ON THE IDENTITY OF MIMOSA MARGARITAE AND M. TRICEPHALA IN BAJA CALIFORNIA, MEXICO

Since its initial discovery, Mimosa margaritae has been known only from the type, a non-reproductive specimen collected 125 years ago on Margarita Island, off the Pacific coast of Baja California Peninsula. Other specimens from the same Peninsula were mistakenly identified as M. margaritae, but are in fact Mimosa tricephala, one of the closest relatives of M. margaritae. During recent expeditions to the island, M. margaritae was rediscovered. Here, we provide the first description of the flowers and fruits of this species. The taxonomic status of the southern peninsular endemic M. lagunensis is also discussed; we conclude that it should be treated as a variety of M. tricephala, and provide a new combination, M. tricephala var. lagunensis.

The Margarita and Magdalena islands are located off the Pacific coast of the Peninsula of Baja California and are considered to be a priority terrestrial region for Mexican biodiversity [i.e., Planicies de Bahía Magdalena; Arriaga et al., 2000]. The mountains of both islands are composed largely of rocks dating from the Triassic to the Jurassic, making them some of the oldest in the Peninsula. According to Ortiz-Hernández et al. (2006), this area is one of 20 Mexican regions with serpentine rocks and magnesite deposits. The soil derived from these rocks is characterized as being limitative and selective for plant productivity (Brady et al., 2005).
The first botanical expedition to visit these islands and the neighboring peninsular region occurred in the fall of 1839, as part of the voyage of the English ship H.M.S. Sulphur. During this journey, the botanist George Barclay and the surgeon Richard Brinsly Hinds collected a great number of plants in the area of Cabo San Lucas and Bahía Magdalena. New species were described by George Bentham (1844) from these collections, but curiously they did not collect any species of Mimosa in such areas.
Between 1889 and 1902, the Californian botanist Townshend Stith Brandegee made several trips to the southern part of the Baja Peninsula. During the course of the first trip, which occurred between January and April of 1889, he explored the Bahía Magdalena region, and visited the Margarita and Magdalena islands (Moran, 1952), gathering collections of ca. 150 taxa, including a dozen of new species. In the checklist of the plants encountered during that trip, Brandegee (1889) listed a collection of an unknown species of Mimosa, with the annotation: "small bush neither in flower nor fruit, common in Santa Margarita Island".
In their revision of North American Mimosaceae, Britton and Rose (1928) treated 116 species of Mimosa, the last of which was published as "Mimosa (?) Margaritae Rose, sp. nov." based on the specimen collected by Brandegee, originally deposited at UC, with a fragment and a photo in NY and US. In his revision of Mimosa, Barneby (1991) maintained M. margaritae, describing it as "an obscure mimosa, resembling in foliage the 2-jugate state of M. caerulea Rose, but remotely allopatric and the flower and pod as yet unknown." Despite many subsequent expeditions by botanists in Baja California, including the Bahía Magdalena islands (Wiggins, 1964(Wiggins, , 1980, the species was not, as far as we know, recollected until very recently. Wiggins' inclusion of M. margaritae in the key to the species of Mimosa in these floras was based on misidentified specimens gath- We re-examined all of the existing material (digital images) of the original collection by Brandegee, comprising a sheet at UC (UC 0084040), and photo and fragment of that sheet at US (US 00000882) and NY (NY 00002571), which were probably taken from the original UC sheet (Barneby, 1991), which should be considered to be the holotype. This bears the annotation Mimosa margaritae n. sp. in Rose's handwriting, whereas the NY material bears the annotation Mimosa margaritae B. & R. [Britton & Rose], examined for North American Flora. The description of the photo and fragment at NY as the holotype in the (GPI) JSTOR website (Anonymous, 2013) was incorrect, since it is a secondary and partial representation of the UC sheet and the leaf fragment originates from US. In addition to Rose's annotation, the UC sheet bears the following annotation from Benjamin Lincoln Robinson: "I do not place this and am inclined to doubt the genus, B. L. R. If you ever get flowers or fruits, please send me some." Although it lacks reproductive structures, in the context of Mimosa, the Brandegee collection has a recognizable combination of branches with brachyblasts, straight to recurved prickles irregularly arranged along the internodes, and leaves with 1(2) pairs of pinnae and 5-7 pairs of obliquely linearoblong leaflets.
