75/125 YEARS

Re-Evaluation and Interpretation of the Fossil Record of Lyonothamnus A. Gray (Rosaceae)

ERWIN, Diane M., and SCHORN, Howard E., Museum of Paleontology, University of California, Berkeley, CA 94720-4780

Extant members of Lyonothamnus are evergreen trees endemic to California's Channel Islands. The modern genus is represented by one species, L. floribundus, with two recognized subspecies, ssp. floribundus and ssp. asplenifolius. Subspecies are distinguished on leaf morphology. Leaves of ssp. floribundus are simple, while those of ssp. asplenifolius are dissected into primary laminar segments (pinnate) that are divided again into triangular- to rectangular-shaped secondary segments. Lyonothamnus is restricted to California's Channel Islands with ssp. floribundus occurring only on Santa Catalina, while ssp. asplenifolius inhabits Santa Rosa, Santa Cruz, and San Clemente. Fossil species occur in Neogene paleofloras throughout the western U.S. However, the record shows a confused taxonomy with species based on two fossil leaf forms. Form B, a newly recognized transitional type, combines morphologies of both subspecies in being pinnate, but the primary laminar segments resemble the simple leaves of ssp. floribundus. Form B is known only from the California Mulholland flora and was identified as L. mohavensis Axelrod. However, L. mohavensis is based on specimens from the California Tehachapi flora and are fragments of Form A. Form A closely resembles ssp. asplenifolius with four species reported: L. mohavensis Axelrod from California, L. parvifolius (Axelrod) Wolfe and L. cedrusensis Axelrod from Nevada, and L. hesperius (Berry) (formerly Comptonia hesperia) from Washington and Oregon.

Review of the collections shows species are based on fragments of isolated primary segments of unknown position, and relative size of the secondaries was used as the basis for species delimitation. However, study of a large suite of complete leaves from Stewart Valley, NV shows size ranges used to segregate species fall within the range of this one species (viz. L. parvifolius), rendering size alone as invalid for species segregation. Results show important characters are number of primaries (critical for unequivocal identification but requiring complete leaves) combined with number of medial secondaries per primary segment (secondary segment index; SSI), length of primaries, and averaged lengths/widths of medial secondaries. Using these criteria we reject L. cedrusensis (=L. parvifolius) for material from the Nevada Purple Mountain and Stewart Valley paleofloras. The Mulholland Form B is a new species. Lyonothamnus mohavensis is retained for the Tehachapi material, as is L. hesperius (Berry) for the Washington and Oregon specimens. Using this revised taxonomy within an updated temporal framework we address questions regarding the evolution, biogeography, and paleoecology of this unusual California endemic.

75/125 YEARS