Otto Warburg and his contributions to the screw pine family (Pandanaceae)

Abstract: Otto Warburg (1859–1938) had a great interest in tropical botany. He travelled in South-East Asia and the South Pacific between 1885 and 1889 and brought back a considerable collection of plant specimens from this expedition later donated to the Royal Botanical Museum in Berlin. Warburg published the first comprehensive monograph on the family Pandanaceae in 1900 in the third issue of Das Pflanzenreich established and edited by Adolf Engler (1844–1930). The aim of this article is to clarify the taxonomy, nomenclature and typification of Warburg's contributions to the Pandanaceae. Considerable parts of Warburg's original material was destroyed in Berlin during World War II but duplicates survived, shared by Engler and Warburg with Ugolino Martelli (1860–1934). Martelli was an expert on the family and he assembled a precious herbarium of Pandanaceae that was later donated to the Museo di Storia Naturale dell'Università degli Studi di Firenze. Warburg published 86 new names in Pandanaceae between 1898 and 1909 (five new sections, 69 new species, five new varieties, two new combinations and five replacement names). A complete review of the material extant in B and FI led to the conclusion that 38 names needed a nomenclatural act: 34 lectotypes, three neotypes and one epitype are designated here. Twenty new synonyms are also proposed. One Freycinetia name and six Pandanus names are considered as incertae sedis. A total of 21 names published by Warburg are accepted: 11 in Freycinetia and ten in Pandanus. In addition, four names published in Pandanus by Warburg serve as the basionyms of accepted names in the genus Benstonea. Citation: Callmander M. W., Vogt R., Donatelli A., Buerki S. & Nepi C. 2021: Otto Warburg and his contributions to the screw pine family (Pandanaceae). – Willdenowia 51: 5–31. Version of record first published online on 15 February 2021 ahead of inclusion in April 2021 issue.


