Additions to the Vascular Flora of San Clemente Island, Los Angeles County, California, with Notes on Clarifications and Deletions

The number of vascular plant taxa reported from San Clemente Island, California, is briefly summarized. Recent additions to the vascular flora are presented and, with one exception, representative voucher specimens are cited as substantiation. Of the taxa discussed , 68 are previously unreported taxa, and six are substantiations of previously dubious reports. An additional dozen taxa are cited in relation to clarifications or deletions. Based on current knowledge, we estimate the known flora to consist of 396 species with an additional 19 infraspecific taxa represented. Of these 415 taxa, 69.2% (272 speciesl15 additional subspecies or varieties) are considered indigenous to the island.


INTRODUCTION
The California Channel Islands are located in the southern California bight and comprise eight islands. These are generally divided into the Northern Channel Islands (San Miguel, Santa Rosa, Santa Cruz, and Anacapa) and the Southern Channel Islands (Santa Barbara, San Nicolas, Santa Catalina, and San Clemente).
San Clemente is the southernmost of the Channel Islands, and has its center at about 32°55'N-118°30'W. With an area of roughly 56 square mi (145 square km), it is the second largest of the Southern Channel Islands, and largely consists of weathered volcanic substrates. The island is narrow-approximately 1 mi (1.6 km) wide at The Isthmus on the N end, 4.3 mi (6.9 km) wide toward the S third, and ca. 20.5 mi (33 km) long-with a NW -SE axis. The NE side of the island is marked by short, deep canyons on a dramatically steep escarpment. The southwesterly slopes of the island are generally more gradual, with longer canyons and a distinctive series of wave-cut terraces formed over geologic time. The maximal elevation on the island is the "Thirst" triangulation point at 1965 ft (599 m), but this is essentially a large knoll, perched atop the eastern escarpment, which forms the culmination of the extensive uplands.
The first botanical collections on San Clemente Is-land were made in April 1885 by William S. Lyon and the Rev. Joseph C. Nevin, amateur botanists of the Los Angeles area (Raven 1963). Their collections were studied by Asa Gray, who described several new species from the material. Lyon (1886), the first to itemize the vascular plants on San Clemente Island, listed 81 taxa for the island as a result of his trip. Over subsequent decades, sporadic collections were made by additional botanists, including T. S. Brandegee, Blanche Trask, Philip Munz, Frank Peirson, LeRoy Abrams, and Ira Wiggins. Floristic summaries of the California Channel Islands were published by Brandegee (1890), Alice Eastwood (1941), and Meryl Dunkle (1950). Dunkle's report was based on 11 years of observations made by field botanists of the Channel Islands Biological Survey of the Natural History Museum of Los Angeles County. He had personally collected on San Clemente Island in April and November of 1939. Over the years, with the various activities of numerous botanists, erroneous reports for the islands began to accumulate. Eastwood's synopsis, for example, was based not only on herbarium collections but also included many "field observations" (without vouchers) and additional undocumented reports. In the early 1 960s, in order to clear the slate of erroneous reports and establish a more accurate listing of the vascular plants on San Clemente Island, Peter (:j:) = Number of taxa have been refigured based on reports and taxonomic schemes currently accepted by the authors, and native vs. no nnati ve status have also been redetermined for some taxa based on current knowledge of phytogeographic patterns; hence, the figures given here may differ slightly from the figures origi nally presented in the cited works. * = Additional infraspecific taxa (subspecies, varieties) if more than one is recognized in the fl ora. ** = To avoid ambiguity, these are total taxa recognized in the flora at or below the species level, but excludes taxa above that level. t = Note: Sporadic natural hybrids (e.g., Abronia maritima X A. umbellata; Lotus argophyllus var. ornithopus X L. dendroideus var. traskiae; Trifolium gracilentum X T. palmeri) are excluded from these figures , but *Pelargon ium Xhortorum. as stabilized " hybrid species," is included. Rhus ovata is tentatively retai ned, although at present no extant voucher specimen is known to exist in substantiation of its former occurrence on the island. Also, two taxa of debatable nati ve status have been arbitrarily tallied here as native, namely (*?) Bromus berterianus (B. trinii) and (*?) Madia sativa. Raven (1963) published a flora which he based on surveys of herbarium specimens at several major herbaria, a thorough review of previous reports for the island, and collections which he made during three trips to the island (totaling l3 days) in 1962. With two exceptions (Rhus ovata and Brassica nigra), previously reported taxa for which no herbarium specimens could be located were necessarily excluded. In his flora, Raven reported a total of 299 vascular plant taxa. Two years later, Raven (1965) added one additional taxon, Gnaphalium luteo-album, which he had collected earlier but overlooked in the preparation of his flora.