In September 2012, under the auspices of the Comisión Nacional para el Conocimiento y Uso de la Biodiversidad (CONABIO), we found, in the northeastern part of Margarita Island, a pair of individuals of Mimosa margaritae, one of them still with flowers and the other with fruits. The two individuals were separated by 2 km distance along a trail, suggesting that the existing population is of low density. Here we provide an updated treatment of the species with the first description of the flowers and fruits.
Another geographically proximal and closely related taxon, Mimosa lagunensis M. E. Jones, inhabits the oak-pine woodland of the Sierra de La Laguna in the heart of the Cape Region in the south of the Peninsula. Since the time of its initial publication (Jones, 1933), the validity of this taxon has been controversial, since it is close to M. tricephala Schltdl. & Cham. var. xanti (A. Gray) Chehaibar & R. Grether, this latter is widespread in the Baja California Cape Region, and in mainland Mexico.
According to Morton (1945), the original description of M. lagunensis was based on mixed material comprising the type collection from the Sierra de La Laguna (Jones 27270, erroneously published as 27290 in the protologue) and another Mimosa from [El] Cayuca [o] Ranch, near Loreto, Baja California Sur (Jones 27272). For details of collection localities and dates, see Jones (1935), Blake (1945) and Morton (1945). The latter collection, although labeled as M. lagunensis was not cited in the protologue of M. lagunensis by Jones and actually is M. distachya Cav. var. distachya, a member of series Distachyae (Barneby, 1991).
However, Morton (1945) considered M. lagunensis to be a synonym of M. xanti [now treated as M. tricephala var. xanti], and he suggested that M. margaritae is also a form of M. xanti. Chehaibar (1988) examined specimens from the Baja California Peninsula and from the entire distribution range (Michoacán to Chiapas, Morelos and Puebla in Mexico; Guatemala, and Honduras) of M. tricephala var. xanti, but not from Margarita Island. On these bases, she adopted Morton's concept for these taxa, thus placing M. margaritae and M. lagunensis in the synonymy of M. tricephala var. xanti.
However, Barneby (1991) treated M. xanti as a species different from M. tricephala. He tentatively recognized taxa A, B, C, and D in a key to the members of the Mimosa xanti complex, including M. lagunensis, among others, and excluding M. margaritae as a distinct species.
Several members of this group have been recognized as varieties of Mimosa tricephala (Grether, 2000) To test the taxonomic validity of the three taxa occurring in Baja California, we gathered a variety of morphological data from specimens of M. lagunensis, M. tricephala var. xanti, and M. margaritae deposited mainly at the herbaria HCIB and MEXU. The specimens were collected from the oak-pine woodland and tropical deciduous forest of the Sierra de La Laguna, from the xerophyllous scrubland in the peninsular lowlands and coastal dunes, as well as from Margarita Island.
The main differences are presented in Table 1 and outlined in Fig. 1. Based on these data, we conclude that: a) M. margaritae and M. tricephala are closely related but distinct species; b) M. lagunensis is conspecific with M. tricephala; and c) within M. tricephala there exist two varieties in the Baja California Peninsula, var. lagunensis and var. xanti.  Fig.1 (Mm), Fig.2.
Phenology: Flowering and fruiting from May to November.
Notes: Mimosa tricephala var. xanti is distinguished from M. tricephala var. lagunensis by the densely hispid branchlets, the hispid stipules, the shorter floral bracts and pubescent corolla lobes, as well as by the strigose to hispid and pubescent legumes (Table 1; Fig. 1).
The typical variety of Mimosa tricephala is distinguished from M. tricephala var. lagunensis and M. tricephala var. xanti mainly by the pubescent stipules and the legumes which are densely and largely hirsute, with setae of 2-4 mm long, not rigid on valves and margin. In