A Monograph of the Genus Geranium L. (Geraniaceae).
Author: Carlos Aedo
2023
Real Jardín Botánico, Madrid.
Consejo Superior de Investigaciones Científicas (CSIC).
Editorial CSIC: publ@csic.es
2. This long awaited monograph deals with the
fascinating group of plants known as geraniums or
cranesbills. It is restricted to the plants covered by
the scientific name Geranium, but does not include
those plants known in English as “geranium” which
botanists have transferred to the genus Pelargoni-
um. The book is the result of 28 years of research by
the author. It is lavishly illustrated with maps, line
drawings and photographs.
There is a gap in the available botanical litera-
ture covering the geraniums. The last monographic
treatment by Reinhard Knuth appeared in 1912. In
the intervening years, much taxonomic information
has been accumulated, rendering this massive and
comprehensive work out of date. There have been
many new species discoveries and much research
has been conducted for revisional works. All these
changes have caused much confussion to the out-
sider, making the need for a comprehensive mono-
graph even more pressing.
For each of the 307 species covered in this book,
complete synonymy and detailed description is
given, as well as its distribution, relationships, vari-
ability and taxonomic history are discussed. Each
species is provided with a map showing its known
distribution based on specimens examined by the
author, and is illustrated with line drawings and,
if available, with colour photographs based on re-
liable specimens. Some introductory sections ex-
plain in detail the morphological structure of these
plants, chromosome numbers, breeding system,
phytochemistry, distribution, habitats and rela-
tionships. For practical reasons and to facilitate the
identification of the species, four identification keys
have been prepared that cover respectively North
and Central America, South America, the Pacific
area and the Old World species. This book will be of
value to botanists, but also for specialists hobbyists
and to anyone fascinated by these plants.
7. CONTENTS
INTRODUCTION 11
HISTORICAL OVERVIEW 12
MATERIAL AND METHODS 15
MORPHOLOGY 16
Duration and Habit 16
Indumentum 17
Leaves 18
Inflorescence 21
Calyx 22
Corolla 22
Androecium 24
Pollen 24
Gynoecium 24
Fruit 26
Seeds 31
CHROMOSOME NUMBERS 32
BREEDING SYSTEM 33
PHYTOCHEMISTRY 34
DISTRIBUTION AND HABITATS 35
RELATIONSHIPS BASED ON DNA MARKERS AND CLASSIFICATION 36
CONSPECTUS OF CLASSIFICATION 39
TAXONOMY 41
Key to species 41
Geranium subg. Erodioidea 66
Geranium subg. Geranium 110
Geranium subg. Tuberosa 649
Geranium subg. Robertium 692
MISCELLANEOUS DOUBTFUL AND NOT VALIDLY PUBLISHED NAMES 786
HYBRIDS 798
ACKNOWLEDGMENTS 799
LITERATURE CITED 800
APPENDIX 1. CHROMOSOME COUNTS OF GERANIUM 813
APPENDIX 2. NAMES OF GERANIUM BELONGING TO OTHER GENERA 830
INDEX TO NUMBERED COLLECTIONS EXAMINED 836
INDEX TO SCIENTIFIC NAMES 883
INDEX TO NEW NAMES AND COMBINATIONS 898
9. 9
“Butanicum herbarum dicitur quod ibi herbae notentur”
[Se dice Butanicum (botánico) de hierbas a (el sitio) donde las hierbas son anotadas;
Herbal Butanicum (botanist) is said to (the place) where herbs are noted down]
Isidoro de Sevilla, Etimologías, Libro IV. 10.4 (c. 625)
[Stearn 1992: 22]
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
11. 11
Introduction
INTRODUCTION
Geranium L., with 307 species, is the largest genus of the
Geraniaceae, a family that according to Stevens (2001),
also contains the genera Pelargonium L’Hér. (280 species),
Erodium L’Hér. (80 species, including California Aldasoro
al.), Monsonia L. (40 species, including Sarcocaulon Sweet)
and Hypseocharis Remy (2 species; Lozada-Gobilard al.
2020). Hypseocharis, however, differs from the other gen-
era by its estipulate leaves, capitate stigma, ovaries with
many ovules per loculus, and unbeaked fruits. This set of
characters may support assignment to a monotypic fam-
ily sister of the Geraniaceae core (Aedo 2012, Palazzesi
al. 2012). Retaining Hypseocharis would result in the Gera-
niaceae losing the main characters (stipulate leaves, style
with 3-5 stigmatic branches, one or two ovules per loculus,
beaked fruits) that define the family. Regarding the generic
division of the family, the synonymization of California, and
especially those of Sarcocaulon seems controversial (Albers
1996, Dreyer al. 1997, Moffett 1979, 1997).
After the segregation of Erodium, Monsonia, Pelargonium
(see Historical overview) from Geranium, the circumscription
of the genus has been stable throughout its recent history.
Three attempts to separate some species into independent
genera have not been followed by most of the authors: Rober-
tium to accommodate some species of sect. Ruberta (Picard
1837), Neurophyllodes to accommodate the species of sect.
Neurophyllodes (Degener Greenwell 1937), and Geraniopsis
to accommodate two species of sect. Trilopha (Chrtek 1968).
The genus Geranium is easily differentiated from the
other genera of the family by its usually actinomorphic
flowers with five free petals, without hypanthium nor nec-
tariferous tube, and with ten free stamens arranged in two
whorls, very rarely the outer lacking the anthers. Geranium
is distributed throughout most of the world except in low-
land tropical areas, but it has diversified more in certain
regions. The mountains around the Mediterranean Basin,
the Caucasus and the Himalayas constitute some of these
regions. Regarding the Old World, also South Africa and the
mountains of New Guinea are very rich in endemic species.
In the New World, the two areas particularly diverse in spe-
cies are Mexico and adjacent regions of western U.S.A., and
the north of the Andes, from Venezuela to northern Peru.
Knuth’s (1912) worldwide monograph of Geranium was
the most complete treatment of the genus during several
decades, and a good starting point for an overview of the
genus. This author accepted 269 species plus 72 more spe-
cies in subsequent papers (Knuth 1913, 1915, 1916, 1922,
1923, 1930a, 1930b, 1931, 1933, 1934, 1936, 1937a, 1937b,
1938, 1939). This impressive monograph causes even more
astonishment when we consider that it covers the whole Ge-
raniaceae, in a broad sense, including small related families.