Introduction
was born on 20 July 1859 in Hamburg (Germany). He studied Natural Sciences and Philosophy at the Universities of Bonn, Berlin and finally Strasbourg, where he completed in 1883 a PhD under the guidance of the renowned Heinrich Anton de Bary (1831 -1888) with a wood morphological dissertation "Über Bau und Entwicklung des Holzes von Caulotretus heterophyllus" (Warburg 1883). In the years 1884 to 1885, postdoctoral studies in München and Tübingen followed where Warburg pursued chemical and plant physiological issues with Adolph von Baeyer (1835-1917) and Wilhelm Pfeffer (1845-1920 respectively (Leimkugel 2005). Influenced by Adolf Engler's (1844Engler's ( -1930 research on plant geography and enthusiastic about the theories and work of the British naturalists Charles Darwin (1809 -1882) and Alfred Russel Wallace (1823Wallace ( -1913, Warburg began to focus more and more on issues of plant evolution and plant geography. He was specially fascinated by Wallace's book Island Life (Wallace 1880) and it was the theory of the so-called Wallace's line, a hypothetical line that separates the Malay archipelago into two distinct parts with different -Indian and eastern Australian -animal and plant origins (Wallace 1880), which inspired him on a four-year research trip to South-East Asia. Otto Warburg came from an extremely wealthy family, was financially independent by inheritance and 6 Callmander & al.: Otto Warburg and his contributions to the Pandanaceae financed the trip from his own assets (Leimkugel 2005). He started in Bombay in December 1885, visited Peninsular Malaysia, Java, Singapore, China, Korea, Japan, the Philippines, the Moluccas islands, German New Guinea and the Bismarck archipelago (now the northern region of Papua New Guinea), and ended up in Australia in June 1889 (see Warburg 1900a). From this four-year expedition, Warburg brought back a collection of more than 25 000 plant specimens (c. 22 000 phanerogams, mosses and fungi and several thousands of wood, fruit and seed samples as well as material conserved in alcohol) later donated to the Royal Botanical Museum in Berlin (Urban 1916).
Warburg settled down in Berlin and worked intensively on his collections, and in a paper on phytogeographic considerations and the rejection of the theory of Wallace's line he included a first, annotated list of the spermatophytes that he gathered in South-East Asia (Warburg 1891). Some years later, he published the first volume of Monsunia: Beiträge zur Kenntniss der Vegetation des süd-und ostasiatischen Monsungebietes (Warburg 1900a) summarizing part of the scientific results of his four-year expedition and including fungi, algae, bryophytes, ferns and lycophytes, conifers, Cycadaceae and Gnetaceae.
As he was now well known as an expert in tropical flora, Warburg was invited to take over the treatments for several plant families in the handbook Die natürlichen Pflanzenfamilien established and edited by Adolf Engler (with Karl Anton Eugen Prantl [1849-1893 until his early death), and in the year 1900 he published his monograph of the Pandanaceae in Das Pflanzenreich (Warburg 1900b). Das Pflanzenreich was established by Engler to furnish a full and comparative account of all known species by plant families (Anonymous 1902). Warburg's interest in the screw pine family started in 1898 when he treated the family in Franz Reinecke's (1866 -unknown) Die Flora der Samoa-Inseln including first descriptions of new species in Freycinetia Gaudich. and Pandanus Parkinson (Warburg 1898). By 1909, Warburg had named 79 taxa of Pandanaceae, most of those described in Das Pflanzenreich (62 spp., Warburg 1900b), the remaining in treatments for Africa (three spp., Warburg 1904a;Volkens 1909aVolkens , 1909b, the Philippines (one sp., Warburg 1904b), the German colonies in the Pacific area (four spp., Warburg 1905), and New Caledonia (five spp., Warburg 1906). Warburg's (1900b) Pandanaceae treatment is the first comprehensive monograph on the family. In this work, he published 19 new names in Freycinetia (two new sections and 17 new species) and introduced 50 new names in Pandanus (three new sections, 35 new species, five new varieties, two new combinations and five replacement names, including several validations of pre-Linnaean or previously not validly published names). The five new sections were: Freycinetia sect. Oligostigma Warb., F. sect. Pleio stigma Warb., Pandanus sect. Foullioya Warb., P. sect. Sussea Warb. and P. sect. Vinsonia Warb.
At the time of Warburg's (1900b) monograph, about 180 Pandanaceae species were known (Stone 1976), among which 79 (c. 43%) were newly named by Warburg. Today the family includes c. 750 species (Pandanaceae Project 2020). Warburg (1900b) was the first to present a coherent infrageneric classification at sectional level for both Freycinetia and Pandanus including identification keys at this taxonomic level. This infrageneric classification started to receive more attention in the mid-20 th century and was subsequently improved and developed by several taxonomists, e.g. St. John (1960) and Stone (1974). The relatively low number of accepted Pandanaceae names introduced by Warburg (24 out of 79; 30 %) is certainly because he "apparently seldom if ever used specimens from other herbaria on loan for study" and this was "a major drawback in his work" (Stone 1973: 267). Most of the material on which Warburg based his studies originated from German New Guinea, from which he had access, in addition to his own gatherings, to further material in the Berlin Botanical Museum collected by botanists who joined the expeditions to this area, e.g. Udo Max Hollrung (1858 -1937) and Carl Adolf Georg Lauterbach (1864 -1937). He also extensively used available literature in validating several Pandanus names. Examples can be found in the validation by Warburg of the numerous not validly published names of Gaudichaud-Beaupré (1841) in his Botanical Atlas of the Voyage autour du monde exécuté pendant les années 1836 et 1837 sur la corvette la Bonite commandée par M. Vaillant. Warburg was aware only of Gaudichaud's illustrations and overlooked the Gaudichaud collections kept mainly in P (see, e.g., under P. delessertii Warb.).
Warburg's interest shifted gradually to plants of economic importance, in particular those of relevance for the colonies of the Deutsches Reich and he acted for many years as a private lecturer for tropical agriculture at the "Seminar für Orientalische Sprachen" [Seminar for Oriental Languages] at the Royal Friedrich-Wilhelms-Uni-versität and as a member of the Kolonialwirtschaftliches Komitee [Colonial Economic Committee] in Berlin. He wrote a highly regarded monograph on the nutmeg family, i.e. the Myristicaceae (Warburg 1897), for which he was awarded the "Prix de Candolle" in Geneva. Warburg founded, and edited for 25 years, Der Tropenpflanzer, a journal dedicated to tropical agriculture; he published his three-volume standard work Die Pflanzenwelt (Warburg 1913(Warburg -1922 and was increasingly engaged in the World Zionist Congress acting as its president between 1911 and 1920 (Leimkugel 2005;Reichert 1938). Otto Warburg died on 10 January 1938 in Berlin.
In the early 20 th century, Ugolino Martelli (1860Martelli ( -1934, an Italian botanist, became interested in the family Pandanaceae. Martelli was the student of the notable Italian naturalist Odoardo Beccari (1843Beccari ( -1920, who introduced him to the systematics of Arecaceae and Pandanaceae (Moggi 2009). Beccari travelled extensively in South-East Asia (Beccari 1877(Beccari -1889(Beccari , 1902) and be-Willdenowia 51 -2021 queathed his impressive Herbarium palmarum to Martelli in 1920(Cuccuini & Nepi 1999. Martelli also assembled a precious herbarium of Pandanaceae that was later donated along with Beccari's Herbarium palmarum to the Museo di Storia Naturale dell'Università degli Studi di Firenze (Italy) following his will (Moggi 2009). Martelli wrote his first observations on the genus Pandanus in 1902 in the Bullettino della Società Botanica Italiana (Martelli 1902; see Moggi & al. 2009) and described his first new Pandanus species in the same journal (Martelli 1904). Later, he published most of his new taxa in Webbia, a journal that he founded in 1905 on the occasion of the 50 th anniversary of the death of Philip Barker Webb (1793 -1854) (Martelli 1905(Martelli , 1907(Martelli , 1910a(Martelli , 1910b