In the late 1970s, the U.S . Navy contracted with R. Mitchel Beauchamp to survey the island for sensitive species. This work also resulted in the addition of several interesting taxa (Beauchamp 1987a). Those reported by Beauchamp were Cyrtomium jalcatum, the exceedingly rare Sibara jilifolia (previously known only from historical collections on Santa Cruz and Santa Catalina Islands), Saxifraga californica, and Orobanche un iflo ra.
Finally, a comparative synopsis of the vascular plants on the California Channel Islands and Guadalupe Island, Mexico, was published by Gary D. Wallace (1985). In it, he included taxa that had accumulated in herbaria but which had not been formally re-ported for the island, along with a couple of taxonomic innovations. Among the taxa that he first reported for San Clemente Island were Apium graveolens, Gnaphalium

General Remarks on the Taxa Presented
In the intervening years, sporadic fieldwork on the island has brought additional records to light. Below, we itemize taxa which are being formally reported from the island for the first time, but we have also added a few taxa which were earlier considered dubious reports. Most of the collections reported here have been made by the authors; however, we have included additions made by other workers if it appears that these records have not been published elsewhere. A couple of previously reported taxa are also necessarily included here for the sake of clarifying taxonomic ambiguities. Otherwise, we have attempted to avoid itemization of taxa which have been merely affected by nomenclatural changes since the works by Raven (1963) or Wallace (1985). Finally, we have taken this opportunity to exclude a few taxa that to us appear to be erroneous reports, and have presented a brief numerical summary of the vascular plant taxa on San Clemente Island (Table 1). The number of addi-tions presented here that are not reflected in the treatments by Raven (1963;1965), Thorne (1969), Wallace (1985), or Beauchamp (1987a1987b) totals 73.
For convenience, the entries in the text are arranged alphabetically by binomial, and species that we consider nonnative are indicated by a preceding asterisk (*). Limited discussion of each addition is provided where relevant. In the citation of label data, elevations of collection sites and distances from locative features have been given in either feet (ft) or meters (m). We have generally retained these as they appear on the original herbarium specimen labels. Elevational units of contemporary United States Geological Survey (USGS) topographic maps covering the island appear in feet, and it is felt that in some cases converting these figures to a metric unit could lead to ambiguities, particularly when reference is made to bench marks (BM) which are identified by their elevation in feet. Since the 1970s, the U.S. Navy has attempted to standardize local place names in use on San Clemente Island, but current USGS topographic maps covering the island still lack most of them. Labels for recent collections on the island generally reflect these standardized Navy names. However, place names applied by collectors to identify previously unnamed physiographic features, or potentially ambiguous names, are generally presented here in quotation marks. For each specimen cited (with the exception of the first, q. v.), herbaria where vouchers are deposited are cited by the standardized acronyms in Holmgren et al. (1990). These are largely the Rancho Santa Ana Botanic Garden (RSA) and the Santa Barbara Botanic Garden (SBBG). The small herbarium at the Stone Field Station on San Clemente Island, in which some voucher specimens are deposited, lacks an official acronym and for convenience is here designated as "xSTO." Duplicates of some collections have been disseminated to additional herbaria, but no attempt is made here to itemize all herbaria housing a particular voucher. Reported on San Clemente by Dunkle (1950 p. 312) as Gilia gilioides var. glutinosa, but discounted by Raven for lack of a documenting specimen. The collection cited above was apparently the first specimen taken to document the species' occurrence on the island. By misfortune, the solitary plant was later lost in transit. Although we still lack a specimen, the report is included here fide the collectors. It is important that the locality be revisited and another herbarium specimen collected for final substantiation of the report. At the site, this taxon was apparently uncommon in the understory of platyopuntia scrub (Opuntia littoralis and 0. oricola) , but was associated with a rich assemblage of herbaceous and suffrutescent species including Sanicula arguta, Galium catalinense ssp. acrispum, Trifolium palmeri, T.  