According to Melchior (1957: 185), Reinhard Knuth re-
ceived his doctorate in 1902 with the dissertation entitled
Über die geographische Verbreitung und die Anpassungser-
scheinungen der Gattung Geranium im Verhältnis zu ihrer sys-
tematischen Gliederung (Knuth 1903, 1904). He collaborated
with F. Pax in the monograph treating Primulaceae pub-
lished in 1905 in the Engler’s Pflanzenreich. Subsequently
he wrote monographs on Geraniaceae for the same se-
ries directed by A. Engler. The main work seems to have
been carried out in a relatively short period, perhaps less
than ten years, taking into account that his first taxonomic
publication on the genus is from 1904 (Knuth 1904). This
monograph should be seen in the context of its time, in
which many areas were still unexplored or undercollected
and loans between herbaria were not routinely done. This
resulted, as an unavoidable consequence, in the scarcity of
material for preparing full descriptions and keys, and for
obtaining detailed distributions of the species. These lim-
itations led to the description of some species that are now
seen as synonyms. This however does not diminish the val-
ue of Knuth’s work (Fig. 1).
Among the relevant contributions after Knuth’s mono-
graph, many floras can be mentioned. The list is, however,
lengthy and it would be difficult to do justice to all the ef-
forts of the authors of these works. One relevant exam-
ple would be Flora of the U.R.S.S. (Bobrov 1949), that covers
more than 22 million km2
, and presents a valuable taxo-
nomic treatment of Geranium (55 species). Other important
works that cover areas poorly known are those of Moore
(1943) for species of Mexico and Central America, Jones
Jones (1943) for United States and Canada, Carolin (1965)
for the southwestern Pacific area, Veldkamp Moerman
(1978) for the Malesian species, and Hilliard Burtt (1985)
for South African species.
A considerable advance in the knowledge of Geranium
occurred with the works of P. Yeo, some focused on classi-
fication and others that studied various groups in particu-
lar. Among works on classification, it should be mentioned
the new proposal for an infrageneric classification by Yeo
Fig. 1. Reinhard Knuth’s handwritten label used to identify the plants
studied in his monograph. Specimen preserved in the herbarium of
University of Zurich (Z-000064950).
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
13. 13
Historical overview
guineum (Fig. 2) are found in the herbarium attributed to
Michelle Merini. A recent study suggests that there is in-
sufficient evidence to attribute this herbarium to Merini
and that it is more appropriate to consider it anonymous
(Cristofolini Nepi 2021). This herbarium is kept at the
Museo di Storia Naturale, University of Florence (FI) and
dated ca. 1544. It seems that it was compiled with plants
cultivated in the Pisa botanical garden. They are the oldest
known herbarium specimens of Geranium. Since that time
it is referred to as the Cibo herbarium, attributed to Fran-
cesco Petrollini. It contains thirteen specimens of Geranium
(G. dissectum, G. lucidum, G. macrorrhizum, G. nodosum, G.
rotundifolium, and G. sanguineum) (Fig. 3). This herbarium,
kept in the Biblioteca Angelica, Rome, is dated from 1550-
53 (Celani 1902; Stefanaki al. 2019). The specimens were
probably collected in Italy.
According to Caruel (1858: 114-115), in the herbarium
of Andrea Cesalpino specimens of G. columbinum, G. dis-
sectum, G. nodosum, G. robertianum and G. sylvaticum are
preserved (Fig. 4). The specimens, probably from northern
Italy, are organized following a new system of classification
based on Aristotle and Theophrastus, later published by
Cesalpino (1583) as De Plantis libri XVI. The geraniums are
mentioned on the pages 558-559. This herbarium is kept at
the Museo di Storia Naturale, University of Florence (FI) and
dated from 1563 (Nepi 2007; Nepi Gusmeroli 2008).
From this time on, the information about the plants of
Europe and the Mediterranean Basin is increased with the
appearance of numerous books and, progressively more
collections preserved in herbaria. Caspar Bauhin (1623:
317-319) in his Pinax Theatri Botanici, mentions 26 gerani-
ums, with a complete synonymy, giving his work an innova-
tive character for its time. He describes different species of
Geranium and Erodium which are widely distributed Europe
that can be easily identified from the herbarium vouchers
kept at the University of Basel, Switzerland (BAS).
John Ray (1688: 1055-1063) in his Historia plantarum
recognized 40 species of Geranium (in a broad sense) that
were expanded considerably in a supplement (Ray 1704:
511-515). In this work he introduces dichotomous keys to
identify the genera. His herbarium vouchers are kept at
The Natural History Museum, London (BM).
Tournefort (1700: 266-270) in his Institutiones rei herbar-
iae accepts Geranium as a genus. The effort of this author
had great influence on consolidation and consistent defini-
tion of the genera. The number of species accepted for the
genus increased to 81 (including Geranium and Erodium),
mostly Europeans but also from the Middle East where
Tournefort himself carried out an expedition. His herbar-
ium, kept at the Muséum National d’Histoire Naturelle, Paris
(P-TRF), can help to clarify the complex polynomial nomen-
clature that Tournefort applied to his species.
The binomials that were used in an irregular way by
previous authors were consolidated with the Linnaeus’s
Species plantarum, which is accepted as the starting point
of modern nomenclature. Linnaeus’s (1753: 676) circum-
scription of Geranium included Erodium, and Pelargonium,
as the previous authors did. However, Geranium s.str. was
recognized implicitly by Linnaeus as the group defined by
“Staminibus decem fertilibus”, in which he included fifteen
species still accepted in Geranium s.str. today. Most of Lin-
naeus’s geraniums are European (except those now trans-
ferred to Pelargonium) but there are two American and one
Siberian. The former are G. carolinianum and G. macula-
tum that Linnaeus cited through Gronovius’s Flora virginica
(1743: 78), which was based on John Clayton’s collections.
Among the Clayton specimens at BM, only the original ma-
terial of G. carolinianum is extant. It is the earliest collection
of the genus in North America and probably dates from ca.
