WORLD CHECKLIST OF GERANIUM L. (GERANIACEAE)

AEDO, C , F. MUÑOZ GARMENDIA & F. PANDO (1998). World checklist of Geranium L. (Geraniaceae). Anales Jard. Bot. Madrid 56(2): 211-252. A checklist of the genus Geranium L. (Geraniaceae) is presented. Four hundred and twenty three species are recognized in 3 subgenera and 18 sections. Our classification differs from Yeo's only in some aspects of subg. Erodioidea and Geranium. Section Brasiliensia is included in subg. Erodioidea, and sect. Neurophyllodes, Paramensia and Azorelloida in subg. Geranium. Section Azorelloida is proposed as an avowed substitute (nom. nov.) for sect. Petraea R. Knuth, nom. illeg. G. collae is proposed as avowed substitute (nom. nov.) for G. intermedium Colla, nom. illeg. An identification key to subgenera and sections is presented. A thorough revisión of available ñames in the genus, and a review of their nomenclatural status were carried out. Correct ñame, place of publication and distribution are given for each species. Geographical distríbutions are given at región and botanical country levéis following the International Working Group on Taxonomic Databases (TDWG) standard. When possible, selected references with relevant information are also included.


iNTRODUCnON
After  published his monograph on Geranium, several regional accounts as well as sotne detailed studies on some sections were published. These papers vastly increased the number of known species. KNUTH (1912) recognized ca. 260 species, and over 423 are currenüy accepted. Information regarding the genus is nowadays scattered over numerous papers, which makes it difficult to grasp composition, circumscription, or any other particular of a given group within Geranium, or even of any species. The exception is CLIFTON'S (1992) worldwide compilation; however, the shallow treatment of nomenclatural and distribution data limits its use.
The aim of this paper is to make accessible the most relevant information about Geranium produced in the last eighty years in a concise way. We mostly follow the infrageneric scheme proposed by , but introduce some changes.
This compilation is primarily based on literature, but also numerous herbarium specimes were reviewed in order to clarify the systematic position of some problematic taxa. Thus, the checklist is the result of collating, interpreting and harmonizing the alternative taxonomies found in the literature and our own work with Geranium (see acknowledgment) ( KNUTH'S ( , 1931 scheme -32 sections for the genus-has been questioned by numerous authors (WARBURG, 1938a(WARBURG, , 1938b, though without advancing an alternative until YEO'S (1984) review. Yeo first subdivides the genus into three subgenera on the basis of fruit-discharge mechanism, as PICARD (1837) already proposed, without acceptance at the time.  distinguishes three principal types of fruit discharge, each characterizing one of the three subgenera. In the "seedejection-type" (which characterizes subg. Geranium) a single seed is actively discharged by the explosive recurvature of the awn, which remains together with the mericarp attached to the columella. The second type of discharge or "carpelprojection-type" characterizes subg. Robertium. Here, the explosive recurvature of the awn also acts as the propelling forcé, but in this case the whole mericarp, containing the seed, is dispersed, whereas the awn remains with the columella. Subgenus Erodioidea is identified by the "Erodium-type" discharge. In this case the mericarp, including the coiled awn, is propelled over a short distance. We follow the scheme proposed by YEO ( , 1990, except for a few changes that we discuss below (Table 1). We also provide an identification key to subgenera and sections, which includes the proposed new sections.
At first, YEO (1984) distinguished two sections in subg. Erodioidea: sect. Erodioidea and sect. Subacaulia, and later YEO (1990) added sect. Aculeolata; all three characterized by the presence of twoflowered cymules. This view, accepted in a recent revisión of the subgenus (AEDO, 1996), is modified with the addition of sect. Brasiliensis, which has one-flowered cymules. We examined specimens belonging to two of the three species of this section (G. arachnoideum and G. brasiliense), and confirmed that both species have "Erodium-discharge-type" fruits, the defining feature of the subgenus Erodioidea. However, this type of discharge,  . 3). Geranium sect. Geranium is a heterogeneous group that comprises over 339 species, which requires a profound revisión. We propose here segregation from it of three small sections on the basis of leaf shape and veining: sect. Paramensia, sect. Neurophyllodes and sect. Azorelloida.
Section Paramensia comprises two very peculiar species, endemic to Northern Andes ( fig. 3). Both species are low shrubs with narrow, coriaceous, glabrous, parallel-veined leaves (linear in G. exallum; cunéate and shallowly 3-lobed at the apex in G. jahnii), have the petiole articúlate with the blade, cymules with one flower and a very distinctive "seed-ejection-type".
Section Neurophyllodes includes 10 taxa, all endemic to Hawaii ( fig. 3), and shares with the previous section a shrubby life-form (up to 2.5 m high), leaf veining and petiole articulation position. However, the leaves are much larger, usually elliptic to ovate (though obovate or cunéate in some species), the petiole is articulated with the stipules, and the cymules are two-flowered. This section is also firmly supported by data from chloroplast DNA . The similarities between the two sections suggest a common origin (STANDLEY, 1915), a possibility that should be explored further, especially the implications of petiole articulation, which is obviously derived and less likely to be homoplasic than leaf veining.
Section Azorelloida contains only one species, known from the Columbian Andes ( fig. 3). It is also a low shrub with oneflowered cymules and leaves very similar to those of G. jahnii: coriaceous, glabrous, cunéate and 3-lobed at the apex, though the petiole is not articúlate. This is just a small contribution towards the clarification of the taxonomy of the subg. Geranium, where more sections need to be recognized.  already pointed out affinities among many species and identified several groups, mostly in Europe, Asia and N America. Obviously, any clarification of the subgenus should take into account the sections proposed by KNUTH ( ,1931, some of which could fit into the YEO'S ( ,1990 scheme. We follow the treatment proposed by YEO ( ,1990 for subg. Robertium. According to his scheme, the subgenus comprises 8 sections and 30 species. Section Polyantha (7 species) is endemic to Eastern Himalayas and Southern China. Section Anemonifolia (2 species) also show a limited distribution, being endemic to Madeira Island. Section Trilopha (5 species) is restricted to mountains of Tropical África, Western Asia and Eastern Himalayas. Distribution of the remaining five sections (Lucida, Ruberta, Divaricata, Batrachioidea and Unguiculata) is focused on the Mediterranean área and Western Asia, though sect. Ruberta extends in the east to Japan, and in the south to mountains of Tropical África (figs. 4,5). This subgenus, morphologically well-defined, is also firmly supported by data from chloroplast DNA (PRICE & PALMER, 1993).