Originally reported for the island by Dunkle (1950), but excluded from the flora by subsequent workers for lack of a voucher. This appears to be the first substantiation of its presence on San Clemente Island, but is known only from this single location. This shrub is common on Santa Cruz, Anacapa, and Santa Catalina islands. A pernicious weed on the mainland that has apparently begun to spread on the island. It has only recently been found outside of the Wilson Cove area. It is likely that this annual to short-lived perennial species will rapidly expand its range if not extirpated soon. Several other weedy bromes (8. diandrus, B. hordeaceus, B. rubens) are already well established throughout the island and beyond the scope of any normal weed-abatement program. Currently known from a single location on San Clemente Island; also known from Santa Rosa and San Nicolas islands. It has been spreading on Santa Rosa in recent years but has not been seen recently on San Nicolas. It will likely come as a surprise to see this common, widespread taxon listed here as an addition to the flora. Claytonia perfoliata sensu lata has certainly been reported for the island earlier, but continuing monographic studies by John M. Miller and Kenton Chambers (Miller 1978;Miller and Chambers 1993) are resulting in a revised classification scheme for the highly variable C. perfohata complex. The collections cited above appear to represent the only typical ones made on the island. Based on annotated specimens at RSA-POM, all remaining collections of the species made on San Clemente Island have been reidentified as C. perfoliata ssp. Dunkle (1950, pp. 292 & 307) reported this species as occurring on San Clemente Island, apparently on the basis of personal observation. In a general discussion of Coreopsis-associated plant communities on the Channel Islands, he mentioned its particular vulnerability to grazing and browsing animals: " Since the branches grow at right angles to the main trunk, the Coreopsis shrub forms an intricate tangle which would be almost impossible to penetrate were it not for the extreme brittleness of the plant. A comparatively light touch will break off branches or even the main trunk. This fact might account for its disappearance where extensive grazing has occurred" (Dunkle 1950, p . 279). At the time that Dunkle visited the island in 1939, Coreopsis must already have been a rare plant. Raven (1963), did not encounter it on his visits to the island and, unable to locate a single herbarium specimen to substantiate its presence, necessarily excluded it as undocumented. With its exclusion from the island flora, this taxon became notable in its absence, si nce it is known on all the other Channel Islands as well as Guadalupe Island to the south.

Additions
The collection cited above apparently represents the first documentation of this species from San Clemente Island. The solitary plant that was encountered probably owed its survival to the fact that it was surrounded by a nearly impenetrable Opuntia scrub and, hence, was protected from browsing by the feral goats on the island. At the time of its discovery, a single flowering branch-tip was taken as a voucher. [For a partial list of associated species at the site, see the entry for Allophyllum glutinosum, above.] The solitary plant was revisited 17 May 1991 by Ross, Orlando Mistretta, and Mike Hammitt and was found to have put on considerable healthy growth in the intervening year. Many flowering and fruiting capitula were present and, while no additional herbarium vouchers were taken, some of the achenes were collected for the RSA Endangered Species Program. We had assumed that if there was one persisting individual, then there could well be more individuals persisting elsewhere on the steep eastern escarpment. This appears to be the case, as Jennifer Stone (pers. comm., May 1995) has now reported observing many plants of C. gigantea on the escarpment directly below " Jack. " With the recent removal of the last feral goats from the island, this species may now be able to gradually re-establish itself. This exotic annual is a common and widespread weed on the southern California mainland in disturbed settings, particularly in yards and gardens, but tends to be sporadic and uncommon in areas of undisturbed native vegetation . This is apparently the first collection of this parasite from the island and was made on Astragalus nevinii, a San Clemente Island endemic. According to Raven (1963), the only previously reported Cuscuta taxon was C. califomica var. califomica, based on Trask 187 (US). This is also the same and only collection of Cuscuta cited by Wallace (1985). It should be noted that if the taxon reported above were recognized at the specific level, as it was by the monographer T. G . Yuncker (1932), the name C. occidentalis has nomenclatural priority. Not previously reported and apparently quite rare on the island, only a very few plants being seen at the collection site. Due to the paucity of specimens, a solitary plant was taken as a voucher. This taxon is extremely rare on Santa Cruz and Santa Catalina islands, and rare on San Nicolas Island (Thorne 1967;Jlmak and Vanderwier 1990;Junak et al . 