H ISTORY

Late Quaternary Paleoecology of Grasslands and Other Grassy HabitatsPeter E. Wigand

Today as in the past, grasslands, and grassy steppe andchaparral, have been essential and dynamic elements of thewestern North American ecosystems. Since the appearance ofgrasses during the Eocene, they have provided both a crucialrole in the recycling of nutrients and an important habitatfor animal populations that have in many cases coevolvedwith them.

Grasslands and grassy steppes are dynamic ecosystem com-ponents that are constantly responding to climate, fire,animals, geomorphic change, and human impact. Grassabundance within vegetation communities, as well as thediversity of grass species, responds to changes in seasonal andannual precipitation and to changes in evaporation rate dueto variations in annual or seasonal temperature. This can beseen historically (e.g., the Dust Bowl) but especiallyprehistorically in the paleobotanical record, where there isabundant evidence that grass abundance, distribution, anddiversity have fluctuated significantly during the lateQuaternary. At times grasses have been much more, and atother times much less, ample within vegetation communitieswhere they presently occur. In the West, and in California inparticular, both pollen and plant macrofossil records provideevidence of the ebb and flow of grasses within late Quater-nary vegetation communities.

Although there is some phytolith evidence from dinosaurcoprolites (Prasad et al. 2005) suggesting the presence ofgrasses during the upper Cretaceous in central India, the firstwell-documented appearance of grass pollen (spherical shapeand single pore) in the evolutionary record suggests an ori-gin on the Gondwana continent (present-day South Americaand Africa) shortly before the beginning of the Paleocene(!65 million years ago) (Jacobs et al. 1999). The earliest

unequivocal grass fossils date to the Paleocene-Eoceneboundary, about 56 million years ago (Jacobs et al. 1999;Kellogg 2001). After grasslands’ appearance (65–50 mil-lion years ago), their expansion seems to have been limiteduntil the middle and late Miocene (ca. 20–10 million yearsago), when grasslands and grass-rich ecosystems becamewidely distributed as a result of either lower atmosphericCO2 content, which gave grasses a photosynthetic advan-tage, or, more likely, climatic changes that created a fireregime suitable for the replacement of woodlands by grass-lands (Keeley and Rundel 2005). Keeley and Rundel (2005)suggest that during the middle and late Miocene, climatesbecame more seasonal, resulting in an annual cycle com-prising a wet season of high plant production followed by adry season during which these materials dried. A monsoonalclimate coupled with the dry season generated storms withabundant lightning, which ignited fires that cleared the pre-viously dominant forest habitats and paved the way for grass-land expansion.

During the Miocene epoch, 20 million years ago, grassspecies developed characteristics that are similar to those ofmodern grasses, even identifiable to modern genera. In partic-ular, they evolved with herbivore grazing. This is especiallytrue of western North America, where grassy habitats and her-bivores characterized the Miocene of much of the region andhad coevolved through the Eocene. In addition, grasses seemto have developed the capacity to respond during the same yearto dramatic increases in either winter or spring/summerprecipitation.

The history and environmental relationships of grassy habi-tats of the West during the late Quaternary is being revealedby three kinds of paleobotanical evidence: pollen, plantmacrofossils (primarily seeds), and phytoliths. Each of theseprovides different yet complementary kinds of evidenceregarding the distribution and abundance of grasses, and in

F O U R

Pleistocene and Pre-European Grassland Ecosystems

3 7

some cases of their importance to human and animal popu-lations. Grasses, like all plants that rely upon the vagaries ofthe wind for pollination, produce relatively large quantities ofpollen. These clouds of pollen settle across the landscape andaccumulate in lakes, bogs, or other locations favorable to theirpreservation. Analysis of samples from such places provides apotentially continuous record of the local and regionalrelative abundance of grasses.

Phytoliths (silica concretions that form in the cells of grassplants) provide a record of their actual distribution on thelandscape. Because they are deposited in the ground and areburied when the plant dies and decays, phytoliths rarelyblow around the landscape. Therefore, they mark the placeswhere grasses actually grew.

Grass macrofossils usually occur in contexts where they havebeen collected either by animals or humans. In most cases,grass macrofossils are in relatively close proximity to the areaswhere they were collected. This is especially true in the case ofsmall mammals, which have a relatively restricted foragingarea. In the case of small mammals, nesting sites are rarely pre-served for very long, so they do not provide a long-term recordof grasses in a region. However, the indurated nests (middens)of woodrats provide an exception. Urine-encrusted woodratnests can preserve plant macrofossils, insects, and pollen fortens of thousands of years (Betancourt et al. 1990).

The relationship between plant macrofossils collected byancient peoples and the sources of these materials is a bit moreproblematic. People can move across great distances in order tocollect raw materials and food for their survival. However, inmost cases plant materials are used, processed, or stored closeto the area of collection (Anderson, Chapter 5). Like bothpollen and phytoliths, plant macrofossils must be deposited inplaces where their preservation is ensured. One exception tothis is if the plant materials are charred by fire. In that case theybecome very resistant to destruction after burial.

General Characteristics of the Late Quaternary History of Grasses

The paleoecological evidence of episodic increases anddeclines in grass abundance and changing distributions duringthe late Quaternary (we will restrict our purview to the last20,000 to 30,000 radiocarbon years before present [rcyr BP]) inthe western Unitied States consists primarily of pollen data.

Paleoecological evidence from the late Quaternary of theWest suggests that grass dynamics are primarily the result ofchanges in precipitation, though temperature may at timesalso play a role. Ideally, for our examination of grassland his-tory, we would examine pollen localities located in the CentralValley in the midst of the grassiest habitats, i.e., the northernportion of the Central Valley. Unfortunately, there are nogood palynological (study of pollen) records from such envi-ronments. Instead, our pollen records are obtained fromenvironments that, in most cases, do not correspond to grass-dominated habitats in the West. Currently, late Quaternarypollen records documenting the dynamics of late Quaternary

vegetation (including grass) have been obtained from avariety of habitats; including coastal estuaries, large andsmall lakes lying at a wide range of elevations on both sidesof the Sierra Nevada/Cascade mountain ranges, bogs andhigher mountain meadows, desert springs, caves, ancientwoodrat nests (middens), and archaeological sites. In mostcases, the archaeological records are either poorly dated orundated and so cannot be used to provide much informationregarding the late Quaternary history of grass. However, sev-eral of the coastal estuary, mountain lake and meadow, anddesert spring records are dated and provide a continuous, andat times detailed, record of grass expansion and contractionin plant communities throughout the West and California inparticular. These are supplemented by well-dated and, in afew cases, well-stratified ancient woodrat midden records(Wigand and Rhode 2002).