1735 (Reveal Pringle 1993: 161).
Cavanilles (1787) published an important revision of the
Geraniaceae, which listed 127 species, following Linnae-
us’s generic concept of Geranium and accepting the new
genus Monsonia (Linnaeus 1767). Cavanilles had access
to important extra-European collections in the Paris her-
barium, which allowed him to work with the South African
geraniums (now under Pelargonium). In his monograph he
Fig. 3. Specimen of Geranium nodosum from Cibo herbarium, attributed
to Francesco Petrollini, prepared in 1550-53, and kept at the Biblioteca
Angelica, Rome [Erbario B 129/498].
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
15. 15
Material and methods
raniaceae, but maintained the narrow circumscription of
Geranium and attributed to the genus about 100 species.
Dumortier (1827) accepted a division of the genus in
three sections which only partially agreed with De Candolle
(1824) and is restricted to Belgian species. Koch (1835) also
proposed three sections, agreeing in part with those of pre-
viously mentioned authors but expanding the geographic
area and the number of species. Picard (1837), Boissi-
er (1867), Rouy (1897), Bubani (1901) and Fiori Paoletti
(1901) also organize the species in their respective work
areas into infrageneric categories. It is necessary, however,
to wait for Knuth’s (1912) monograph to find a proposal for
a worldwide infrageneric classification.
After De Candolle’s synthesis, a large number of new
species were published, generally linked to the explora-
tion of little-known areas. These were evaluated in Knuth’s
(1912) monograph. More recent relevant treatments are
noted in the Introduction.
MATERIAL AND METHODS
This revision is based on 25,953 herbarium specimens
from the following herbaria: A, AAU, ABS, AD, AK, ALA, ANG,
APP, ARAN, ARIZ, ATH, B, BAF, BC, BCN, BEOU, BH, BIRM,
BISH, BM, BO, BOL, BOLV, BP, BR, BRIT, BRNM, BRNU, BRSL,
BUCA, C, CAL, CAN, CANB, CAR, CAS, CDBI, CGE, CHR, CL,
CLU, CM, COFC, COI, COL, COLO, CONC, CORD, CTES, DAO,
DBN, DUKE, E, ENCB, ERE, F, FI, FT, G, GA, GB, GDA, GH,
GOET, GRA, GRM, GZU, H, HA, HAL, HAO, HBG, HBIL, HGI,
HO, HUA, HUB, HUJ, HULE, HVR, IBF, IBSC, ICEL, IEB, ILL,
INB, IND, ISC, ISTE, ISTO, JACA, JBAG, JBSD, JE, JEPS, K, KANU,
KASH, KE, KR, KRA, KUH, KUN, KUS, KW, KY, KYO, L, LD, LE,
LEB, LIL, LINN, LISE, LISI, LIV, LOJA, LOU, LP, LPA, LPB, LUX,
LW, LWG, LY, LZ, M, MA, MACB, MADJ, MADM, MAF, MEL,
MER, MERL, MEXU, MFU, MICH, MIN, MO, MONTU, MOR,
MPU, MSB, MSC, MSTR, MU, MUB, MW, NA, NAP, NBG, NDG,
NEB, NEU, NHMC, NMC, NMW, NSK, NSW, NU, NY, O, OBG,
OBI, OKL, ORE, OSC, OXF, P, PACA, PAL, PAVL, PE, PERTH,
PH, PO, PORT, PR, PRC, PRE, QCA, QFA, QPLS, R, RB, RENO,
RO, RSA, S, SALA, SAPS, SBBG, SD, SEV, SGO, SI, SIB, SKO,
SLBI, SOM, SP, SS, STU, SYD, SZ, TAA, TAI, TAUF, TBI, TENN,
TEX, TFC, TI, TK, TNS, TUB, U, UB, UC, UNM, UPA, UPS, US,
USM, V, VEN, VLA, VT, W, WA, WAG, WELT, WIS, WRSL, WTU,
WU, WVA, XAL, Z, ZT, and the private herbaria of J. Bouch-
ard, G.G. Guittonneau, S. Halloy and G.H. Parent.
The revision includes 652 herbarium specimens of Gera-
nium belonging to 110 species and assessments of natural
populations obtained by the author in field trips to Argen-
tina, Bolivia, Brazil, Chile, Costa Rica, Ecuador, Nicaragua,
Peru, the United States, and Venezuela in the New World
and to Armenia, Austria, Bulgaria, Cyprus, Equatorial Guin-
ea, France, Germany, Great Britain, Greece, Italy, Morocco,
Norway, Portugal, Romania, Slovenia, South Africa, Spain,
Switzerland, Tunisia, and Turkey in the Old World.
The account presented in the Taxonomy includes a com-
prehensive synonymy. All names have been verified in
their original publication. The names that share the same
type are indicated in the same paragraph, ordered by pub-
lication date. In each paragraph there is a heading named
“type locality” in which the geographical information of the
protologue is literally transcribed. These data are followed
by the type, giving the information that comes from the la-
bel in a standard way, and supplying the geographic or col-
lector information that may be missing. The disposition of
names for which no type material was available is based on
the opinions of previous authors or in a few cases, on the
original description. Invalidly published names and names
that could not be fixed with certainty are listed under Mis-
cellaneous doubtful and not validly published names (see also
Appendix 2 for names which now belong to other genera).
One hundred nine (109) characters, 72 qualitative
traits, and 32 quantitative traits, as well as five ratios were
scored for the description of the species. At least 15 spec-
imens were scored for each species, if available. Quanti-
tative characters were assessed with a Mitutoyo CD-15CD
digital caliper (Mitutoyo, Kawasaki, Japan). The most fre-
quent values are percentiles (between the 25th and 75th
percentiles), are shown without brackets, and the ex-
treme values are enclosed in brackets. All measurements
were taken from herbarium specimens. The line drawings
of leaf laminas and petals show the adaxial surface, and
drawings of the sepals show their abaxial surface, unless
stated otherwise. In the descriptions and key, the length
of the sepal does not include the length of the mucro. The
use of the term “Leaves” refers to the lamina of the leaf
unless stated otherwise. Similarly, the description of the
leaf segments refers to the middle segment unless other-
wise indicated.