1995). This weedy annual has probably been established on the island for many years and overlooked by previous workers due to its superficial resemblance to E. botrys (Cav.) Bertol., with which it sometimes co-occurs. Erodium botrys appears to be the more common taxon on San Clemente Island, however. Erodium brachycarpum is reportedly native to southern Europe (Taylor 1993 Currently known on the island at Wilson Cove and at the two locations described above. In July 1981 , there were four live trees, one dead trunk, and several saplings along the main island road near the location of Ross 5363 (Juilak, pers. observation). Of the various eucalypts introduced to southern California, this species is the one most commonly seen naturalizing (McClintock 1993;Junak et al. 1995). Future workers on the island should be prompt in removing any saplings which appear. Currently known from a single location, where two large trees are persisting without cultivation. Flynn (1942)   This weedy taxon is now scattered in the central portion of the island, mostly on the E side of San Clemente Ridge Road between camera pads " Stone" and " Boulder." It has spread rapidly on Santa Cruz Island (Junak et al. 1995 This taxon is now common in disturbed sites (especially roadsides) throughout much of San Clemente Island; it has apparently spread rapidly during the last 10 years. It is also known from San Miguel, Santa Cruz, and Santa Catalina islands (Junak et al. 1995 This weedy taxon, which was apparently introduced to at least one site on the island with road gravel or construction materials, should be eradicated as soon as possible. It has recently been collected at two locations on Santa Catalina Island as well. This weedy taxon is becoming locally common in several locations on San Clemente Island and already poses a potential threat to the endemic Cryptantha traskiae which occupies similar habitats. It is also known from San Miguel, San Nicolas (Junak and Vanderwier 1990), and Santa Catalina islands, as well as Guadalupe Island, Mexico (Wallace 1985). We strongly recommend that eradication measures be taken as soon as possible. Although considered an annual or biennial species in Flora Europaea (Fernandes 1968), several of the plants seen on San Clemente Island were clearly perennating. However, they probably do not persist as individuals for more than 3 to 5 years. This taxon is also known from Santa Rosa and Anacapa islands; it has been spreading rapidly on Middle Anacapa Island. These are apparently the only reports of this weedy species on the island. The " sweet alyssum" of gardens, this taxon was likely introduced as an ornamental at the site and subsequently has been self-seeding and persisting in the vicinity.   Beauchamp in 1986(Beauchamp 1987. Subsequent molecular studies by Liston, Rieseberg, and Mistretta (Liston et al. 1990) using nuclear ribosomal DNA markers confirmed the hybrid origin of four individuals in the Wilson Cove area out of 38 plants sampled.
The Ross collection was from one of two hybrid individuals observed on a gentle, stabilized slope of coarse gravels. It was morphologically intermediate between the putative parents (presumably an F,), with twiggy spreading, glabrescent branches. The voucher material taken was from a plant growing interlocked with L. argophyllus (presumably the seed parent), the nearest L. dendroideus being about 4 m away where it showed a preference for outcropped rocks. Other taxa associated with the hybrid at the site included Poa secunda (P. scabrella), Stipa lepida, S. cernua, Melica impeifecta, Hemizonia clementina, Opuntia lilloralis, Calystegia macrostegia ssp. amplissima, Eriophyllum conJertiftorum, Malacothrix Joliosa, and Rafinesquia californica. It is likely that sporadic hybrids between these two taxa may be expected wherever they occur sympatrically.
The same putative hybrid was recently collected on a ridge S of Guds near the southern extreme of the island, amid a mixed population of Lotus argophyllus var. ornithopus and L. a. var. adsurgens (20 May 1995, Ross 8595 [RSA]). The solitary anomalous plant was growing out from under the margin of an individual of L. a. var. ornithopus and stood in stark contrast to the two Lotus taxa present. Lotus dendroideus was not observed in the vicinity, however, and is not known from the extreme southern portions of the island. [The most southerly documented population appears to be " E side of island near Malo Point, third canyon S of Bryce, 8 April 1990, Boyd, Ross, & Arnseth 4346 (RSA) ; this would put it 5800 m NW of the putative hybrid on the ridge near Guds.] Such a hybrid near Guds would apparently have to be the result of long-distance pollen dispersal by an insect vector. Because additional supporting evidence for the hybridity of thi s plant was not present at the site, this anomalous collection has not been cited with the other collections above.