TH E PLE I STOCE N E

Thus far, there are four long pollen sequences recording localand regional vegetation change:

• The three-million-year sequence from Tulelake on theeastern edge of the Modoc Plateau in northeasternCalifornia (Adam et al. 1989)

• The pollen sequence from Owens Lake on the lowerportion of the Owens River east of the Sierra NevadaMountains in southeastern California, comprising anapproximately 180,000-year-long section (Woolfenden1993, 2003), and a lower section extending from thebase of this section to over 870,000 years ago (Litwinet al. 1993)

• The !130,000-year-long sequence from Clear Lake in thecoast range of northern California (Adam 1981, 1988)

• The ongoing analysis of a fragmentary million-year longrecord from the Buena Vista Lake Basin southwest ofBakersfield, California, at the mouth of the Kern River,which will also eventually provide some information ongrass history (Wigand 2006, unpublished data)

These records provide an indication of the response ofgrasses under California’s natural variation in precipitation,temperature, and climate. Variance in the environment rangesfrom annual (Reever-Morghan et al., Chapter 7) to decadalpatterns (e.g., El Niño/La Niña cycles) or to millennial scales(e.g., glacial cycles). Temperature-driven changes based uponorbital-scale climate change underlie many of these precipita-tion cycles (Ruddiman 2001; Liu and Herbert 2004), but theycan also drive grass response by reducing evaporation rates,thereby increasing effective water availability.

At scales of tens and hundreds of thousands of years, theearth’s orbital characteristics, including axial precession(precession of the equinoxes), obliquity (or tilt of the earth’saxis), and eccentricity, have resulted in differences in solarinsolation affecting global temperature (Milankovitch 1930).

3 8 H I S T O RY

Variations in the amount of heat accumulated in variousregions (oceans or land, Northern or Southern Hemisphere)have driven the global climate system (Cronin 1999: 560).Changes in millennial-scale solar insolation directly impactthe amount of moisture evaporated from the oceans, the pathsof this moisture across continents, its condensation as precip-itation, and finally its accumulation as glaciers. Glaciers, oncethey begin to grow, create their own local climates that furtherimpact the amount and nature of precipitation. Over the spanof the late Cenozoic the rise of the Sierra Nevada/CascadeMountains has further affected long-term grass distributionand abundance through changing the distributions of mois-ture on opposite sides of their crest. A general rule of thumbis that the Sierra Nevada Mountains have risen roughly 100meters per 100,000 years since about 3 million years ago. Givenorographic forcing of precipitation on the western slopes ofthese rising mountains, each successive glaciation forced moremoisture out of storms west of their crests and increased therain shadow effect on their lee. The eventual result of thisprocess might suggest the gradual decrease in grass abundanceeast of the Sierra Nevada crest and its gradual increase at siteson the west slope of the Sierra Nevada Mountains. This has not,however, been observed in the pollen records from Tulelakeand Owens Lake.

The three-million-year record of grass abundance fromTulelake in northeastern California’s Modoc Plateau clearlyreflects the trend to progressively cooler glacial cycles duringthe last 3 million years (Adam et al. 1989). Currently Tulelakeis much smaller than it was during the Pleistocene. In pre-settlement California, it was surrounded by marsh habitatsdominated by sedges and cattails. Now, lowlands beyond themarsh are composed of grassy sagebrush steppe. Whereasintermediate elevations are covered with western juniper-dominated woodlands, higher elevations are characterized bypine- and fir-dominated montane forests. Compared withareas further south and east in the Great Basin, the ModocPlateau region surrounding Tulelake is rich in grasses.

During the Quaternary, the pollen record indicatesincreasing abundance of grass in the Tulelake core. Thisappears to correspond directly to both increasingly cold andlong glacial cycles, as recorded in the !18O of Atlantic Oceansediment cores (Ruddiman 2001). In the deeper sections ofthe Tulelake core, greater grass abundance is both low andsporadic in its frequency (Adam et al. 1989). Grass abundanceappears to correlate only with the coolest of the glacial cyclesprior to 0.6 million years ago. However, at the top of the coreit is not only much more abundant, but abundant for muchlonger spans of time (Adam et al. 1989). The correspondenceof increased grass pollen in the Tulelake record with the cur-rent timing of late Pleistocene glacial cycles is given addi-tional support by coincident increases in pelagic lake algaeduring these periods, suggesting a deeper lake (Adam et al.1989). It appears that the early and later portions of theHolocene (the modern interglacial) have been cool enough tosupport more relatively abundant grass communities, com-pared to those found in previous interglacials (Figure 4.1a). At

Tulelake, abundant grass primarily reflects increased effectiveprecipitation due to reduced evaporation rates, althoughthere may have been a slight increase in real precipitation.Especially during the last 600,000 years, grass has been moreabundant than in the previous 2.6 million years (Adamet al. 1989). That transition occurred when 100,000-year gla-cial cycle dominance replaced 41,000-year glacial cycle dom-inance (Ruddiman 2001). Since 600,000 years ago, glacialcycles have been both more severe and longer in duration,factors that have encouraged the proliferation of grasses inareas where they would normally be rare.