Under the heading Representative specimens examined a
selection of specimens representative of the distribution of
the species is indicated. When the species is rare or few
specimens have been studied, all of them are listed under
the Additional specimens examined section. In no case are
the type specimens repeated, as they have already been
mentioned together with their corresponding names. The
specimen labels have been transcribed in the language
in which they were written. Only in those cases written in
non-Latin alphabets (Chinese, Japanese, Korean or Rus-
sian), have they been transliterated to the Latin alphabet,
either to English or to Spanish. For country names, the 1999
edition of The Times comprehensive Atlas of the World [Times
Books, London] has generally been followed. Changes after
that date have been incorporated when considered useful.
In the case of the Democratic Republic of the Congo, how-
ever, the name of Zaire has been maintained, because it is
widely used in the literature and avoids confusing with the
neighboring Republic of the Congo.
Each species has been mapped based on the studied
specimens. When the specimens lacked geographic coordi-
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
17. 17
Morphology
habit is found in the Macaronesian perennials G. palmatum
and G. reuteri, and the monocarpic G. maderense.
Some species, mainly from sect. Geranium, but also from
sect. Aculeolata and sect. Neurophyllodes, are characterized
by prostrate to ascending stems occasionally rooting at the
nodes. An unusual pattern is seen in G. stoloniferum, which
has a basal rosette of leaves giving rise to one erect main
stem with a well-developed dichasial inflorescence and one
or two herbaceous procumbent stolons. These epigeal sto-
lons, borne from the inner part of the rosette, reach up to
20 cm long, have several pairs of leaves, and terminate in
a bud. They seem to result from the growth of one year.
Geranium stoloniferum has also some hypogeous, lignified
stolons. These are borne from the external part of the ro-
sette, are ramified, and reach up to 100 cm long. The ligni-
fied stolons bear several secondary rosettes and terminate
in a rosette as well. It is possible that these structures orig-
inate from herbaceous stolons. Epigeal stolons have also
been found in five species of the Papuanum Group.
Some species from the northern and central Andes (the
Sibbaldioides Group and the Multiceps Group) and from New
Guinea (the Papuanum Group) have structures that have
been termed “vegetative stems” in order to facilitate de-
scription. These species produce a kind of bud towards the
apex of the branches, sometimes quite long and covered
by stipules (usually without petiole remains), which bears
a rosette of leaves and an inflorescence (see drawings for
a better understanding). This structure may represent the
growth of each vegetative period.
The underground parts of perennials (roots and root-
stocks) are generally not thoroughly studied owing to their
poor representation in herbaria.
Probably as a response to arid climatic conditions from
the Mediterranean Basin and central Asian highlands
some groups have developed tuberculate rootstocks (sect.
Tuberosa and two species of the Pylzowianum Group). In a
similar way, the turnip-shaped rootstocks found on species
of the Sessiliflorum Group, the Solanderi Group, the Poten-
tillifolium Group, the Diffusum Group, the Rupicola Group,
the Mexicanum Group, the Renifolium Group and the sect.
Incana may have evolved to withstand drought conditions
in South Africa, South America, Mexico and Oceania.
Species with rootstocks that have thickened roots are
widespread among different groups of Geranium. Normally
these roots are arranged along the rhizome but in some
species of the Palustre Group, the Krameri Group, the Sibiri-
cum Group and one of the Pylzowianum Group (sect. Gerani-
um) plus in one species of sect. Dissecta the thickened roots
are grouped in a fascicle.
Indumentum
Most Geranium species are covered in variable density by
various types of hair. Exceptionally some species are al-
most entirely glabrous (e.g., G. glaberrimum and G. lucidum).
The following types of hairs, all simple and uniseriate
(Theobald al. 1979), have been found in Geranium (Figs.
6-8). (1) Eglandular hairs (0.1-5.4 mm long), unicellular,
smooth or papillose, appressed, more or less uncinate or
patent; these occur in a variable frequency in most species.
In G. schiedeanum and G. lozanoi some hairs have a circular
basal portion and others a narrowly elliptic one. The second
type appears under the microscope as “plane” or “scale-
like”, particularly if it is very short. According to Payne (1978),
all these hairs belong to the “subulate” type. (2) Glandular
Fig. 6. SEM photographs showing indumentum
types found in Geranium. a. Eglandular hairs on
the leaf of G. biuncinatum (sect. Trilopha). b. De-
tail of an eglandular hair on the pedicel of G.
core-core (sect. Geranium). c. Eglandular hooked
hairs on the mericarp of G. lasiopus (sect. Ru-
berta). d. Eglandular hairs on the mericarp of G.
libanoticum (sect. Tuberosa). (Based on: a, Wood
3126, MA; b, Aedo 7073, MA; c, Mutlu 848, HUB;
d, Zederbauer s.n., June 1902, WU).
100 um
100 um
a
10 um
b
100 um
c d
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
19. 19
Morphology
Geranium ornithopodioides has peltate leaves (with the
petiole attached to the blade not by the margin), an un-
known trait in other species of the genus.
The Hawaiian endemic species of sect. Neurophyllodes
have a distinctive leaf lamina. It is unlobed, with an ovate to
obtriangular-cuneiform outline, with 3-7 teeth restricted to
the apex or almost to the base and with a cuneate base. In
the sect. Paramensia, G. jahnii has quite similar leaves, obtri-
angular in outline, cuneate and with a tridentate apex, and
G. exallum differs from all other species in the genus in its
entire, narrowly lanceolate, ericoid lamina. In the sect. Gera-
nium, four species from the northern Andes (G. digitatum, G.
maniculatum, G. rhomboidale, and G. trujillense; the Sibbaldi-
oides Group) plus G. cavanillesii (the Multiceps Group), have
digitate laminas, with the lateral segments turned upward.
Geranium campii (the Sibbaldioides Group), also from this re-
gion, has a very distinctive lamina, which is tripartite with
broadly lanceolate segments. Geranium azorelloides (the
Sibbaldioides Group) has a rather different tridentate lamina
that is quite similar to that of G. jahnii (Figs. 12, 13).