[Although Lotus argophyllus var. ornithopus has been recently treated as a synonym of L. a. var. argenteus Dunkle (Isely 1993 On the basis of morphology, this collection appears best treated as this taxon, although the spheric fruits are as much as 5 cm in diameter, rather than 1.5-2.5 cm (Ferris 1960 Because the genus Marah is a common, widespread, and unshowy taxon in cismontane California, it has generally been taken for granted. Very little taxonomic work has been conducted in the genus, and very few herbarium specimens contain mature fruits and seeds, which are often critical to proper identification. For marahs on the California Channel Islands, the situation is particularly acute. The assumption has often been made that there is only one species to be found on the islands; hence, the variations exhibited by insular material have been explained away by oversimplifications rather than critical study. At present, there appear to be at least three taxonomic entities now documented on San Clemente Island alone. These three taxa are recorded in the citations above. The senior author is currently re-examining some of the type material of Marah from the islands, and has the intention of presenting elsewhere a slightly broader discussion of the taxonomic problems relating to the genus on California's Channel Islands. This annual to short-lived perennial taxon is apparently rare on San Clemente and has not been reported from any of the other Channel Islands (fide Wallace 1985). According to information provided with the collection by Elizabeth Kellogg, the species was "found in a soil depression of heavy clay in grassland, seasonally wet, open, with Malvella leprosa, Verbena bracteata, Avena barbata, and a few Typha latifolia." On the mainland, it ranges from Los Angeles County (where probably now extirpated) southward to San Diego County and Baja California, Mexico, and eastward to Texas and northern Mexico. In California, the species tends to occur on muddy shores of pools and merits concern due to increasing loss of habitat. Known from a single location on San Clemente Island, Where it persisted for at least 15 years; all known plants have now been removed. This invasive taxon is also known from Santa Cruz, San Nicolas, and Santa Catalina islands (lunak and Vanderwier 1990;lunak et al. 1995), as well as from Guadalupe Island, Mexico (Wallace 1985). The population reported here consisted of one large colony well established just above the bed of the drainage. There appeared to be some introgression in some local plants of Opuntia oricola, but this could not be ascertained with confidence. Hybridization between O. ficus -indica and native Californian Opuntia species is generally discredited, but requires further study. Currently known on San Clemente Island only from the location described above and from a small canyon nearby. Currently known from a single collection on San Clemente. This taxon is also known, primarily from disturbed sites near habitations, on Santa Cruz, San Nicolas, and Santa Catalina islands (Junak and Vanderwier 1990;lunak et al. 1995 ;Thome 1967 This is the first report of a Paspalum on the island, this species having only been reported previously from Santa Catalina Island (fide Wallace 1985). The occurrence was a dense, localized patch which may represent a fairly recent introduction. Reportedly this was one of several weeds on the island targeted for eradication (E. Kellogg, pers. comm., 1992) The "garden geranium" of the horticulture trade, this hybrid species has been planted as an ornamental about several buildings in the Wilson Cove settlement and about the airport at North Head. While not generally considered weedy by nature, its wind-disseminated seeds and drought-tolerance allow this plant to establish itself outside of cultivation. To date, the escapes noted have not traveled very long distances from their original introduction sites. Reported on San Clemente by Dunkle (1950, p. 309), but discounted by Raven (1963) for lack of any documenting herbarium specimens. While we cannot know how common this taxon may once have been on the island, its occurrence now appears to be relictual and quite restricted. This weedy perennial grass has already established itself in several areas on the N third of the island and appears to be spreading rapidly, particularly along the margins of the main N-S road. Serious efforts need to be made to extirpate this species from the island. Based on the rapid dispersal already evident, this will become a very serious, competitive weed in the years ahead unless it is eradicated promptly. These apparently represent the first formal reports of this taxon from the island, making it the second Psilocarphus species known to occur here. This species appears to be largely restricted to the margins of vernal pools, although P. tenellus var. tene/lus may also occur in this habitat as well as more widely on vernally moist adobe flat s. This taxon had originally been reported for the island by Dunkle ( 1950). In his discussion of plant ecology on the Channel Islands, he twice mentioned the taxon in regard to woodland and savanna communities as follows: "A few trees of Q. chrysolepis and Photinia [HeteromelesJ arbutifolia macrocarpa are to be found in protected canyon heads on San Clemente .. . " (p. 