At Owens Lake the record of grass is less clear. Counts ofgrass pollen are very low, and many samples do not even havegrass in them. This could indicate poor preservation or simplyits rarity. Despite this, there is a general correspondence ofgrass pollen appearance in both the Tulelake and Owens Lakerecords. The major difference is that the increase in grasses atOwens Lake between 500,000 and 1,000,000 years ago is muchmore dramatic than that at Tulelake. The two most abundantperiods of grass at Owens Lake during this period are centeredat !550,000 and at 750,000–800,000 years ago. Both of theseperiods correspond to major glacial episodes, as indicated inthe !18O record from the eastern Pacific Ocean (Ruddiman2001: Figure 12.16).

The pollen record from pluvial lake Buena Vista, southeastof Bakersfield, contains a discontinuous pollen record span-ning the last million years. Although analysis is ongoing,the record indicates several episodes of increased grass pollenduring the last 250,000 years, between 370,000 and380,000 years ago, about 440,000 years ago, 560,000 yearsago, and 680,000 years ago. (Wigand 2006, unpublished data).The most dramatic increase occurred at 560,000 years ago,roughly coinciding with the dramatic increase in grass in theOwens Lake record centered at 550,000 years ago. This latterevent corresponds with the longest-duration cool, moistepisode, as recorded in the Owens Lake pollen record byplants reflecting cooler, moister climate (our diagrams plottedfrom data presented by Litwin et al. 1993).

Therefore, both the Tulelake and Owens Lake pollen recordsindicate that cooler temperatures and increased effective pre-cipitation, due to reduced evaporation rates during glacialepisodes, appear to have been crucial in dramatically increasedgrass in semiarid woodland and shrub steppe habitats in thecurrently dry interior regions of the West. Was this also true ofthe region west of the crest of the Sierra Nevada Mountains?

Clear Lake in the coast range of northwestern California isthe only published record from western California thatextends beyond the latest Pleistocene (Adam 1988: 86).Currently, this lake is surrounded by blue oak (Quercusdouglasii)/gray or foothills pine (Pinus sabiniana) forest withsmall patches of chaparral near its southern shore. Grassesoccur primarily as understory in both the forest andchaparral habitats, and as a result are not as abundant as inmore open habitats. A 130,000-year-long pollen record fromthe lake provides a pollen record extending back into thepenultimate glaciation (Illinoian) (Adam 1988). Grass pollen

P L E I S T O C E N E A N D P R E - E U R O P E A N G R A S S L A N D S 3 9

has usually constituted less than 4% of the terrestrial pollenat the site. During the late Quaternary, there is a clear corre-spondence between grass pollen abundance and vegetationcommunity structure that can be tied indirectly to climate.Grasses seem to be slightly more abundant during interglacialperiods, when oak forests are dominant. This may reflect aslightly more open structure in oak forests, as opposed to themixed conifer forests that characterized the glacial periods.More light and precipitation was probably available for agrass understory to develop. This is confirmed by greater

abundance during interglacial periods of alders, willows,hazel, buckthorn, and composites (Adam 1988).

However, increased sagebrush during glacial cycles in theClear Lake record also suggests either that there were openareas nearby or that the conifer forest may have been moreopen in some areas. Because values of sagebrush pollen werenever greater than 10%, there probably was never a real sage-brush steppe community within the region (Adam 1988).Episodes of greater moisture during glacial periods seem tohave resulted in sagebrush decline and grass increase. These

4 0 H I S T O RY

F IG U R E 4.1a. Grass-to-sagebrush and grass-to-total-terrestrial (grass pollen percentage) pollen ratios for five of the longest pollen records fromCalifornia (the ratios from the other two long pollen records from California, Clear and Owens Lakes, are not illustrated here). The ExchequerMeadow data are available online from the National Climate Data Center, NOAA, at the North American Pollen Database (NCDC 2007). TheTulare Lake and Playa Vista core 1 data are available online from the Department of Geosciences at the University of Arizona (Davis 2002).The Tulelake data were obtained from Adam and Vagenas (1990: 307). The axis on the right of each plot records the ratio of grass to terrestrialpollen; the axis on the left records the ratio of grass to sagebrush. Courtesy of Dr. Peter E. Wigand.

periods may also have been slightly warmer as well, as is sug-gested by declines of fir during such episodes. In summary,over the span of the Pleistocene, grass abundance becamegreater in shrubby steppe environments east of the SierraNevada Mountain crest during each successive glacial period.This seems to correspond to a shift toward cooler, wetter, andlonger-duration glacial cycles. Cooler temperatures duringthese cycles resulted in greater effective soil moisture. In addi-tion, the wettest portions of these cycles were not during theglacial maxima, but during the slightly warmer onsets anddeclines (Wigand and Rhode 2002). During glacial maxima,reduced global temperatures caused orographic rainfall tocommence and be more intense at lower elevations than isthe case today on the western slopes of the Sierra Nevada andCascade mountains. As a result, much of the moisture thatreached the coast of California during the Pleistocene wasforced out of Pacific storms before they reached the crest ofthe mountain ranges. This resulted in dramatically reducedprecipitation in the intermountain interior during the glacialmaxima, when increased precipitation might be expected.However, reduced precipitation was somewhat balanced bymuch reduced evaporation rates, due to the lower globaltemperatures, as well. The ultimate result was a cold, drycontinental climate during glacial maxima in the inter-mountain West. However, just before and after the glacialmaxima were periods of warmer temperatures, when the rateand amounts of precipitation wrung out of Pacific storms onthe western slopes of the Sierra Nevada and Cascade moun-tains was reduced. That is, because of warmer temperatures,condensation of moisture occurred at higher elevations andmore slowly than during the glacial maxima. More and wetterstorms were able to cross the Sierra Nevada/Cascade moun-tain ranges and bring precipitation to the interior. Althoughevaporation rates were slightly higher, there was a net gainin effective precipitation. It is during these periods that wesee grass expansion in sagebrush steppe and juniper wood-lands in the north and in the middle and higher elevationshrub steppe communities in the south (Wigand and Rhode2002). At the same time, pluvial lakes reached their highestlevels (Benson et al. 1990), and glacial advances in theSierra Nevada reached their greatest late Pleistocene extent(Phillips 1996).