Results
Among the 81 Pandanaceae names linked to Warburg, 38 names needed a nomenclatural act: 34 lectotypes, three neotypes and one epitype are designated here. Twenty new synonyms are also proposed. One Freycinetia name and six Pandanus names are considered as incertae sedis. A total of 21 names published by Warburg are accepted: 11 in Freycinetia and ten in Pandanus. In addition, four names published in Pandanus by Warburg serve as the basionyms of accepted names in the genus Benstonea Callm. & Buerki (Callmander & al. 2012).
In his Botanical Atlas of the Voyage autour du monde, Gaudichaud-Beaupré (1841) introduced a number of new generic and species names that were validated by an il-lustration with analysis. A monotypic genus and species can be validated simultaneously by a single illustration with analysis (Turland & al. 2018: Art. 38.5 -38.9), e.g. Roussinia Gaudich. and R. indica Gaudich. on t. 21, fig.  1 -4 of the Atlas. However, the new genus must be monotypic. On other plates in the Atlas, where Gaudichaud introduced a new generic name and illustrated more than one species in that genus, none of the names at either rank was validly published. These generic names are "Barrotia" (t. 13), "Dorystigma" (t. 13, 31), "Eydouxia" (t. 18), "Fisquetia" (t. 4, 5), "Foullioya" (t. 26), "Sussea" (t. 24,25,38) and "Vinsonia" (t. 17,23,31). For the same reason, these generic and species names were not validated by the later-published accompanying text (Alleizette 1866: 113 -134). Kurz (1869) and Balfour (1878)  Remarks -A single specimen of original material collected by Theodor Pentzke is extant in B. This specimen is considered here as the holotype. Three duplicates are extant in MEL, but none in FI.
According to Stone (1982), Freycinetia australiensis is a synonym of the New Guinean F. marginata. Remarks -A single specimen of original material is extant in B. This specimen is considered here as the holotype. A photograph of the holotype and a fragment packet containing part of an infructescence are extant in FI.

Freycinetia biroi
The holotype has a determinavit from Stone (18 Jul 1975) as Freycinetia marantifolia. This synonymy is accepted here. Many forms of F. marantifolia have been described as new species. Remarks -A single specimen of original material collected by Warburg is extant in B. This specimen is considered here as the holotype. A fragment packet containing fragments of infructescence is extant in FI. Martelli's handwriting on the packet indicates that it was sent by Warburg to Martelli. Freycinetia candeliformis is considered as a synonym of F. celebica (Stone 1969a Remarks -A single specimen of original material of Freycinetia coriacea is extant in B. This specimen is considered here as the holotype. A duplicate in FI consists of a leaf, a packet containing a fragment of an infructescence and a photograph of the holotype in B. Freycinetia spectabilis is a distinctive species in New Caledonia characterized by its coriaceous leaves with straight, terminal infructescences on thick peduncles ( Fig. 2A). This species shows variations in leaf width and size of syncarps, which is common in Freycinetia. No significant morphological character was found to keep F. coriacea, F. cylindracea and F. subulata distinct, and they are placed here in synonymy with F. spectabilis. Freycinetia cylindracea and F. spectabilis described by Hermann Graf zu Solms-Laubach (1842-1915  Remarks -Original material of Freycinetia ferox has been located in B. This material consists of bracts and staminate inflorescences mounted on two sheets and in the spirit collection. The leaf mounted on B 10 0367708 belongs to the genus Pandanus and has to be excluded. The two sheets and spirit material comprise a single specimen and are considered here as the holotype. Stone (18 Jul 1975) annotated B 10 0367708 with "Warburg's description of the leaf is to be ignored". Earlier, Stone (1969b) considered Freycinetia ferox as having leaves gradually attenuate to the apex, referring to Warburg's description, which "is very different from the appearance of F. minahassae, whose leaves are quite abruptly acute-acuminate" (Stone 1969b: 605) and he did not know "what F. ferox is". After visiting B in 1975 and studying the type specimen, Stone (1982) considered F. ferox as having very abruptly caudate leaf apices and considered F. maxima Merr., also from Luzon, as a synonym. Freycinetia maxima was described by Merrill  Remarks -Freycinetia globiceps was described based on two gatherings: Lauterbach 2119 and Hollrung 857, both cited in the protologue and therefore syntypes. A single specimen of both gatherings is present in B. The most complete specimen is that of Lauterbach, which is therefore designated here as the lectotype. Both type gatherings have duplicates in FI.
Freycinetia globiceps is considered as a synonym of F. beccarii following Keim (2009 Remarks -Warburg provided a replacement name for Freycinetia beccarii Hemsl., which is a later homonym of F. beccarii Solms. Remarks -Freycinetia hollrungii was described based on three gatherings: Hollrung 218, Lauterbach 1522 and Warburg 20997, all cited in the protologue and therefore syntypes. Specimens of the three gatherings are extant in B. The most complete specimen is Hollrung 218, which is therefore designated here as the lectotype. The two sheets in B comprise a single specimen, which is designated here as the lectotype. A duplicate has been located in FI. Freycinetia hollrungii is considered as a synonym of F. scandens (Stone 1969a). Remarks -Original material of Freycinetia jagorii has been located in B. This original material is mounted Willdenowia 51 -2021 on two sheets representing a single specimen, which is considered here as the holotype. A duplicate is extant in FI.
Freycinetia jagorii is listed as an accepted species by Stone (1969a) but a critical taxonomic revision of Pandanaceae is required for the Philippines. Remarks -Freycinetia lagenicarpa was described based on two gatherings by Friedrich Richard Rudolf Schlechter (1872Schlechter ( -1925 in Papua New Guinea: Schlechter 14571 and Schlechter 14572. Specimens of the two gatherings are extant in B. The most complete specimen, Schlechter 14572 is designated here as the lectotype. Duplicates have been located in FI for both gatherings. Remarks -A single specimen of original material, Sarasin 663, is present in B. This specimen is considered here as the holotype. A duplicate in FI represents a fragment packet containing a fragment of an inflorescence and a photograph of the holotype in B. Freycinetia latispina is considered as a synonym of F. minahassae (Stone 1969b). Remarks -A single specimen of original material is present in B. This specimen is considered here as the holotype. A duplicate is extant in FI.
Freycinetia lauterbachii is considered as a synonym of F. funicularis (Stone 1969a Remarks -A single specimen of original material is present in B. This specimen is considered here as the holotype. A photograph of the holotype is extant in FI. Remarks -As already stated by Martelli (1910c: 313), the original material of Freycinetia novo-guineensis is a mixed gathering. The leaves are from F. pseudoinsignis and the fertile material from F. funicularis.
We designate here the fertile material of Warburg 21000 as the lectotype.
Freycinetia novo-guineensis is considered as a synonym of F. funicularis (Stone 1969a). Remarks -The extant material in B represents a fragment packet with Warburg's handwriting "Freycinetia papuana Warb." Despite the fact that Hollrung 218a is not mentioned, we do consider this material as original. This specimen is considered here as the holotype. A duplicate is extant in FI.
Freycinetia papuana is considered as a synonym of F. funicularis (Stone 1969a Remarks -Freycinetia reineckei was described based on four gatherings: Reinecke 255, 255a, 255b and 362, all cited in the protologue and therefore syntypes. Only Reinecke 255a is still extant in B, mounted on two sheets. B 10 0367684 is wrongly labelled both as 255a and 355a, the latter clearly a typographical error. The two sheets in B comprise a single specimen, which is designated here as the lectotype. Other syntypes have been found in G (Reinecke 255) and FI (Reinecke 362). A sheet is present in B (B 10 0673256) with a copy of the illustration published in Warburg (1898: t. 8, fig. B). Remarks -Warburg described Freycinetia samoensis based on two gatherings: Reinecke 355c and 362a, both cited in the protologue and therefore syntypes. A single specimen of original material is extant in B and bears three labels: two as 362a and one as 355c. We consider the correct collecting number to be 362a. A fragment packet in FI, which Martelli most likely received from Berlin, bears collecting number 355, but with the locality of 355c. This should be considered as a typographical error for 355c. B 10 0673226 bears the original pencil drawing published in Warburg (1898: t. 8, fig. A). A copy of this drawing is also extant on another sheet in B (B 10 0673248).

