MADRONÄ O, Vol. 52, No. 4, pp. 222±257, 2005 SERPENTINE ENDEMISM IN THE CALIFORNIA FLORA: A DATABASE OF SERPENTINE AFFINITY H. D. SAFFORD1,2,J.H.VIERS3, AND S. P. HARRISON2 1 USDA-Forest Service, Paci®c Southwest Region, 1323 Club Drive, Vallejo, CA 94592 [email protected] 2 Department of Environmental Science and Policy, University of California, Davis, CA 95616 3 Information Center for the Environment, DESP, University of California, Davis, CA 95616 ABSTRACT We present a summary of a database documenting levels of af®nity to ultrama®c (``serpentine'') sub- strates for taxa in the California ¯ora, USA. We constructed our database through an extensive literature search, expert opinion, ®eld observations, and intensive use of accession records at key herbaria. We developed a semi-quantitative methodology for determining levels of serpentine af®nity (strictly endemic, broadly endemic, strong ``indicator'', etc.) in the California ¯ora. In this contribution, we provide a list of taxa having high af®nity to ultrama®c/serpentine substrates in California, and present information on rarity, geographic distribution, taxonomy, and lifeform. Of species endemic to California, 12.5% are restricted to ultrama®c substrates. Most of these taxa come from a half-dozen plant families, and from only one or two genera within each family. The North Coast and Klamath Ranges support more serpentine endemics than the rest of the State combined. 15% of all plant taxa listed as threatened or endangered in California show some degree of association with ultrama®c substrates. Information in our database should prove valuable to efforts in ecology, ¯oristics, biosystematics, conservation, and land management. Key Words: serpentine, ultrama®c, California, endemism, diversity. INTRODUCTION lain by ultrama®c rocks (6000 km2/406,280 km2). Ultrama®c rocks, often called ``serpentine'' by In addition, because they tend to have small geo- ecologists, botanists and pedologists, underlie more graphic ranges and because many of them occur in than 6000 km2 of the land area of the State of Cal- the rapidly urbanizing San Francisco Bay Area, ser- ifornia (Harrison et al. 2000). The edges of conti- pentine endemics are overrepresented among the nental plates often include bands of these vestiges state's rare, sensitive, and listed plant taxa (Skinner of oceanic mantle rock, accreted during the geolog- and Pavlik 1994). The ecology of California's ser- ic process of subduction, and later uplifted and ex- pentine plants has been extensively studied at the posed during mountain building and subsequent University of California's Sedgwick Ranch Reserve erosion. Ultrama®c rocks and the soils that develop (e.g., Seabloom et al. 2003; Gram et al. 2004) and on them are characterized by critically low levels McLaughlin Reserve (e.g., Harrison et al. 2003; of most principal plant nutrients (N, P, K, Ca), and Safford and Harrison 2004) and Stanford Univer- exceptionally high levels of Mg and Fe and a suite sity's Jasper Ridge Reserve (e.g., McNaughton of toxic trace elements including Cr, Ni, and Co. 1968; Huenneke et al. 1990; Hobbs and Mooney Outcrops of ultrama®c rocks support high numbers 1991). of edaphic-endemic taxa throughout the world Botanists have relied for two decades on the (Brooks 1987). The California serpentine ¯ora is monograph by Arthur Kruckeberg (1984) for most the richest in the temperate zone, and consists of of their information on Californian serpentine-en- hundreds of species and subspecies that are largely demic plant taxa. Since then, publication of the Jep- or entirely con®ned to ultrama®c substrates. son Manual (Hickman 1993), and a proliferation of Serpentine endemism is a key feature of the di- new botanical research and name changes have left versity of the California ¯ora (Raven and Axelrod this list in need of updating. Our initial aim was to 1978; Kruckeberg 2002). Of about 1410 full spe- modify Kruckeberg's (1984) list, primarily using cies endemic to the State (Hickman 1993), Kruck- information from Hickman (1993), to use in our eberg (1984) estimated that about 180 were endem- research on diversity patterns (Harrison et al. 2000, ic to serpentine. If these numbers are at least ap- 2004). However, it soon became clear that we proximately correct, then about 13% of the plant would have to expand and intensify our search for species endemic to California are serpentine en- the best available information. Complicating this demics. This is a remarkably high number when effort, plants show a continuum in degrees of ser- one considers that only 1.5% of the State is under- pentine restriction, and are sometimes more restrict- 2005] SAFFORD ET AL.: SERPENTINE ENDEMISM IN THE CALIFORNIA FLORA 223 ed in some parts of their geographic ranges than marks (``?'') were attached to those taxa for which others, thus