The pollen record from Tulare Lake at the southern end ofthe Central Valley is the longest record currently availablefrom southern California west of the crest of the SierraNevada Mountains (Davis 1999). There the pollen recordreveals three periods during the last 27,000 years whengrasses were relatively more abundant. During the last glacialcycle of the Pleistocene it appears that grass was more abun-dant between 26,000 and 19,000 rcyr BP than during theglacial maximum 18,000 to 19,000 rcyr BP (Figure 4.1a). Thisis similar to the record from Clear Lake, where decreasedgrass abundance corresponded to cooler, drier episodesaround the glacial maximum.

The transition from the Pleistocene to the Holocene is notwell documented. Unfortunately, the longer pollen records

discussed in the foregoing paragraphs have very low-resolution(wide spacing between samples) during the late Pleistocene/Holocene transition. In some cases, such as Tulare Lake, thePleistocene/Holocene transition is missing (Figure 4.1a).

The Playa Vista 1 pollen record from the Ballona Estuaryreveals a late Pleistocene increase in grasses between!12,500 and 11,500 rcyr BP that corresponds to increasedpine, and moist-climate shrub pollen values from the samecore (Figure 4.1a; Wigand in press, a). We know from pollenand woodrat midden macrofossil records from the north-ern Mojave Desert that climates were much wetter between13,000 and 12,000 rcyr BP than they had been during the gla-cial maximum (Wigand and Rhode 2002). Unfortunately,there is no preserved pollen record between 10,700 and 9,500rcyr BP from the Playa Vista 1 locality, and only poorlypreserved pollen from the Playa Vista 8 locality for this period(Figure 4.1a). Therefore, we do not have a good picture of whatmay have been happening with grasses during the transitionfrom glacial to postglacial climates in southern California.

The record from Exchequer Meadow from the westernslope of the Central Sierra Nevada Mountains provides arecord of the transition from late glacial alpine/subalpinegrassy habitats to Holocene Sierran Montane Forest. Currently,the meadow is dominated by sedges (Scirpus spp.) and rushes(Carex spp.) (Davis and Moratto 1988). During the latest Pleis-tocene (between !13,000 and 12,000 rcyr BP), ExchequerMeadow was dominated by grasses, sagebrush, and compos-ites. This suggests a climate considerably cooler and drierthan that of today. Grasses were more abundant at ExchequerMeadow at that time than at any time later during theHolocene. This is in part due to the fact that slopes sur-rounding the meadow were still almost entirely dominatedby subalpine grasslands. By 10,500 rcyr BP, grass abundancehad declined precipitously as sedges and rushes became moreabundant in the increasingly wetter meadow. The grasslandssurrounding the meadow were slowly invaded by pine andfir as they readvanced into higher elevations of the SierraNevada to areas surrounding Exchequer Meadow. Thesechanges signaled both warmer and moister climates, asstorms that, during the glacial period, had lost much of theirmoisture at lower elevations now carried increasing amountsof precipitation to higher elevations.

The Tulelake record indicates that during the last glacialcycle grasses were more abundant than during most of theHolocene. In addition, a dramatic increase of grass thatoccurred at !10,000 years ago at Tulelake may correspondwith an episode of effectively globally more moist climate dueto very cold temperature conditions that occurred between11,200 and 10,200 rcyr BP. Allowing for the vagaries of radio-carbon dating and changing deposition rates, this coolepisode is recorded in the Tulelake record at 10,000 rcyr BP.This cool episode may also appear in the record from TulareLake, where it is recorded as the highest early Holocene grassvalues centered at 10,000 rcyr BP (Figure 4.1a). At about thesame time, the European Younger Dryas event was bringingdramatically lower global temperatures.

P L E I S T O C E N E A N D P R E - E U R O P E A N G R A S S L A N D S 4 1

TH E HOLOCE N E

The Holocene record of grass in California is documentedby substantially more pollen records than during the Pleistocene(Figures 4.1a–d). It is clear that there are both long- (century-scale) and short-term (multidecade) cycles in grass abundance.

However, it is only the highest-resolution pollen records thatcan reveal what seem to be decade-long episodes of dramaticgrass increase (only one or two such records currently exist inthe West). In addition, the available pollen records show thatgrass abundance is highly variable. These differences may reflectmicroclimatic differences as well as the impact of local topog-raphy. Finally, it should be noted that the age assignments arenot exact. Differences in calculation of the deposition rate ofeach site due to the position and number of radiocarbon dates

4 2 H I S T O RY

FIGURE 4.1b. Grass-to-sagebrush and grass-to-total-terrestrial (grasspollen percentage) pollen ratios for four pollen records from Californiathat span much of the middle and late Holocene. The San JoaquinMarsh, Shellmaker, and John Wayne Freshwater Marsh records arefrom coastal southern California and are available from theDepartment of Geosciences at the University of Arizona (Davis 2002).The Dinkey Meadow record is from the west slope of the central SierraNevada Mountains and is available online from the National ClimateData Center, NOAA, at the North American Pollen Data Base (NCDC2007). The axis on the right of each plot records the ratio of grass toterrestrial pollen; the axis on the left records the ratio of grass tosagebrush. Courtesy of Dr. Peter E. Wigand.

FIGURE 4.1c. Grass-to-sagebrush and grass-to-total-terrestrial (grasspollen percentage) pollen ratios for four pollen records fromCalifornia that span the late Holocene. These pollen records are allfrom southern California west and northwest of the Los AngelesBasin. The San Nicholas Island data are available from theDepartment of Geosciences at the University of Arizona (Davis 2002).The data for Zaca Lake, Carpinteria Marsh, and Cleveland Pond arefrom Mensing (1993). The axis on the right of each plot records theratio of grass to terrestrial pollen; the axis on the left records the ratioof grass to sagebrush. Courtesy of Dr. Peter E. Wigand.

and the errors associated with them can result in slightdifferences from site to site in the apparent position of specificclimatic events. In general, there appear to be several cycles ofgrass increase during the Holocene that seem to reflect regionprecipitation patterns, and mirror their change through time.