Freycinetia samoensis
Freycinetia samoensis is considered as a synonym of F. storckii Seem. (Stone 1969a). Remarks -A single specimen of original material is present in B. This specimen is considered here as the holotype. No original material has been located in FI. Remarks -A single specimen of original material is present in B (Fig. 3). This specimen is considered here as the holotype. Two duplicates are extant in FI. Remarks -A single specimen of original material is present in B. This specimen is considered here as the holotype. A duplicate is extant in FI.

Freycinetia schlechteri
Freycinetia stenophylla is considered as a synonym of F. angustissima (Stone 1969a Freycinetia streptopifolia is considered as a synonym of F. beccarii (Keim 2009 Remarks -A single specimen of original material is present in B. This specimen is considered here as the holotype. A duplicate is extant in FI. Freycinetia involuta shows no morphological differences from F. sulcata, an endemic species from New Caledonia (Fig. 2C), and is considered here as a synonym. Remarks -A single specimen of original material is present in B. This specimen is considered here as the holotype. A duplicate is extant in FI.

Freycinetia verruculosa
Freycinetia delicata shows no morphological distinction from F. verruculosa, a distinctive endemic species from New Caledonia (Fig. 2D), and has to be considered as a synonym. Remarks -The original material of this name in B and FI is very fragmentary. It consists of a bract, part of a syncarp and a leaf. Based on the limited available material, it is not possible to define the taxon to which this name applies. Remarks -Stone (1992: 53) considered the holotype of Pandanus amboinensis to be deposited in L. This could have been treated as an error to be corrected to lectotype (Turland & al. 2018: Art. 9.10), but original material is still extant in B and that specimen should be considered as the holotype. The original material of Pandanus krauelianus considered here as the holotype consists of a single specimen mounted on three sheets.

Genus
Pandanus amboinensis is a synonym of P. krauelianus (Jebb 1992;Keim 2009 Warburg as "Rev. Rowlands, Baron n. 4931", which is the same gathering cited by Baker and which is represented by a specimen still extant in K. This specimen is considered here as the holotype, with duplicates in FI and P. Pandanus cyaneoglaucescens was described by Martelli based on Perrier de la Bâthie 12406, also collected in the southern part of the highlands of Madagascar. Careful study of original material and recent collections have shown that these represent the same species as P. bakeri and the name is therefore considered here as a synonym. Remarks -No original material is extant in B. A fragment packet consisting of several phalanges is extant in FI. This packet was sent by Warburg to Martelli in 1903. This specimen is designated here as the lectotype. An original pencil drawing is extant in B (B 10 0673234). It represents the illustration published in Warburg (1900b: 47, fig. 13A -C). Remarks -A single specimen of original material is present in B. This specimen is considered here as the holotype. Duplicates are extant in G, FI and P.
A complete description of Pandanus unipapillatus accompanied with illustrations can be found in Nadaf & Zanan (2012: 35 -39).   Remarks -The original material still extant in B consists of a single leaf. A duplicate in FI consists of two fragment packets with drupes and two small leaves that must be excluded from the original material as they do not belong to Benstonea celebica (Callmander & al. 2012: 332). A photograph of a syncarp located in B, now destroyed, is extant in FI.