contributing to inconsistencies among more information was necessary to con®dently as- reports from different sources. This led us to adopt sign their status. Some of the ``tentative'' endemics a semi-quantitative procedure for scoring plant taxa were included in the indicator appendix as well, on their reported degree of serpentine af®nity. thus these taxa occur twice in Kruckeberg's lists. In this contribution, we present a summary of our We combined Kruckeberg's two scales, and added current database of serpentine af®nity in the Cali- two levels to yield six levels of ultrama®c af®nity, fornia ¯ora. The database was constructed via an where 6 represents a ``strict endemic'' ($95% of extensive literature search, expert opinion, ®eld ob- occurrences on ultrama®cs), and successively lower servations, web research, and intensive use of ac- values signify lower af®nity to the substrate (5 5 cession records at key herbaria. It provides data on 85±94% of occurrences; 4 5 75±84%; 3 5 65± levels of serpentine endemism, rarity, geographic 74%; 2 5 55±64%; 1 5 45±54%). By this de®ni- distribution, taxonomy, and lifeform. tion, ``1'' thus represents a species found about half of the time on serpentine. We consider scores be- METHODOLOGY tween 1 and 2 to indicate ``weak indicators'', and a score of about 1 to mean an ``indifferent'' taxon. We began by conducting a database search of the The Kruckeberg ®delity scale crosswalks to ours in electronic Jepson Manual (Hickman 1993) main- the following fashion: ``111'' 5 6; ``11'' 5 5; tained by the Jepson Herbarium at the University ``!!'' 5 3; one ``!'' 5 2. Those taxa which occurred of California-Berkeley (UC-JEPS 2004a). The da- in both Kruckeberg's endemic and indicator tables tabase was queried for all taxa with ``serpentine'', had their two scores averaged: these all fell be- ``ultrama®c'', or related (e.g., ``asbestos soils'') ref- tween ``3'' and ``4'' on our scale. For example, Cu- erences in the habitat description. Taxa containing pressus macnabiana was rated ``111'' in Kruck- ``non-serpentine'' in the description were removed eberg's Appendix C (i.e., ``6'' on our scale), and afterward. We cross-checked the 391 serpentine-re- ``!!'' [i.e., ``2'' in our scale] in Appendix D; these lated taxa found in the Jepson Manual with Kruck- were averaged to ``4'' on our scale. eberg (1984), who listed those taxa he believed to We attached our categorical levels of ultrama®c be endemic to ultrama®c substrates in California, af®nity to all of the species in our hybrid Jepson- and those that were either local or regional ``ser- Kruckeberg database. In the case of the Kruckeberg pentine indicators'' (i.e., nonendemic taxa whose taxa, we simply cross-walked the Kruckeberg ®- distributions are nonetheless skewed toward occur- delity codes to our scale as described above, mak- rences on ultrama®cs). Taxonomic updates in the ing some adjustments based on more recent taxo- Jepson Manual (Hickman 1993) were applied to the nomic revisions and combinations. In the case of Kruckeberg list (which included 377 taxa after the Jepson Manual taxa, we were forced to interpret these revisions), and then those taxa not on the Jep- the language used in habitat descriptions to deter- son-derived list were added to our database. This mine levels of af®nity. We used the following in- resulted in a list of 529 taxa; of these, 287 were not terpretations of description language to assign af- shared between the two sources. We then added to ®nities: a ``6'' was assigned where the habitat de- the list a number of taxa that we considered to be scription categorically stated ``serpentine'' or ``ul- likely endemics or indicators but which were not trama®c'' (a ``5'' if there was some indication that indicated as such by either Kruckeberg (1984) or this restriction was not absolute); a ``4'' was as- the Jepson Manual (1993). Finally, published lit- signed where the modi®ers ``generally'' or ``usually erature (e.g., Meinke and Zika 1992; Nelson and serpentine'' were used; ``especially'' or ``often'' Nelson 2004; Baldwin 1999 and 2001; Barkley equaled ``3''; ``sometimes'' or ``occasionally'' 1999; Porter and Johnson 2000; Zika et al. 1998) equaled ``1''. In a few cases, af®nity levels were and the online Jepson Interchange Jepson Flora assigned based on ancillary information in the hab- Project (UC-JEPS 2004b) were consulted for tax- itat and/or range description rather than on explicit onomic revisions and taxa newly described since statement of serpentine af®nity. the publication of the Jepson Manual. We then conducted a broad survey of the litera- To score the af®nity of taxa to ultrama®c sub- ture, regional botanical experts, and herbaria rec- strates, we adopted a modi®cation of Kruckeberg's ords to obtain as many sources as possible for each measures of ultrama®c ``®delity''.