After the conclusion of the early Holocene Younger Dryasevent, there was a period of summer-shifted precipitationbetween about 9,500 and 8,000 rcyr BP, coincident with thepost-glacial thermal maximum (Wigand and Rhode 2002).Although this is most strongly manifested in the southernportion of the intermountain West, it is evidenced as far northas the northern Great Basin. A middle Holocene increase inprecipitation that occurred between 8,000 and 5,000 rcyr BPwas concentrated primarily in southern California during aperiod of warmer temperatures (Wigand in press, a). A briefcool, winter-wet event centered ! 5,500 rcyr BP, because of its

brevity, is only seen in higher-resolution pollen records fromthe West (Wigand and Rhode 2002). The cool, winter-wet Neo-pluvial period occurred between !4,000 and 2,000 rcyr BP(Wigand 1987; Wigand and Rhode 2002). The evidence forthese events is much more pronounced in pollen sites in thenorthern half of the West than in the southern half. A dra-matic increase in precipitation at the same time that muchcooler temperatures occurred !2,000 rcyr BP. The impact ofthis event appears to have been similar throughout the West(Wigand and Rhode 2002; Wigand in press, a). Between!1,600 and 1,000 rcyr BP the West was characterized byanother period of wetter climate. However, this period wascharacterized by warmer temperatures and increased late sea-son (summer) precipitation (Wigand and Rhode 2002). A suc-ceeding period of warm, moist winter climate between !800and 700 rcyr BP resulted in a brief, though dramatic, increasein biotic productivity in the northern intermountain Westand in some areas of the southwestern United States as well.Finally, a period of cool, wet winters during the last phase ofthe Little Ice Age, between about !350 and 180 rcyr BP, resultedin the most recent surge of grass abundance in the West.

Pollen from Tulare Lake in California’s Central Valley pro-vides one of the most sensitive proxy records of Holocene veg-etation change for south-central California (Davis 1999). Thatrecord is tied to changes in lake level and marsh history of thelake as well (Negrini et al. 2006). At Tulare Lake there was anearly and middle Holocene episode of more abundant grassbetween 10,000 and 5,000 rcyr BP, with a major break about8,000 rcyr BP (Figure 4.1a). The middle Holocene grass eventclearly suggests more significant increases in precipitation thandoes the early Holocene event. Evidence from the MojaveDesert suggests that the early Holocene event is the result of anincrease in summer precipitation (Wigand and Rhode 2002).An increase in grass between 8,000 and 6,000 rcyr BP corre-sponds to a hypothesized middle Holocene increase in summerprecipitation that is revealed in recent reanalyses of two coresfrom the Ballona Estuary in the southwestern corner of the LosAngeles Basin (Wigand in press, a). The Playa Vista 1 and8 records from the Ballona Estuary provide some of the mostdetailed information on both local and regional vegetationchange for coastal southern California currently available.Although the timing and magnitudes of individual responsesin the two cores are variable, there is clear evidence of a middleHolocene increase in grass around the Ballona Estuary between8,000 and 6,000 rcyr BP (Figure 4.1a). This corresponds tomiddle Holocene increases in oak and chaparral species aswell (Wigand in press, a). A late Quaternary synoptic climatemodel reconstruction of southern California climate byDr. Reid Bryson of the University of Wisconsin suggests thatthis period was characterized by warmer annual temperatureand increased annual precipitation, including increases in bothwinter and summer precipitation (Wigand in press, a).

Farther north, at Exchequer Meadow, there are only slightincreases in grass pollen at that time (Figure 4.1a). However,they are insignificant when compared with early Holocenegrass abundance at Exchequer Meadow. Yet farther north, at

P L E I S T O C E N E A N D P R E - E U R O P E A N G R A S S L A N D S 4 3

FIGURE 4.1d. Grass-to-sagebrush and grass-to-total-terrestrial (grass pollenpercentage) pollen ratios for three pollen records from California thatspan the latest Holocene. The Playa Vista and Long Beach Campus pollenrecords are both from southern California: Playa Vista from the south-western corner of the Los Angeles Basin, and the Long Beach Campusrecord from the Long Beach area. The pollen data from these sites areavailable through the University of Arizona Department of Geosciences(Davis 2002). The Woski Pond record is from the west slope of the centralSierra Nevada Mountains, and the data are available through the NorthAmerican Pollen Database (NCDC 2007). The axis on the right of eachplot records the ratio of grass to terrestrial pollen; the axis on the leftrecords the ratio of grass to sagebrush. Courtesy of Dr. Peter. E. Wigand.

Tulelake, this event does not appear in the pollen record(Figure 4.1a). This suggests that this event did not extendmuch beyond the latitude of central California.

Evidence for increased grass in response to a brief, but dra-matic cool, winter-wet episode centered around 5,500 rcyr BPis evident only in the higher-resolution records from theShellmaker and Playa Vista 8 sites from southern California(Figures 4.1a and b). An event at Dinkey Meadow thatappears to have occurred at about 4,700 rcyr BP may also bethis episode (Figure 4.1b). In this case, however, the episodemay simply be shifted as a result of the problems previouslymentioned in calculating an accurate chronology for thecore. In the Intermountain West the 5,500-rcyr BP event isstriking. It can be traced from the eastern Washington Plateauto the northern Mojave Desert. It is manifested in data asdiverse as increased vegetation density in eastern Washington,shifts from desert shrub to sagebrush steppe vegetation insouthern Oregon, dramatic rises of lake levels in Lake Tahoe,and renewed spring activity in the northern Mojave Desert(Wigand and Rhode 2002).

Although several small-scale increases in precipitationhave been recorded in the West between 5,500 and 4,000 rcyrBP, the first significant ones occurred as part of what is calledthe Neopluvial. Three episodes of cool, winter-wet climatecentered at !3,700, !2,700, and !2,200 rcyr BP characterizethe Neopluvial period (Wigand 1987; Wigand and Rhode2002). The Tulare Lake, Playa Vista, Shellmaker, and, appar-ently, Dinkey Meadow sites may all record one or more of thethree episodes associated with this period (Figures 4.1a–c). Inthe interior West these events are much more stronglymanifest in the northern half of the region (Wigand andRhode 2002). Expansions of woodlands and grassy habitatsduring this period were the greatest since the end of the Pleis-tocene. These data suggest a cool, winter-wet precipitationpattern that probably originated in the northern Pacific.