287); and, " On San Clemente Island Quercus tomentella, Q. chrysolepis, and Photinia arbutifolia form very small areas of savanna. These savannas are usually on terraces or on rolling hills about wide upland watercourses. The trees most frequently occupy slopes or swales where they receive some protection from the wind" (p. 297). Twelve years after Dunkle's report, Raven collected oaks throughout the island and had his collections examined by Cornelius Muller. All were identified as Q. tomentella. As Raven was apparently unable to locate any vouchers for this taxon in herbaria, he concluded that " records of Q. chrysolepis Liebm. from the island are evidently erroneous" (Raven 1963) and necessarily omitted it from the flora. Subsequently, Wallace (1985) listed the species as occurring on San Clemente Island and cited Dunkle 7361 (LAM). Many botanists, however, apparently regarded this report as an error and dismissed it. This voucher is indeed extant and has been cited above with the Boyd et al. collection as substantiation. As with much of the island's native flora , this species is now only relictual. Certainly the savanna formations are gone and the canyon woodlands are only a shadow of what they once must have been. Boyd indicated on his collection label " large tree, infrequent in canyon bottom." In fact, there may have been no more than one to three individuals encountered during our trek down China Canyon. This species is in need of careful conservation and replenishment efforts on the island. This pernicious annual weed apparently represents a recent introduction to San Clemente. In recent decades it has spread rapidly on the southern California mainland; it is also known from Santa Cruz, Santa Barbara, and Santa Catalina islands. Every effort should be made to extirpate this weed before it spreads more widely. This is an inconspicuous native annual that is often overlooked in the field. Based on these collections and field observations, the taxon appears to be restricted to the southern third of the island. Until recently, the common Suaeda species of southern California coastal areas was known as Suaeda californica S.Watson. Within this taxonomic framework, two infraspecific taxa were recognized: var. californica, a glabrous form ; and var. pubescens Jeps. , with pubescent herbage. Raven (1963) had indicated that " all material from San Clemente Island is pubescent and would thus be referable to var. pubescens Jeps., if that variety were recognized. " While the glabrous and pubescent forms of the species are not always recognized by taxonomists, the pubescence (or lack thereof) may result in rather different physiognomies. Recently, Wayne Ferren (UCSB) has determined that the name S. californica S.Watson has been misapplied and is applicable only to populations in the vicinity of San Francisco Bay (now apparently extirpated) and Morro Bay (San Luis Obispo County), and that the correct name for the S. californica auct .. non S.Watson, in southern California is S. taxifolia. Under thi s species name, infraspecific epithets are not available to di stinguish between the pubescent and glabrous forms of the taxon. Despite the lack of a taxonomic designation for this entity, the presence of the glabrous form on San Clemente Island is noted here should the character be deemed of taxonomic significance in the future. Currently known from the N and central portions of San Clemente Island, thi s weedy taxon is already spreading rapidly and should be eradicated as soon as possible. It is also known from Santa Cruz, San Nicolas, and Santa Catalina islands; it has spread rapidly on San Nicolas (Junak and Vanderwier 1990;G. Smith, pers. comm. , 1996) The common " garden nasturtium" of cultivation has been known to establish itself along coastal bluffs in southern California. The second collection cited here represents a small colony naturalized on an eroding embankment, but with no escapes from the general vic inity noted.

Excluded Taxa
A few taxa mentioned by workers post-Raven appear to be erroneous reports and are here excluded. The widely dispersed European weed, Euphorbia pep-Ius L., was reported from the island by Wallace (1985, p . 15) based on Dunkle 7216, " Pyramid Cove, E [side], lower slope, locally common" (RSA, ex LAM). This specimen consists of a 3.5 cm stem tip with one cyathium and one parasitized capsule. Re-examination of this material indicates that it is a "top-snatch" of Euphorbia misera Benth. Consequently, Euphorbia peplus appears to be undocumented on the island and should be deleted from the flora.
Wallace also included Phacelia cicutaria Greene ssp. hispida (A. Gray) Thome in his listing for the island (1985, p. 19), but cited no vouchers. We have not encountered this taxon in our fieldwork on the island, are not aware of any herbarium specimens to document its presence there, and have therefore excluded the report as a clerical error.
Gilia angelensis V.E.Grant was reported for San Clemente by Wallace (1985, pp. 23, 74) on the basis of an unspecified specimen at SBBG. The only possible voucher found at SBBG was Piehl 62-318, which was determined as G. angelensis by Alva Day in August 1963; this same specimen was redetermined as G. nevinii by Alva Day in April 1990. Without further evidence, G. angelensis should be excluded.