The cauline leaves of Geranium are opposite in most spe-
cies but alternate leaves are also common. In species of dif-
ferentgroupsthemiddlecaulineleaf(usuallyone,sometimes
two) is alternate but the more distal leaves are opposite. In
contrast, in five species of the Papuanum Group the cauline
leaves are opposite while those of the buds alternate. Gera-
nium oaxacanum shows a trait unknown in any other species
of Geranium: a whorl of 3-6 leaves at the first branch point
of the inflorescence. Sometimes a second whorl of leaves is
present between the base and the first branch point.
The six species of sect. Neurophyllodes share a singular
feature: an abscission zone at the base of the petiole, just
at the junction with the stipules. It should be noted that
two species from sect. Paramensia also have leaves with an
abscission zone. In these species, however, the abscission
zone is between the lamina and the petiole.
Fig. 9. Photographs showing palmatifid leaf laminas of Geranium. a.
G. atlanticum (sect. Geranium). b. G. renardii (sect. Lanuginosa). c. G.
thessalum (sect. Subacaulia). d. G. wilfordii (sect. Geranium). (Based
on: a, Quintanar 3244, MA; b, Aedo 4478, MA; c, Herrero al. 3600,
MA; d, Farges 110, MA).
a b
c
d
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
21. 21
Morphology
quire a unique aspect because of the persistent, sheathing
stipules (Fig. 16). The stipules of the South African species of
sect. Incana are lanceolate to flabelliform in outline, with 2-9
narrow and deep lobes (Fig. 16).
Inflorescence
Acaulescent species (e.g., sect. Subacaulia, Sessiliflorum
Group or Sibbaldioides Group), have a reduced inflores-
cence of cymules arising directly from the rootstock. In the
caulescent species the usually 2-flowered cymules are in a
dichasially arranged inflorescence, sometimes with mono-
chasial branches (e.g., sect. Tuberosa, Caespitosum Group,
Multiceps Group, Sylvaticum Group). It is more frequent,
however, that 1- or 2-flowered cymules are borne at the
nodes along the stem, i.e., in a monochasial cyme. In both
cases some species may have umbel-like aggregates of cy-
mules at the top of each branch in which the peduncles are
usually short or absent.
The shrubby species of sect. Neurophyllodes (from Ha-
waii) have a complex inflorescence. Along each branch
some floriferous branchlets can occur. These partial in-
florescences have 1-flowered cymules grouped into
(1)7-12(60)-flowered cymes at the end of a brachyblast.
Some axillary leaves occur on the basal part of these in-
florescences. From each node usually one single-flowered
cymule is borne. All nodes of the inflorescences, i.e., the
base of the peduncle and the partial branchlets have ab-
scission lines. The abscission of peduncles together with
the fruits may contribute to seed dispersal as a comple-
mentary mechanism of autochory, superposed upon the
ballistic seed-ejection system of this group of Geranium.
Some robust species from Macaronesian islands also ex-
hibit a complex inflorescence. Geranium palmatum and G. re-
uteri (sect. Ruberta) produce a terminal inflorescence with ax-
illary branches arising below the uppermost rosette leaves,
which is basically dichasial with 2-flowered cymules. Howev-
er, the suppression of internodes, the absence of growth in
certain axes, and the abortion of flowering peduncles com-
bine to display a complex inflorescence with numerous flow-
ers together. Geranium maderense (sect. Ruberta) has a sim-
ilar but immense inflorescence with four to eight crowded
terminal branches at the apex of the rosette leaves.
In some shrubby species from South Africa (sect. In-
cana) the inflorescence shows peduncles opposite to the
Fig. 12. Photographs showing obtriangular-cuneiform or lanceolate leaf laminas of Geranium. a. G. azorelloides (sect. Geranium). b. G. cuneatum subp.
cuneatum (sect. Neurophyllodes). c. G. jahnii (sect. Paramensia). d. G. exallum (sect. Paramensia). (Based on: a, Cuatrecasas Willard 26343, MA; b, Remy
828, MA; c, Riina al. 597, MA; d, Sklenář 15822, MA).
a b
c d
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
23. 23
Morphology
Fig. 17. Photographs showing whorled, opposite or overlaping bracteoles of Geranium. a. G. nodosum (sect. Geranium). b. G. sanguineum (sect. Ge-
ranium). c. G. cuneatum subp. cuneatum (sect. Neurophyllodes). (Based on: a, Calvo al. 6606, MA; b, Galán Cela 1261, MA; c, Remy 828, MA).
Fig. 15. Photographs showing free or connate, ovate stipules of Gerani-
um. a. G. maderense (sect. Ruberta). b. G. reuteri (sect. Ruberta). c. G. arabi-
cum subsp. arabicum (sect. Geranium). d. G. wallichianum (sect. Geranium).
(Based on: a, Sequeira Góis-Marques 7961, MA; b, Panero Ortega 6942,
MA; c, Aedo 18093, MA; d, Aedo 15739, MA).
a b d
c
Fig. 16. Photographs showing sheathing or flabelliform and deeply lobed stipules of Geranium. a. G. cuneatum subp. cuneatum (sect. Neurophyl-
lodes). b. G. multisectum (sect. Incana). c. G. pulchrum (sect. Incana). (Based on: a, Remy 828, MA; b, Aedo 15092b, MA; c, Aedo 15016, MA).
a b c
a b c
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
25. 25
Morphology
50 um
10 um
a b
10 um
c
50 um
d
10 um
a
30 um
b
10 um
c
5 um
a
5 um
b
50 um
c
50 um
d
50 um
d
Fig. 18. SEM photographs showing pollen grains of the Geranium-type in different groups of Ge-
ranium sect. Geranium. a. G. antrorsum. b. G. holosericeum. c. G. schrenkianum. d. G. tablasense.
(Based on: a, Woodward s.n., 21 Jan. 1958, SYD; b, Uribe 7027, MA; c, Karamuisheva 62, LE; d, Cárde-
nas 758, US).
Fig. 19. SEM photographs showing pollen grains of the Geranium-type in different groups of Ge-
ranium subg. Geranium. a. G. exallum (sect. Paramensia). b, c. G. arboreum (sect. Neurophyllodes).
d. G. sintenisii (sect. Dissecta). (Based on: a, Prieto 279, BH; b, c, Degener 18007, A; d, Nisa 712, MA).