Although the impact of the dramatic, but brief cold, winter-wet episode of climate centered around 2,000 years ago is mostdramatic in higher-resolution pollen records of the inter-mountain West, it also appears as an episode of increased grassin the Tulare Lake, Shellmaker, and John Wayne FreshwaterMarsh records (Figures 4.1a and b). The event is also recordedin the Playa Vista 1 and 8 records, but not as an increase ingrass. In the Ballona record it is characterized by dramaticregional increases in both pine and sagebrush pollen (Wigandin press. a). Because the increase in these pollen types was sogreat, they may have masked a similar increase in grass pollen.This episode is recorded in pollen sequences from DiamondPond in south central Oregon to Lower Pahranagat Lake insouthern Nevada (Wigand and Rhode 2002). This event mayhave been caused by a major volcanic eruption that occurredat that time and was recorded as the highest Holocene sulfurvalues in ice cores from both Antarctica and Greenland(Wigand in press, a). The dramatic increase of precipitationassociated with this event not only affected vegetation distri-butions but also resulted in dramatic changes in stream flowregimes, desert lake levels, and even shifts of stream channels

on the Mojave River (Ely et al. 1993) and the Humboldt River(House et al. 2001) for the next 150 years. In addition, althoughthe increase in grass in southern California does not seem asdramatic in the pollen record as it does in the intermountainWest, it must have been significant. Its importance in southernCalifornia is hinted at in the charred plant macrofossils recov-ered from Native American archaeological sites. A dramaticincrease in the number of radiocarbon-dated sites around theLos Angeles Basin attest to a sudden increase in native popu-lation (Wigand in press, b). Grasses constituted 20 to 25% ofthe seeds used by these Native Americans at that time (Wigandin press, b). Although some of the grass seeds were of varietiesthat might be found in sandy, or dune, areas, a significant pro-portion were of types that are typically associated with the ver-nal pools found on the Los Angeles Prairie (Wigand in press, b).This suggests that vernal pools may have been less ephemeralat that time and that the grasses around them may have beenmore abundant. Both factors may have played a role in draw-ing Native Americans to the Los Angeles Prairie at that time.

Perhaps the most interesting event of the last two millenniaoccurred between !1,600 and !1,000 rcyr BP. During thisperiod, which was contemporaneous with the EuropeanMedieval Warm period, warmer climate and decreased winterprecipitation characterized the West. However, this period wasalso characterized by increased late spring through early sum-mer precipitation in the northern half of the region andincreased middle to late summer precipitation in the south(Wigand and Rhode 2002). In the northern intermountainWest, grass abundance relative to winter-loving plant speciessuch as juniper increased dramatically (Wigand 1987). Theeffect of this episode on vegetation, animals, and people wasnothing short of spectacular. In an area stretching from north-ern Nevada to eastern Washington, grasses became much moreabundant in the sagebrush steppe and semiarid woodlandscharacteristic of the area at that time (Wigand and Rhode2002). In response to this, bison expanded into the lush newgrazing habitats. Radiocarbon dates from archaeological siteson bison remains records Native American pursuit of these ani-mals into areas where bison had not been since the earliestHolocene (Wigand and Rhode 2002). In the northwesternGreat Basin, late spring to early summer–shifted precipitationresulted in the final northward expansion of piñon pine andpromoted a major shift from root crops to pine nuts in thenative economy. In the eastern Great Basin the shift in seasonalprecipitation enabled Fremont horticulturalists to raise maizein areas where previously it could not be grown. Because thesepeople had no major irrigation systems but relied upon floodfarming on alluvial fans, they could grow corn only wherethere was sufficient summer rainfall available. When thisperiod ended, !1,000 to 900 rcyr BP, these peoples disappeared.In southern California the archaeological evidence for the useof grasses from vernal pool areas noted in the previousdiscussion of the 2,000 rcyr BP cool winter-wet episode con-tinued through this period as well (Wigand in press, b). By!1,000 rcyr BP it appears that Native American populationsdeclined precipitously around the southwestern corner of the

4 4 H I S T O RY

Los Angeles Basin, and evidence for their extensive use ofgrasses wanes.

During the last millennium two wet-climate events areevident in the paleoecological record. A warm, wet climaticevent !700 to 800 years ago is most strongly recorded inpollen records currently being analyzed in the northernGreat Basin (Sardine Meadow in northeastern California westof Reno, Nevada, and Summit Lake, north central Nevada).This event also appears in the Exchequer and DinkeyMeadow records and in the Tulare Lake record. It is also evi-dent in the results of a recent study of lake levels in theTulare Lake Basin (Negrini et al. 2006). In northeastern Cal-ifornia this period is marked by increased grass abundanceand greater spring discharge in the Sardine Meadows southof the Sierra Valley. At Summit Lake there are dramaticincreases in aquatic algae productivity, indicating warmerwater temperatures, and increases in floating aquatic plantabundance, indicating deeper water. At Grays Lake, Idaho,floating aquatic plants became more abundant relative tolittoral plant species, indicating slightly deeper water con-ditions as well. This event seems to be one that is primarilyrestricted to a relatively narrow swath across the northernGreat Basin stretching from the Sierra Valley in California toGrays Lake.

Finally, it is tempting to associate an increase of grass atTulare Lake during the last 300 years to the cool, moist LittleIce Age (Wigand and Rhode 2002; Figure 4.1a). Tulelake alsorecords the final Little Ice Age cool, moist climate event -(Figure 4.1a). However, a record of the Little Ice Age event ismore difficult to confirm in the other pollen records availablefrom California. An accurate age assignment to a finalepisode of grass expansion apparent in the upper sections ofmany of the cores from southern California, and its correla-tion to the Little Ice Age event, is almost impossible. However,it is highly probable that many of these grass increases docorrespond to the cool, winter-wet Little Ice Age. Further eastpollen records from the intermountain West contain abun-dant evidence of increased grass abundance and expansionof winter moisture-loving plant species during the Little IceAge (Wigand and Rhode 2002).