Navarretia atractyloides (Benth.) Hook. & Am. was included in the flora by Wallace (1985, p. 24) based on House & Grumbles s.n., August 5-13, 1930, (RSA, ex USC). This collection, however, represents robust material of N. hamata Greene ssp. leptantha (Greene) H.Mason (annotated by A G. Day 1993). Additional Navarretia material was reported (ioc. cit.) as N. hamata var. hamata. Examination of the other extant Navarretia collections housed at RSA-POM indicates that they are all representative of N. hamata ssp . leptantha. This taxon, to date, appears to be the only Navarretia documented on the island.
A report of Delphinium parryi AGray, cited by Wallace (1985, p. 25), is based on Raven 17820 housed at RSA Examination of that specimen, however, indicates that it was collected on Santa Catalina Island. The species is not otherwise known from San Clemente Island, and is therefore excluded.
Sanicula crassicaulis DC. is mentioned casually in text by Beauchamp (1987a, p. 576) in discussing pigrooting damage at the type locality of Delphinium kinkiense ("canyon north of Nanny," now known as Larkspur Canyon). However, S. arguta 1.M.Coult. & Rose is the only sanicle known to occur on the island and is still reasonably common at the site mentioned above. No specimens of S. crassicaulis have been seen by us at the site mentioned, or in herbaria, and the taxon has not otherwise been reported from the island. Its mention is therefore assumed to be an inadvertent error.
Cneoridium dumosum (Nutt.) Baill., was reported from the island by Roderick (1967) and thence by Munz (1974). This report has been perpetuated most recently in The Jepson Manual (Shevock 1993). However, Thome (1969) already pointed out that the report was based on misidentified material of Lycium calijomicum. We are aware of no collections that would substantiate the presence of this monotypic genus on San Clemente Island, and once again recommend deletion of the taxon from the flora.
The delicate annual grass Muhlenbergia microsperma (DC.) Trin. has been reported for the island by Dunkle (1950), Raven (1963), and Wallace (1985). Reexamination of the Muhlenbergia collections from San Clemente Island indicates, however, that all the material collected thus far represents the very similar Muhlenbergia appressa C.O.Goodd. This species is apparently restricted to the southern half of San Clemente Island and is disjunct in California to the eastern Mojave Desert, thence SE to southern Arizona and Baja California, Mexico. Hence, reports of M. microsperma on the island should be supplanted with M. appressa.
Finally, Polygonum aviculare L. was included in the island flora by Wallace (1985) citing only "RSA-POM." Examination of RSA-POM holdings in 1992 revealed two collections, Raven 17343 and Raven 17996, which had been identified by Raven as P. arenastrum Boreau, but each of which bore a penciled annotation by R. E Thome stating simply "probably P. aviculare L." According to James C. Hickman (1993), true P. aviculare has not been documented in California, the name being generally misapplied to P. arenastrum. Here we accept the original determinations for these two collections and delete P. aviculare from the known flora of San Clemente Island.

Need jor Continued Fieldwork
It is important to note that several of the species listed above (Allophyllum glutinosum, Brickellia calijornica, Coreopsis gigantea, Pellaea mucronata, Quercus chrysolepis) were originally reported by Dunkle in 1950, apparently based on his San Clemente Island fieldwork of 1939, but were excluded from the flora by Raven because voucher specimens could not be located. While Dunkle's work is not error-free, these affirmations of his reports lead us to speculate on some of the other taxa that he cited for the island but for which there appear to be no extant voucher specimens. Among the additional taxa reported by Dunkle, but not yet re-encountered on the island, are Stellaria nitens, Lupinus albifrons (also reported by Eastwood 1941), Salvia mellifera (also Eastwood 1941), Platystemon californicus (also reported by Brandegee 1890), and Solanum wallacei. These additional reports may represent taxa that have been extirpated on the island since that time, or they may yet persist on a remote ridgetop or canyon slope awaiting rediscovery.
Several native taxa reported in this paper had apparently never been reported for the island by previous workers (e.g., Draba cuneifolia, Stylocline gnaphaloides, Verbena bracteata, Nama stenocarpum). This underscores the fact that the island has not yet been thoroughly explored botanically, and that continued exploration by knowledgeable field botanists should reveal the presence of other rare native taxa. Admittedly, some of these may now be restricted to precipitous slopes or life-threatening canyons on the eastern escarpment, but the Verbena and Nama provide examples of two vernal pool species that have persisted on a readily accessible clay terrace, but which apparently eluded earlier workers due to their restricted distributions.