Pandanus celebicus
The taxonomic identity of this species has been resolved and was presented in Callmander & al. (2014: 162 -163), where an epitype was designated.

Remarks -No original material is extant in B.
A fragment packet consisting of a single phalange has been located in FI and is designated here as the lectotype.
Pandanus dammannii is considered as a synonym of P. spiralis (Wilson 2011: 222). Remarks -Pandanus delessertii was based on "Eydouxia ? delessertii" of Gaudichaud-Beaupré (1841), which was not validly published (see above). A single phalange originating from Mauritius [not Réunion] that served for Gaudichaud's drawing is extant in P (P00867902 carpol.). This specimen cannot be considered as original material because Warburg clearly specified "nur aus der Abbildung im Atl. Bon. bekannt" [known only from the illustration in Atl. Bon.]. Gaudichaud's illustration is therefore considered as the holotype. According to Bosser & Guého (2003: 38), this phalange is morphologically closely related to P. eydouxia. This phalange is nevertheless clearly an immature phalange of the latter species endemic to Mauritius. No collections are known from the nearby island of Reunion and this specimen is clearly from Mauritius.

Pandanus delessertii
Gaudichaud's epithet probably refers to Adolphe Delessert (1809 -1869) and not to his renowned uncle Remarks -Pandanus ellipsoideus was described based on four gatherings: Warburg 16150, 16151, 16152 and 16887, all cited in the protologue and therefore syntypes. One syntype of each gathering is extant in B, of which Warburg 16151 was designated as the lectotype by Stone (1978: 22). Warburg 16152 is the only original material located in FI.
Pandanus ellipsoideus is the basionym of Benstonea ellipsoidea (Callmander & al. 2012 Remarks -A single specimen of original material of Pandanus engleri is present in B. This specimen is considered here as the holotype. No duplicate has been located in FI. Pandanus engleri is considered as a synonym of P. rabaiensis (see Beentje & Callmander 2014 Ridl.) (Stone 1978;Callmander & al. 2012) (Fig. 4B) Remarks -Original material has been located in B. This single specimen mounted on two sheets is considered here as the holotype. A duplicate has been located in FI.
Pandanus goetzei is considered as a synonym of P. rabaiensis (Beentje 1993 Warburg's (1900b) monograph. An illustration is extant in FI, which served Martelli to illustrate this species in his Enumerazione delle Pandanaceae (Martelli 1913b: t. 24, fig. 4, 5). This drawing shows a side view and a top view of a phalange and is annotated in Martelli's hand "In herb. Berolinense, communicavit Engler 1904". This drawing is designated here as the neotype of P. hahnii because it represents the only link to the original material.
Pandanus hollrungii was tentatively put in synonymy of P. cominsii Hemsl. by Kanehira (1936: 544). The same author later questioned his decision by listing P. hollrungii as an accepted species but writing "this is possibly a synonym of Pandanus cominsii Hemsl. but we have no authentic material for comparison" (Kanehira 1940: 253). Merrill & Perry (1939) and Stone (1982) accepted P. hollrungii.
Study of the original material shows that Pandanus hollrungii has large leaves and a flattened pileus, with a scarcely raised stigma, which is characteristic of P. subumbellatus (Jebb 1992). Pandanus hollrungii is therefore considered as a synonym of P. subumbellatus. Finally, P. cominsii has much narrower leaves and infructescence and is therefore considered as a different taxon (Jebb 1992). Remarks -Balfour (1878: 66) rightly underlined that Gaudichaud-Beaupré (1841) mixed two species under his t. 21: Roussinia indica for fig. 1 -4 and Pandanus palustris Thouars for fig. 5 -9. The phalanges that served for the illustration of R. indica ( fig. 5 -9) are extant in P (P00867954, P00867955) and belong to P. palustris. The lectotype designated here is Gaudichaud's t. 21, fig. 1 -4  (Fig. 5). This illustration is a copy of the original illustrations in Fontana (1792) designated as the lectotype of P. leram by Turner (2013: 167). A copy is also extant in P (P01183280).
Pandanus indicus is therefore a synonym of P. leram. Remarks -Pandanus kaernbachii was described based on two gatherings: Hollrung s.n. and Kaernbach s.n., both cited in the protologue and therefore syntypes. Only material of Hollrung s.n. has been located in the spirit collection in B. This specimen is designated here as the lectotype. A reproduction of the drawing published in Warburg (1900b: fig. 13D) is also extant in B (B 10 0673255). The sheet in FI bears two fragment packets and a reproduction of the drawing published by Martelli (1913b: t. 6, fig. 3, 4). Only the fragment packet on the lower part of the specimen has an annotation in Martelli Pandanus kamerunensis is considered as a synonym of P. candelabrum (Beentje & Callmander 2014 Remarks -Original material has been located in B. This material consists of two gatherings: Kersting 52 in the wood collection and Kersting 693 in the herbarium. Kersting 693, a single specimen is mounted on three sheets, is designated here as the lectotype. There is a duplicate in FI.