Although the record of grass abundance from the Holoceneappear to be more variable from the West, there is good cor-respondence to episodes of wetter climate (associated withboth warm- and cold-temperature climate) and, in a few cases,to shifts in seasonal distribution of precipitation. This corre-spondence is evident in late Quaternary grass pollen recordsfrom east of the Sierra Nevada Mountains as well. In the morearid regions of the West, plants (and their pollen records) aremore sensitive to slight variations in precipitation, so appar-ent responses can be very dramatic (Figure 4.2). There is alsoclear evidence that fluctuating precipitation resulting in veg-etation change also affected the distribution of the woodrats.The timing, abundance, and spatial distribution of woodratmiddens are clear evidence of the impact of climate uponwoodrat populations (Wigand and Rhode 2002; Betancourt etal 1990). A comparison of grass pollen from both sediment

cores and ancient woodrat middens reveals some of the samemajor episodes of grass abundance that were discussed above(Figure 4.2). In addition, the actual periodicity of ancientwoodrat midden occurrence reflects the recurrence of climatesfavorable to woodrats. As previously discussed, the grass eventsseen in Figure 4.2 were due to a variety of climatic factors.Some of these episodes were related to slightly warmer, wetterclimates at the onset and conclusions of glacial maxima(!40,000, 34,000, 21,000, and 10,600 years ago), others weredue to increased summer precipitation during episodes ofmuch warmer climate (!9,000 and 1,600 years ago).

Regional Climate Differences

From the review of the Holocene portion of this record it isclear that there are regional differences in the response of grass(and other plant species) that suggest regional differences inclimate. Differences in precipitation amount, and more impor-tantly in the seasonality of precipitation, are the factors mostclearly associated with these changes in vegetation. A compar-ison of several average monthly precipitation records fromweather stations in northern and southern California on bothsides of the Sierra Nevada Mountains reveals several clearpatterns (Figure 4.3). In northern California January andDecember tend to be the wettest months. East of the SierraNevada Mountains a slightly increased May/June precipitationis clearly evident. This provides an early growing season forgrasses. If precipitation increases significantly during thisperiod, as the paleovegetation record clearly indicates that ithas, it results in dramatic increases in grass abundance in shrubsteppe and semiarid woodland habitats east of the mountains(e.g., between 1,600 and 1,000 rcyr BP). In south-centralCalifornia maximum monthly average precipitation occurslater in winter than in the north and is centered aroundFebruary (Figure 4.3). This pattern is similar on both sides ofthe Sierra Nevada Mountains at lower elevations. Closer tothe Sierra Nevada Mountains precipitation is again centeredaround January, suggesting that the foothills have an earlieronset of heavy winter precipitation than the surroundinglowlands do. East of the Sierra Nevada Mountains at thesouth end of the Owens Valley, an August/Septemberincrease in average monthly precipitation (related to thesouthwestern monsoon) is evident (Figure 4.3). At thelatitude of Blythe, California, in the central Mojave Desert,August and September have the highest monthly average pre-cipitation (Figure 4.3). In southern California west of the SierraNevada, winter-dominated precipitation is again the typicalpattern (Figure 4.3).

This is roughly the current seasonal distribution ofprecipitation in California. However, as the paleovegetationevidence discussed in the preceding sections suggests, thispattern has shifted significantly in the past. For example, atLower Pahranagat Lake in the northeastern Mojave between1,600 and 1,000 rcyr BP, middle to late summer precipitationmay have increased significantly (Figure 4). Late spring andearly summer rainfall also allowed water levels in marshes to

P L E I S T O C E N E A N D P R E - E U R O P E A N G R A S S L A N D S 4 5

4 6 H I S T O RY

remain deeper well into the summer months, promoting agreater abundance of floating aquatic plants (Wigand andRhode 2002). Episodes of increased grass between 4,000 and2,000 rcyr BP suggest that the cold, winter-wet climatecharacteristic of the northwestern Great Basin may havepenetrated as far south as the northern Mojave during the Neopluvial (Figure 4.4). Even the warm, wet episode ofprecipitation between 800 to 700 rcyr BP is evidenced in theLower Pahranagat Lake record, suggesting that the May/Juneincrease in monthly average precipitation currently seen inthe northern Great Basin may also have extended into thenorthern Mojave during that period.

Conclusions

In summary, late Cenozoic grass populations in the West haveresponded to changes in both long- and short-term variationsin climate. On the millennial scale of glacial to interglacial

climates these variations have coincided with major shifts,not only in grass abundance and species composition but alsoin vegetation community composition and structure as awhole. This process was driven by variations in solar insolation(amount of solar radiation reaching the earth), resulting fromongoing changes in the earth’s orbit and tilt as well asvariations in solar output (radiation output by the sun), andmountain building. During the Holocene, variations primarilyin grass abundance (rather than in species composition) haveresulted from changes in precipitation on the scale of decadesto centuries caused by variations in solar output.

During the last 200 years the history of grass has been com-plicated by the impact of Euro-American activities and by theiranimals, primarily cattle and horses. For example, east of theSierra Nevada/Cascade crest the impact of the horse may havebeen felt as early as the 1680s, when Spanish horses escapedduring the Pueblo uprising. By the 1740s Native American

F IG U R E 4.2. Late Quaternary grass pollen records from east of the Sierra Nevada Mountains. Pollen records from the more arid sideof the Sierra Nevada/Cascade mountain crest are more sensitive to even small variations in precipitation. These records comparepollen from both sediment cores and ancient woodrat middens. Periodicity in the occurrence of ancient woodrat middens has beenshown in the past to reflect recurrence of climates favorable to woodrats. The pollen from these records show clear periodicity ingrass abundance during the last 50,000 years. Some of these peaks are due to glacial advances or cooler climate episodes (!40,000,34,000, 21,000, and 10,600 years ago), and others are due to increased summer precipitation (!9,000 and 1,600 years ago). Diagrammodified from one in Wigand and Rhode (2002).