Conservation Concerns
The efforts to document the floristic diversity on San Clemente Island will be of little ultimate consequence unless they go hand-in-hand with well-planned and competently executed conservation activities. The documentation of newly introduced aggressive weeds, for example, should be followed with a quick effort toward extirpation before they spread from their points of introduction. Fortunately, the U.S. Navy has shown an interest in weed abatement, and several weed control projects have been undertaken on the island in recent years.
Control of invasive exotics is only one portion of the equation, however. Many of the native taxa on San Clemente Island have been driven to the brink of extinction by decades of relentless activity by feral goats and pigs. Removal of these pernicious herbivores was finally completed in about 1992, and a few species of vascular plants already appear to be recovering. For many other taxa, however, recovery may not be possible unless there is concerted intervention. As an example, the shrub Crossosoma califomicum Nuttall is endemic to Guadalupe Island, Mexico, San Clemente and Santa Catalina islands, and the Palos Verdes Peninsula on the California mainland (where there are only three or four shrubs known). This plant is still reasonably widespread and common on Santa Catalina Island, but on San Clemente the species is now reduced to a handful of solitary, isolated plants in several remote canyons. The plants appear to be capable of self-pollination, but seed-set also appears to be poor among the individuals (Ross, pers. observ. 1995). In such a case, it should be worthwhile to give each plant an identification number, root cuttings or germinate seeds from each of these individuals, and the following winter transplant the progeny to the vicinity of the persisting individuals in such a manner as to eventually promote cross-pollination between genetically divergent individuals (presumably increasing the seedset of the individuals and promoting greater adaptability among the seedlings). Several of the woody species now reduced to dangerously low numbers (such as Adenostoma Jasciculatum, Quercus chrysolepis, Rhamnus pirifolia, Ceanothus megacarpus, Ribes malvaceum, Malosma laurina, Lonicera hispidula var. vacillans, Sambucus mexicana vel ajf., and Malacothamnus clementinus) may benefit from this type of replenishment effort. Material used in such a program should originate on San Clemente Island, and under no conditions should it be supplemented or replaced with genetic material imported from the mainland or one of the other islands. This would be critical to maintaining the genetic integrity of the plants long isolated on San Clemente Island.

Floristic Summary Jor San Clemente Island
Although a complete itemization of the vascular plants on San Clemente Island has not been presented in this paper, we offer figures that we believe most accurately reflect the known flora. These are in line with substantiated reports since Raven's flora, deletions of erroneous reports and synonyms, and taxonomic innovations accepted by the authors. The summary presented in Table 1 should serve to illustrate the continued need for floristic documentation on the island.
Several native taxa are thought to have been extirpated on the island over the last few decades, largely as an effect of feral herbivores, but also as a result of more direct human disturbances. Among them are Lomatium insulare (Apiaceae), Malacothrix incana (Aster.), Senecio flaccidus var. douglasii (Aster.), Batis maritima (Bat.), Trifolium Jucatum (Fab.), Dissanthelium californicum (Po.), Dendromecon rigida ssp. rhamnoides (Rhamn.), Anemopsis californica (Saurur.), and Lycium brevipes var. hassei (Solan.). Comprehensive fieldwork will be required to verify the status of these and other taxa that have not been seen or documented in recent decades. ACKNOWLEDGMENTS We wish to acknowledge the valuable assistance and cooperation of Jan Larson, Natural Resources Officer with the U.S. Navy at North Island, San Diego, and the commanding officers of the Navy for allowing us the opportunity and privilege of collecting botanical specimens on San Clemente Island. Botanical exploration on the island was also greatly facilitated during the 1995-96 season by Jennifer Stone, recently appointed as Staff Botanist at North Island. Tim Ross and Steve Boyd wish also to thank Orlando Mistretta, until recently Coordinator of the Endangered Species Program at RSA, for his important role in the Garden's cooperative efforts with the U.S. Navy in recent years. Steve Junak is grateful to Mitch Beauchamp, Howard Ferguson, Ralph Philbrick, and Larry Sward for sharing their observations and knowledge of the island. Finally, we wish to acknowledge the help and pleasant company of various colleagues, associates, and field assistants whose names appear as co-collectors on many of the specimens cited above.
. LITERATURE CITED