Pandanus kaernbachii
Pandanus kerstingii was mentioned in the first part of Volkens's treatment of the economic plants of Togo (Volkens 1909a (Volkens 1909b). Volkens (1909b) ascribed the name Pandanus kerstingii to Warburg, but the latter was not mentioned as being an author of the treatment of Pandanaceae. We therefore consider the author of this name to be "Warb. ex Volkens" (see also P. togoensis).
Pandanus kerstingii is considered as a synonym of P. candelabrum (Beentje & Callmander 2014), but see notes under P. togoensis. Remarks -Original material has been located in B and is considered here as the holotype. Duplicates have been located in FI and WRSL.

Pandanus lauterbachii
Pandanus lauterbachii is the basionym of Benstonea lauterbachii (Callmander & al. 2012 Remarks -Balfour (1878: 52) had previously attempted to transfer "Dorystigma madagascariense" illustrated in Gaudichaud-Beaupré (1841) to Pandanus, but neither name was validly published. Warburg (1900b) validated the name P. madagascariensis by providing a Latin diagnosis, which was based on Gaudichaud's illustration. Warburg had no specimen in B: "Man kennt nur die Abbildung" [One knows only the illustration]. Three phalanges have been located in P in the carpological collection. Two phalanges have a label referring to "Vinsonia purpurascens" (P00867360 carpol.) but determined correctly by St. John as "P. concretus Baker of Madagascar". The label is probably wrongly associated to these phalanges. The third phalange (P00867905 carpol.) was probably collected by Jean Michel Claude Richard (1787 -1868). This drupe has a label referring to both "Dorystigma mauritianum" (see under P. mauritianus below) and "Dorystigma madagascariense". Bosser & Guého (2003: 39) already considered the illustration in Gaudichaud-Beaupré (1841) as the "type". The first author realized only recently that Warburg's Pandanus madagascariensis was a validly published name. This name was considered as "illegitimate" by Stone (1970c: 112) and was not mentioned in recent works (e.g. Callmander & Laivao 2002;Callmander & al. 2011). Pandanus madagascariensis, along with P. dauphinensis (Martelli & Pichi-Sermolli 1951), P. centrifugalis, P. circularis and P. erectus (St. John 1968), are all synonyms of P. concretus, a morphologically variable species distributed along the eastern coast of Madagascar (Fig. 4C). Callmander & Laivao (2002) revised Pandanus sect. Dauphinensia Martelli and accepted most of these species based on slight differences in the number of stigmas per drupe, size of syncarp and presence/absence of a filament (1 -2 mm) below the anther. After studying all the available material, we conclude that these differences do not justify recognizing different species along the eastern coast of Madagascar and they are considered here as synonyms.  (Kurz)

Callm. & Buerki
Remarks -Original material has been located in B and is considered here as the holotype. Duplicates have been located in FI, NY and US.
Pandanus merrillii is considered as a synonym of Benstonea affinis (Callmander & al. 2012). Remarks -No original material of Pandanus micracanthus is extant in B. A fragment packet containing immature drupes has been located in FI. This packet was received by Martelli from Engler in 1904 and undoubtedly represents original material seen by Warburg. It is designated here as the lectotype.

Remarks
Original material of Pandanus microstigma has been located in FI and P. The FI material consists of a photograph of the Pervillé s.n. specimen in P, a syncarp and a fragment packet received from P with a few loose drupes and another fragment packet that Martelli received from Berlin with a fragment of a staminate inflorescence.
The complete specimen in P that served for the good engraving of "Sussea microstigma" in Gaudichaud-Beaupré (1841) is designated here as the lectotype of Pandanus microstigma. The fragment packet in FI originating from Berlin with fragment of a staminate inflorescence is not considered as original material.

Pandanus militaris
No original material is extant in B. Original material has been located in several herbaria, and Gaudichaud 110 in P is designated here as the lectotype.

Pandanus petersii
No original material is extant in B (as mentioned by Beentje 2009), but a duplicate has been located in FI. This specimen is designated here as the lectotype. Remarks -The original material in B was destroyed and the FI duplicate is designated here as the lectotype.

Pandanus platycarpus
Pandanus platycarpus is considered as a synonym of P. kirkii (see Beentje & Callmander 2014). Martelli (1933) considered P. platycarpus as originating from Java and cultivated in Zanzibar. The Javanese P. odorifer (Forssk.) Kuntze is cultivated in Zanzibar, but pistillate plants seem to be absent; only staminate plants are cultivated for their fragrant inflorescences (Stone 1973). Remarks -Pandanus pseudolais was described based on two gatherings: Hasskarl s.n. and Junghuhn s.n., both cited in the protologue and therefore syntypes. No original material has been traced in B and FI. Rahayu (2011) mentioned that original material of both gatherings is extant in BO. Despite a thorough search, we could not locate any original material in BO. The only original material we could trace is a sterile specimen of Junghuhn s.n. in L. This specimen is designated here as the lectotype.