Fig. 20. SEM photographs showing pollen grains of the Geranium-type in Geranium subg. Erodioidea,
subg. Tuberosa and subg. Robertium. a. G. aculeolatum (sect. Aculeolata). b. G. glanduligerum (sect.
Brasiliensia). c. G. malviflorum (sect. Tuberosa). d. G. maderense (sect. Ruberta). (Based on: a, Witte
1840, P; b, Ule 1717, R; c, Retz 84449, H; d, Gonçalves 9688, MADJ).
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
27. 27
Morphology
a b c
Fig. 23. Photographs showing fruits of the “Erodium-type” of discharge in species of Geranium
subg. Erodioidea. a. G. argenteum (sect. Subacaulia). b, c, G. thessalum (sect. Subacaulia). (Based on:
a, Aedo 20891, MA; b, c, Herrero al. 3600, MA).
a
1 mm
c
300 um
d
100 um
b
Fig. 24. SEM and optical photographs showing mericarps of the “Erodium-type” of discharge in
species of Geranium subg. Erodioidea. a, b, G. aculeolatum (sect. Aculeolata). c, d, G. brasiliense
(sect. Brasiliensia). (Based on: a, Jansen 4277, G; b, Lebrun 7308, P; c, d, Lahtivirta s.n., H-1680478).
Fig. 25. Photographs showing fruits of the “carpel-projection type” of discharge in species of Geranium
subg. Robertium. a. G. albanum (sect. Divaricata). b, G. divaricatum (sect. Divaricata). c. G. pusillum (sect.
Batrachioidea). d. G. cataractarum (sect. Ruberta). (Based on: a, Gonzalo 251, MA; b, Herrero 1446, MA;
c, Sunding s.n., MA-774016; d, Fernández Casas s.n., MA-394574).
a
c d
b
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
29. 29
Morphology
In the second subclass the seed is retained in the meri-
carp during the pre-explosive interval by a part of the meri-
carp wall, which projects as a prong. This prong has the
same texture as the rest of the mericarp wall. This type is
found in the sect. Dissecta (subg. Geranium) (Figs. 28, 30).
Yeo (2002a: 39) has suggested that subclasses first and sec-
ond might deserve the rank of subgenus. In the present
work this range has only been accepted for the first group
due to the strong morphological differentiation of its fruits.
In the third subclass the mericarp has a basal callus with
a cluster of bristles at the lower end, which prevents the
seed from falling out prematurely during the pre-explosive
interval. This type is found in the remaining groups classi-
fied into the subg. Geranium (Figs. 30-32).
In the three subclasses of the “seed-ejection type” the
mericarp is smooth, occasionally with 1-3 transverse veins
at the apex.
The rostrum of some species of Geranium is gradually
narrowed toward the apex, whereas in others it terminates
abruptly in a narrow apex, in which case the apical portion of
the rostrum is easily differentiated from the basal one. This
narrowed apex is of taxonomic utility at the species level.
Structural aspects of the hygroscopic movements of the
awninGeraniaceaeareanalyzedbyAbrahamElbaum(2013).
Fig. 28. Photographs showing fruits of the “seed-ejection type” of discharge in species of Geranium subg. Tuberosa and sect. Dissecta. a. G. platypeta-
lum (sect. Lanuginosa). b. G. lanuginosum (sect. Lanuginosa). c. G. dissectum (sect. Dissecta). (Based on: a, Aedo 11454, MA; b, Aedo 24060, MA; c, without
voucher, from Madeira, Portugal, photo: M. Sequeira).
Fig. 29. SEM photographs showing mericarps of the “seed-ejection type” of discharge in species
of Geranium subg. Tuberosa. a. G. bohemicum (sect. Lanuginosa). b. G. ibericum (sect. Lanuginosa).
c. G. kotschyi subsp. charlesii (sect. Tuberosa). d. G. malviflorum (sect. Tuberosa). (Based on: a, López
s.n., MA-213282; b, unknown collector, BR-825853; c, Lipsky 3432, LE; d, Davis 59451, BM).
2 mm
1 mm
1 mm 1 mm
a
a
d
c
c
b
b
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
31. 31
Morphology
Seeds
The seeds are more or less ellipsoid and slightly com-
pressed. Most species of the genus show a finely reticulate
surface. The depth and length of the alveolus vary within
some species, suggesting that this feature has little taxo-
nomic value. In many species of subg. Robertium (but also
in few cases of those of the subg. Geranium and subg. Tu-
berosa) the surface appears smooth at low magnification,
but a reticulum is revealed at high magnification (Fig. 33).
Exceptionally, the seeds of G. rotundifolium and few oth-
er species show a distinct reticulation even to the unaided
eye (Fig. 34). The outer layer of the seed-coat has cells with
anticlinally thickened walls, forming a polygonal pattern.
These anticlinally thickened walls, usually straight in the
genus, are more or less curved in the seeds of the sect.
Neurophyllodes, giving the seed-coat surface a distinctive
reticulate-sinuous aspect. Seeds are glabrous in Geranium,
except in G. rupicola, which has scattered hairs on the seed
surface (Fig. 35).
According to Albers Van der Walt (2007) the seeds are
campylotropous with large embryos and strongly folded cot-
yledons, when viewed in transverse section, and have no or
very little endosperm; they are rich in lipid substances and
poor in starch (see also Boesewinkel Been 1979; Boesewin-
kel 1988, 1997; Yeo 1990). Cotyledons usually have an entire
margin but in some species they have one or two notches on
each side. This feature has been found only in G. aculeolatum
(with two notches on each side) (subg. Erodioidea), G. albanum
(subg. Robertium), G. aristatum (subg. Erodioidea), G. bohemic-
um (subg. Tuberosa), and G. divaricatum (subg. Robertium).
Specific studies about Geranium seeds production, devel-
opment, dormancy, germination and seed survival are ad-
dressed in the following papers: Abrams Dickmann (1984),
Ågren Willson (1992), Baskin Baskin (1974), Gama-Arach-
chige al. (2010, 2011, 2012), Herrera (1991), Meisert (2002),
Meisert al. (1999), Milberg (1994), Raunkiaer (1888), Rob-
erts Boddrell (1985), Van Assche Vandelook (2006), Van-
delook Van Assche (2010) and Varga al. (2015).