ARRANGEMENT OF THE CHECKLIST
Subgenera and sections are arranged according to the scheme by YEO ( ,1990. New sections are listed at the end of their respective subgenera. Within sections, species are in alphabetic order. For each accepted species, place of publication, selected references, and geographic distribution are provided. We have examined the original publication for all accepted ñames and basionyms and checked the nomenclatural status of each one. We have verified all synonyms in the current literature (ca. 2600 ñames) as well, but they are not listed here for the sake of brevity. The checklist has been generated from a datábase with a computer program developed by us, NOMENFMI, which sorts, formats and ensures nomenclatural consistency of the listed ñames. The list of references (1300) used for the compilation of the checklist would be too large to be included here in full, and only selected references for each species -where descriptions, identification keys, illustrations, nomenclature or distributional data on each one can be found-is provided. These references are listed after the species ñame by an identifying number in square brackets. For some species no references, other than that for the protologue, are given. This is due either to their recent publication or to their remote distribution, as is the case with most species from South America.
Geographic distribution is recorded following the standard by HOLLIS & BRUMMTTT (1992). We use level 2 (regions, grouped by continents) and level 3 (botanical countries) (see Appendix 1,2). In a few cases, a more detailed distribution is provided using level 4 (basic recording units), or listing the specific localities between round brackets. Introduction in an área is indicated by enclosing its code in square brackets; extinction by appending the symbol f at the end the code; when presence is questionable in an área, a question mark follows its code.
Finally, we believe that others in the systematic community will find the checklist useful. We hope that it will assist those currently conducting research on Geranium and encourage others to undertake badly needed revisions of certain infrageneric groups. KEY TCS  TEN  TEX  THA  TKM  TOE  TUN  TUR  TVA  TVL  TZK