Pandanus pseudolais
Pandanus pseudolais is part of a complex of species of the taxonomically difficult Pandanus sect. Rykia (de Vries) Kurz. Backer & Bakhuizen van den Brink (1968) did not recognize P. pseudolais and considered it as a synonym of P. furcatus Roxb. This synonymy is not accepted here, because P. furcatus is endemic to the western coast of India (Nadaf & Zazan 2012). Further taxonomic work on this section is needed in Java. A recent study on this section showed that three species should be accepted: P. bantamensis, P. pseudolais and P. scabrifolius (Rahayu & al. 2011). This taxonomy is in agreement with Stone (1972) and is followed here. Remarks -No original material is extant in B, but duplicates of original material have been located in FI and L. Martelli received the material in FI from Leiden (two sheets) and from BM (two fruits in the carpological collection). The material in FI received from Leiden is the most complete. FI017865 bears a label "Java n o 3253 Forbes" and FI017864 "Borneo, leg. Forbes". We designate here FI017865 as the lectotype. FI017864 is considered as a duplicate and the locality "Borneo" on the label is certainly an error because all specimens in L are labelled as Forbes 3253 from Java.

Pandanus radula
Pandanus radula is considered as a synonym of P. helicopus (Stone 1972). Remarks -The original material of Pandanus reineckei in B consists of a single drupe in a fragment packet. It is considered here as the holotype. No material has been traced in FI.

Pandanus reineckei
This very characteristic species is endemic to the American Samoa islands and has been collected on the summits of several ranges in Savaii, Upolu and Tutuila (Stone 1967) (Fig. 4D). Remarks -Warburg published Pandanus rumphii citing as a synonym P. cermamicus var. sylvestris Kunth (1841: 98). Warburg's species name can therefore be treated as a replacement name for Kunth's varietal name. Warburg also cited as a synonym "P. montanus Rumph." with a reference to Rumphius (1743: 145, t. 77), where two entities were recognized: "Pandanus silvestris sive terrestris" or "keker wassi" (to which t. 77 belongs) and "Pandanus montanus" or "keker ewan" (not illustrated, see under P. terrestris). Kunth cited "P. sylvestris" and the same illustration from Rumphius and mentioned Rumphius after the validating description; he gave no indication that he knew the species from any source except Rumphius. Therefore we consider the Rumphius illustration to be the holotype. Warburg's name is illegitimate because it is a later homonym of P. rumphii Gaudich. (Gaudichaud-Beaupré 1841: t. 22, fig. 11 Remarks -Pandanus samoensis was described by Warburg based on both staminate and pistillate individuals. The locality mentioned in the protologue is "Upulu: Vaiele Mai 1894". Only a single specimen of original material is extant in B, representing staminate material together with a sketch of an immature infructescence, which was published by Warburg (1898: t. 8, fig. C). This specimen is therefore designated here as the lectotype. Martelli (1934: 23) wrote "I saw the type specimen in the Botanical Museum in Berlin, and I am sure that no confusion took place there. For male and female flowers belong to the same species." Martelli nevertheless left P. samoensis as an undefinable name due to the immature pistillate original material, but stated that the staminate material "probably [belongs] to P. tectorius". Earlier, Martelli (1913a: 30)  Remarks -A single specimen of original material is present in B and represents a single leaf. A duplicate is extant in FI and represents two photographs of the leaf extant in B with two fragment packets containing parts of an infructescence received by Martelli from Warburg. The FI specimen (Fig. 6) is designated here as the lectotype as it is more complete than the original material deposited in B. Remarks -A single specimen of original material is present in B. This specimen is considered here as the holotype. A duplicate has been located in FI.

Pandanus schlechteri
Pandanus schlechteri is a synonym of P. decumbens (see Callmander & Buerki 2013). Remarks -The description of Pandanus setistylus was based on two gatherings: Lauterbach 2505 and Hollrung 175, both cited in the protologue and therefore syntypes. Specimens of both gatherings are extant in B. Hollrung 175 (B) was designated as the lectotype by Stone (1978: 22). It consists of a single specimen mounted on two sheets with additional material in the spirit collection.

Duplicates of both gatherings have been located in FI.
Pandanus setistylus is the basionym of Benstonea setistyla (Callmander & al. 2012). Remarks -No original material of Pandanus spinifer is extant in B but original material has been located in BM and FI. The material in FI consists of a herbarium sheet with a fragment packet containing loose drupes and three syncarps and parts of leaves in the carpological collection. This original material representing a single specimen is designated here as the lectotype. Three syncarps have also been located in the carpological collection in BM. The material in FI served for Martelli to describe Pandanus sect. Acanthostyla Martelli (Martelli 1933). This section, endemic to Madagascar, is characterized by its distinctive "coniferoid" habit (see Stone 1970b).