Fig. 33. SEM photographs showing seeds and seed-coat details of Geranium. a, b. G. holosericeum.
c, d. G. maderense. (Based on: a, b, Fernández Alonso 7742, MA; c, d, Navarro 3396, MA).
Fig. 34. SEM photographs showing seeds and seed-coat details of Geranium. a. G. dissectum. b.
G. grande. c, d. G. rotundifolium. (Based on: a, Quintanar 3888, MA; b, Cunninghan 30, MEL; c, d,
Herrero 1306, MA).
a
a
d
d
c
c
b
b
1 mm
1 mm
1 mm
100 um
1 mm
50 um
300 um
1 mm
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
33. 33
Breeding System
sic chromosome number of G. lanuginosum is probably x
= 14, and its number 2n = 48 may have originated from a
dysploid. Only five more species of this section have been
counted: G. gracile with 2n = 26, G. gymnocaulon, G. iberi-
cum, G. libani and G. platypetalum with 2n = 28 in all cases,
but with some additional numbers to be confirmed in G.
ibericum (2n = 56), and G. platypetalum (2n =42).
There is uneven information regarding the six sections
of the subg. Robertium. The four species of sect. Batrachi-
oidea have 2n = 26. It seems that some different numbers
attributed to these species should be disregarded (Van
Loon 1984b, 1984c). In the sect. Divaricata, G. divaricatum
has 2n = 28 while G. albanum has 2n = 20; in both species
there are old counts which are difficult to verify (Van Loon
1984b, 1984c). The sect. Trilopha and sect. Polyantha are
poorly studied. In the former only two species were count-
ed, G. brevipes with 2n = 56 and G. favosum with 2n = 50. In
the latter only G. polyanthes was studied resulting in 2n =
28. In sect. Unguiculata, 2n = 46 was found in G. dalmaticum,
while G. macrorhizum showed a variable number of chro-
mosomes, in some cases 2n = 46 and in others from 2n =
87 to 93 (see Van Loon 1984b: 269 for further details). All
species of sect. Ruberta were studied. The figures obtained
in this group are quite heterogeneous. According to Van
Loon (1984c: 283) G. lucidum has a polyploid series (2n =
20 to 44) with the basic chromosome number x = 10. The
tetraploid cytotype (2n = 40) is the most common and it
comprises an ascending dysploid series. According to Yeo
(1973a: 314) and Wilder-Kiefer Yeo (1987), G. purpureum
(2n = 32), G. robertianum (2n = 64) and G. yeoi (2n = 128)
probably represent a polyploid series. This author indi-
cates that the triploid hybrid G. purpureum × G. robertianum
formed n univalents and n bivalents at the first division of
meiosis, strongly suggesting that G. robertianum is an allo-
polyploid with G. purpureum and an unknown species as
parents. Yeo Wilder-Kiefer (1990: 2) pointed out that the
results of their study on the presence/absence of amino ac-
ids are congruent with the hypothesis of the allopolyploid
origin of G. robertianum and the autopolyploid origin of G.
yeoi [as G. rubescens]. The remaining species of the section
are: G. glaberrimum (2n = 30), G. cataractarum (2n = 36), G.
lasiopus (2n = 48), G. maderense (2n = 68), G. palmatum (2n
= 68) and G. reuteri (2n = 128). Wilder-Kiefer Yeo (1987:
287) reported that the hybrid of Geranium maderense × G.
cataractarum formed ca. 16 univalents and ca. 18 bivalents
at the first division of meiosis, which strongly suggests an
allopolyploid origin for G. maderense, with G. cataractarum
and an unknown species as parents.
According to Van Loon (1984b, 1984c) most species of
Geranium are diploid with 2n = 28 and the main basic chro-
mosome number is x = 14 (Van Loon 1984b, 1984c; Warburg
1938a, 1938b), but other numbers (x = 9 to x = 23), mainly in
annual species, have also been reported (Van Loon 1984b,
1984c). These data are probably due to a sampling focused
on Eurasian species and could change when more species
from Africa, America and Oceania can be studied. Polyploi-
dy appears to be an important pathway far evolution in
some groups, like sect. Ruberta (Yeo 1973a: 308). It may be
interesting to note that in some species from different sub-
genera (e.g.: G. asphodeloides, G. caeruleatum, G. lucidum, G.
macrorrhizum, G. macrostylum, G. potentilloides, G. reflexum,
G. subcaulescens) a variable number of chromosomes have
been found. These cytotypes do not differ from each other
morphologically.
BREEDING SYSTEM
It is widely accepted that the species of the genus Geranium
are predominantly protandrous with an internal whorl of
stamens that matures earlier than the external one, and a
stigma which matures after pollen dispersal, which favours
decreasing selfing processes (Aedo 1996; Knuth 1908;
Sprengel 1793; Wildler-Kiefer Yeo 1987). Several authors,
however, have found different variations on this pattern.
Philipp (1985) found that plants of G. brevicaule [as G. ses-
siliflorum] were protandrous at one locality, homogamous
at another, and protogynous at a third locality. Martin
(1965: 127) reported that G. maculatum is predominantly
protandrous but also may be partially homogamous. Tofts
(2004a: 549) summarized the available information about
G. robertianum, that may be homogamous, slightly protan-
drous or protogynous. Protogyny is also reported by Bertin
(2001) for G. robertianum, and by Knuth (1908) for G. dissec-
tum and G. pusillum.
Although protandry (or protogyny) suggests a strong
preference for outcrossing, in plants with a large number
of simultaneously blooming flowers, pollination can occur
between different flowers of the same individual. Such is
the case reported by Hessing (1986: 467) in protandrous
plants of G. caespitosum, where “geitonogamous pollina-
tion is an extremely common event, probably overwhelm-
ing the effects of dichogamy”. Moreover, Hessing (1988:
1328) suggested that the low decrease of fertility after self-
ing of G. caespitosum is not caused by self-incompatibility.
Asikainen Mutikainen (2005b: 488) also pointed out that
hermaphrodite flowers of G. sylvaticum are self-compati-
ble. Wildler-Kiefer Yeo (1987: 288) indicated that flowers
in Geranium are fully self-compatible.
