOL.
31,
No.
1,1985
80
and
open
along
the
dorsal
suture.
The
number
of
ovules
in
Magnolia
kachirachirai
was
described
as
two.
I
found,
however,
4
ovules
in
each
carpel
(
Liao
&
Lai
10534,
Taiwan,
Hengchun
Peninsula).
To
be
certain
that
this
species
does
not
belong
to
Manglietia
(although
it
is
glabrous
whereas
all
Manglietia
species
possess
hairs),
Dr.
P.
Baas
(L)
examined
the
leaf
anatomy.
The
epidermis
is
weakly
sclerified,
the
hypo-
dermis
absent,
spongy
parenchyma
('arm
parenchyma')
not
sclerified.
The
vein
sheath
cells
are
lignified
(it
is
not
clear
in
sections
whether
there
is
an
additional
un-
lignified
sheath
of
lobate
cells).
The
leaf
margin
is
not
sclerified.
Terminal
cells
are
difficult
to
observe
in
sections.
The
conclusion
is
that
the
leaf
anatomy
rules
out
that
this
specimen
belongs
to
Manglietia.
position
of
of
flower
carpels
prefoliation
of
bud
stipitate
gynoecium
growth
pollen
Dugandiodendron
terminal*
concrescent
or
free
heliciconvolute
no
sympodial
magnolia
type**
Magnolia
terminal
concres-
cent
or
free
convo-
lute
some-
times
sympo-
dial
magnolia
type
Talauma
Aromadendron
terminal
concrescent
convolute
or
heliciconvolute
no
sympodia