Pandanus spinifer
Pandanus fetosus from the highlands of Madagascar (where P. spinifer has certainly been collected by Johann  10 0673282), was annotated by the first author as a syntype (Apr 2016), but it does not bear the locality "Dar es Salam", as indicated in Warburg's protologue and cannot be considered as original material.
The specimen in FI (barcode FI003586) consists of two fragment packets with loose drupes and two photographs of a now destroyed infructescence in B. Another fragment packet on the lower right of the same sheet in FI (barcode FI003585) is not considered as original material.
Pandanus stuhlmannii is considered as a synonym of P. rabaiensis (Stone 1973;Beentje & Callmander 2014 St. John (1965). Warburg did not cite any specimen and based his concept on earlier names and publications (St. John 1965). A specimen in B annotated by Warburg as "Pandanus fascicularis Lam var laevis Warb" is designated here as the neotype. Warburg (1900b: 46) considered P. fascicularis Lam. as a synonym of P. tectorius. A duplicate has been located in Philip Barker Webb's (1793 -1854) herbarium in FI. This gathering was collected by Gaudichaud in the Calcutta Botanic Garden and sent to Carl Sigismund Kunth (1788Kunth ( -1850 for study in 1841 in B (St. John 1965). Several Gaudichaud specimens in P possibly represent isoneotypes (P02131259, P02131262, P02131263, P02131264), but none bears the number "33", which may represent a numbering within the Gaudichaud material sent to Kunth and Webb. Willdenowia 51 -2021 This cultivated species with unarmed leaves originated from the Moluccas (named by Rumphius [1743: 147 -148] as "Pandanus moschatus seu laevis"). Roxburgh (1832: 744) already mentioned its cultivation in the Calcutta Botanic Garden and its origin in Amboyna (Ambon) and named it P. inermis Roxb. St. John (1965: 232)  Remarks -No original material is extant in B, but duplicates of original material have been located in FI. The FI material consists of four fragment packets and a glued phalange mounted on three sheets. The material on two sheets (FI014765, FI015004) originated from Berlin. The third sheet (FI017932) bears a fragment packet annotated by Martelli: "Pandanus tectorius var. pulposum Warb. det. Warburg. Con questo nome trovasi nelle collezioni di Berlino misto ad altre forme sembra tutte con la stessa determinazione. Fra quelle forme vi è pure il mio P. Fischeri. U. M." [Under this name it is located in the Berlin collections mixed with other forms maybe all with the same identification. Among them my P. Fischeri as well]. This specimen also originated from Berlin but it is uncertain whether it is original material. We therefore consider only the four phalanges mounted on FI014765 and FI015004 as belonging to the same gathering and comprising a single specimen, and these are designated here as the lectotype.
Pandanus tectorius var. pulposus represents a synonym of the widespread and morphologically variable P. tectorius (see Wagner & al. 2020b Pandanus samak Hassk. was described based on the Javanese vernacular name "Pandan samak" (Rahayu & al. 2008). This name is used for a cultivated plant, the leaves of which are used for weaving, especially mats. The cultivated species was derived from P. odorifer (Stone 1972), of which P. samak is here considered a synonym (see Rahayu & al. 2008 for an illustration of the cultivated species). Remarks -No original material is extant in B, but duplicates of original material have been located in FI. The FI material was received by Martelli from Berlin and represents both staminate and pistillate plants (Hillebrand s.n.). The fragment packets with loose phalanges received from Berlin (FI017961), Warburg in 1902 (FI017962) and Engler in 1904 (FI017863) originate from a single gathering and are mounted on the same sheet; they comprise a single specimen and are designated here as the lectotype. The fragment packet on the same herbarium sheet with staminate material (FI017960) is also considered to be original material.
Pandanus tectorius var. sandvicensis represents a synonym of the widespread and morphologically variable P. tectorius (see Wagner & al. 2020a -St. John (1967) carefully studied the original material of Pandanus tectorius var. sinensis in B. He designated the only pistillate material as the lectotype: "It is from Hainan, 1889, A. Henry. One sheet bears two young pistillate inflorescences. A second bears a packet with several very young phalanges, and one mature phalange but, since it is the only fruiting one in the series, it is designated here as the lectotype of the var. sinensis." Other syntypes are either sterile (Naumann s.n. and Warburg 5482) or represent staminate material (Henry 8290).
Remarks -Original material of Pandanus togoensis has been located in B. This original material is mounted on two sheets comprising a single specimen and is considered here as the holotype. The material is sterile and represents only young leaves. A leaf is not sufficient to determine a Pandanus species, and therefore the fertile specimen Grüner s.n., collected at the type locality and described in detail by Huynh (1987: 15), is designated here as the epitype. Volkens (1909b) ascribed the name Pandanus togoensis to Warburg, but the latter was not mentioned as being an author of the treatment of Pandanaceae. We therefore consider the author of this name to be "Warb. ex Volkens" (see also P. kerstingii).
Pandanus togoensis is currently considered as a synonym of P. candelabrum (Beentje & Callmander 2014). Further taxonomic work is needed in Togo to determine if more than a single species can be recognized. Recent field work at the type localities of P. kerstingii and P. togoensis tend to confirm that more than a single species is present in Togo (Ton Rulkens, pers. comm.). Remarks -This species was described based on a leaf collected from a plant introduced by William Bull (1828 -1902) originating from New Caledonia and described by Karl Koch (1809-1879. Koch (1872) named it Pandanus ceramicus K. Koch, which is an illegitimate later homonym of P. ceramicus Kunth (≡ P. conoideus Lam.), which was first described by Georg Everhard Rumphius (1627Rumphius ( -1702 in his Herbarium amboinense (Rumphius 1743: 149, t. 79). The descriptions of both Koch and Warburg mention a leaf with a midrib unarmed, which may represent P. tectorius Parkinson, but in the absence of any original material P. bullii will remain doubtful.