Cruden (1976, 1977) emphasized that the Log pol-
len-ovule ratios are correlated with the breeding system. A
substantial decrease in Log P/O suggests a pass from out-
cross pollination to self-pollination. Aedo al. (2002) point-
ed out that some perennial species from the Andes (such
as G. multipartitum and G. sibbaldioides) show low Log P/O
ratios, suggesting facultative self-pollination. This is con-
gruent with homogamy observed in herbarium specimens
of these perennial species.
Shirk Hamrick (2014) found that G. carolinianum, an
annual species with short petals, is incompletely protogy-
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es
35. 35
Distribution and habitats
The compilation of the preceding paragraphs is neces-
sarily incomplete because only the taxonomic literature
could be examined, but it seems useful to provide a gener-
al overview about this topic. As Stuessy al. (2014) point-
ed out phytochemical data have been used successfully in
systematics but in recent times they have flagged consid-
erably. Regarding Geranium, it seems that most the studies
have been focused on flavonoids. Their results are obvi-
ously interesting to understand the species that have been
studied, but at the moment they seem of little taxonomic
utility, either to characterize the infrarageneric groups or
the species.
DISTRIBUTION AND HABITATS
The genus Geranium is distributed throughout most of the
world except in lowland tropical areas (Fig. 5). In the New
World most of the 123 native species belong to sect. Gerani-
um. The exceptions are the two species of sect. Paramensia
(subg. Geranium) growing on the northern Andes and the
three of sect. Brasiliensia (subg. Erodioidea) found in south-
ern Brazil and northern Argentina. Only two species of sect.
Geranium are native in the New World as well as elsewhere:
G. erianthum also occurs in eastern Asia, and G. sylvaticum
is widespread in Europe and western Asia. The Andes in
Colombia and Ecuador harbor many endemic species but
the greatest diversity is found in Peru (34 species) and in
Mexico (36 species).
The Pacific area that includes the islands of Japan, Tai-
wan, Indonesia, East Timor, Papua New Guinea, Australia,
New Zealand, and Hawaii harbor 42 native species, all from
sect. Geranium except the Hawaiian species of the sect. Neu-
rophyllodes (subg. Geranium). The greatest diversity is found
in New Guinea and Japan, both with 11 species; the species
of New Guinea are all endemic, while the diversity of Ge-
ranium in Japan includes three endemic species and eight
species shared with mainland Asia, one extending to North
America. Other areas of high diversity of Geranium in the Pa-
cific area are Australia and New Zealand (nine species, all en-
demic), and the Hawaiian Islands (six species, all endemic).
In Sub-Saharan Africa the genus is generally poorly rep-
resented. The richest area is South Africa where most of
the 22 species of sect. Incana are endemic, some of them
reaching the mountains of southeastern Africa. The only
representative of sect. Aculeolata (subg. Erodioidea) is
spread across the highlands of eastern Africa from Ethiopia
to Malawi. Most of the seven species of the sect. Trilopha
(subg. Robertium) are found in the Horn of Africa area and
the nearby Arabian peninsula, but one reaches the eastern
and western mountains of tropical Africa and another is
endemic to the Himalayas. The sect. Ruberta (subg. Rober-
tium) is represented in eastern mountains of tropical Afri-
ca by one non-endemic species, and the sect. Geranium by
three species, one endemic from Madagascar.
The richest area both in number of species and in diver-
sity of sections and subgenera is the region that includes
the mountains around the Mediterranean Basin, the Cau-
casus, central Asia, the Himalayas and China. In this area
the six sections of the subg. Robertium are found, the two
sections of the subg. Tuberosa, two of the four sections
of the subg. Erodioidea and two of the five sections of the
subg. Geranium.
The 15 species of sect. Subacaulia (subg. Erodioidea) and
the three species of the sect. Erodioidea) are found in the
mountains around the Mediterranean Basin, although a
species of the latter group reaches northern Europe.
The sect. Tuberosa (seven species), found from the west-
ern Mediterranean to the highlands of central Asia and the
western Himalayas, is better represented in the central
area of this region. The sect. Lanuginosa (subg. Tuberosa; 10
species) is better represented in the Caucasus and Iranian
mountains, but a couple of species reach the western Med-
iterranean and northern Europe.
Regarding the subg. Robertium, it has already been men-
tioned that sect. Trilopha is focused on the Horn of Africa
area and the nearby Arabian peninsula, and in southern
Iran (five species) plus one species endemic to the Himala-
yas. The sect. Polyantha (seven species) is found through-
out the Himalayas reaching Burma and the mountains of
western China. One of the two species of sect. Divaricata
is restricted to the Caucasus and northern Iran while the
other is widespread, occurring from western and central
Europe to central Asia and the Himalayas. The sect. Batra-
chioidea (four species) is also widespread, occurring from
western and central Europe to central Asia and the Himala-
yas. The sect. Ruberta has four species endemic from Mac-
aronesian islands, three endemic from the Mediterranean
Basin and the remaining three widespread throughout the
Mediterranean Basin, central Europe, Asia and Africa. The
native status of G. robertianum in northeastern U.S.A. and
Japan is controversial (Aedo 2012; Nishida al. 2012). Fi-
nally, the sect. Unguiculata has two species restricted to the
north side of the Mediterranean Basin.
The sect. Geranium is represented in Europe and main-
land Asia by 53 native species. This section comprises three
species widely distributed through Europe and Asia. Be-
sides, six species are restricted to the mountains of south-
ern Europe and northern Africa and seven more occur also
in lowlands of central Europe and extend to Caucasus and
West or central Asia. The greatest diversity of section Ge-
ranium is found in southern China and adjacent areas of
the Himalayas with 22 species, and in eastern Siberia and
adjacent regions of northeastern China and Korean penin-
sula with 10 species. Four species extend throughout Sibe-
ria, Mongolia and nearby areas and one is restricted to the
mountains of south India. The sect. Dissecta (four species)
is better represented in the eastern Mediterranean Ba-
sin, although one species has spread to Macaronesia and
northern Europe.
Copia
gratuita
/
Personal
free
copy
http://libros.csic.es