FIGURE 4.3. Average monthly precipitation records for weather stations near some of the pollen localitiesdiscussed in the text. These provide an indication of the differences in the precipitation pattern withinCalifornia. (Source: Western Regional Climate Center, Desert Research Institute, UCCSN Reno, NV,www.wrcc.dri.edu/climsum.html).

horse cultures were well established, and grazing impacts uponnative grasses were in full swing. These impacts, together withfire, have resulted in significantly changed habitats. Not onlyhave vegetation community compositions changed, but theirstructure as well. Only in a few isolated places have relativelyintact native plant communities with rich grass understoriessurvived the onslaught of cheatgrass and other Eurasianinvaders (Figure 4.5). However, with the onset of global warm-ing, even these few communities may be in jeopardy.

Rancholabrean Mammals of California and Their Relevance for Understanding Modern Plant EcologyStephen W. Edwards

The term Rancholabrean is rightly associated with the spec-tacular latest Pleistocene fossil assemblage recovered fromthe Rancho La Brea tar pits in the Los Angeles Basin. But thatis only one among many assemblages of late Pleistocene agein California, not to mention a host of localities producingisolated mammal fossils. Savage (1951) defined Rancholabreanas a continent-wide mammalian provincial age, recognized bythe appearance in North America, approximately 150,000 yearsago, of Bison, an immigrant from Asia (Woodburne andSwisher 1995), and ending with the demise of “charismatic”megafauna such as mammoths and sabrecats at the end ofthe Pleistocene.

By Rancholabrean time the California flora consistedalmost entirely of genera (and, at least in woody plants, alsoof species) that are still extant in the state. There were differ-ences in distribution, and there were combinations of planttaxa that no longer occur together, but the flora and vegeta-tion were nevertheless distinctly Californian and recognizablymodern (Edwards 2004). The fossil flora recovered from theRancho La Brea tar pits, for example (Akersten et al. 1988;Templeton 1964; Warter 1976) is basically of central-southernCalifornia aspect. That paleoflora, as well as many others,show that cismontane California was not a Pleistocene arctic

waste, but experienced a temperate-maritime climate thatserved as a refuge for evergreen hardwoods (Johnson 1977).

The Rancholabrean fauna, looking at first encounter likesomething belonging in Africa or Asia, was adapted to aCalifornia flora. Mastodons browsed trees and shrubs, horsesgrazed needlegrass and the other familiar perennial grasses,which must have been more abundant and productive inthe more mesic Pleistocene climate than they are in today’sfully Mediterannean regime. Although human hunters prob-ably sealed the fates of many megafaunal species, it is likelythat a climatically induced type conversion from lushPleistocene grasslands to arid Holocene landscapes domi-nated by native annuals had already diminished megafaunalpopulations. The relative contributions of climate changeand hunting are still debated, but the result is clear. TheRancholabrean fauna, with so many large animals, represent-ing the zenith of the Age of Mammals in North America andwhat could be considered the true fauna of California, waswiped out forever in the space of about 2,000 years. Whilethe fauna that had coevolved for millions of years with theflora and vegetation of California thus disappeared, the floraand vegetation fared better. Though there are many gaps inthe fossil record and in particular herbaceous taxa are verypoorly represented, all indications are that late Pleistocenewoody species persisted to make up modern associations,and the same is true at least of herbaceous genera. Therefore,when it is evident that a modern native plant species is welladapted to grazing and/or browsing or even to fire, it makessense to recall the relations that were forged through millionsof years of coevolution between late Cenozoic (and especiallylate Pleistocene) mammals and California native plants.Adaptations affording grazing tolerance among living grass-land plants include, among others, widespread occurrencesof toxic or unpalatable defense compounds (e.g., Fabaceae,Madiinae, Ranunculaceae, Scrophulariaceae) and basal (ornear basal) meristems (probably in all Poaceae). Their pres-ence through numerous genera within families reflects thedeep histories of such adaptations. It is unlikely that they

4 8 H I S T O RY

F IG U R E 4.4. Grass-to-sagebrush and grass-to-total-terrestrial (grass pollen percentage) pollenratios for a pollen record in the northern Mojave Desert northeast of Las Vegas, Nevada. It isthe highest resolution pollen record currently published for the western Hemisphere andspans the latest Holocene. Sample spacing is decadal (Wigand, unpublished data).

suddenly appeared in response to decreased grazing pres-sures of the Holocene after the megafauna disappeared. Butthey would have constituted excellent preadaptations forfire, more intense in the arid Holocene, and grazing byhypsodont microtine rodents, rabbits, hares, grazing avifauna,and elk.

Much basic research needs to be done on dietary habits ofRancholabrean mammals, but enough data are available(Edwards 1996, 1998) for preliminary indications to be given.This will be done, telegraphically, in the following list of thelarger Rancholabrean mammals of California.

There is no way accurately to estimate population numbersof extinct mammals, but two lines of evidence suggest thatlarge Rancholabrean herbivores were very abundant and thusmust have had dramatic and pervasive impacts on vegetationand flora. First, the array of carnivorous mammals and largescavenging birds (two condors, four vultures, one teratorn, andsix eagles at Rancho La Brea) equals or exceeds that of the EastAfrican Pleistocene, which exceeded that of the game reservesof East Africa today. There must have been plenty of meat onthe hoof to support this diversity. Second, Davis and Moratto(1988) found abundant spores of the dung-consuming fungusSporormiella in sediments dated to 11,600 rcyr BP at ExchequerMeadow in the Sierra Nevada. They noted:

Sporormiella spores are abundant in modern sediments onlywhere introduced grazing animals are plentiful, and they areeven more profuse in sediments older than 11,000 yr B.P. inseveral sites. (Davis and Moratto 1988: 146)

The Larger Rancholabrean Mammals of California

E DE NTATA (G ROU N D S LOTH S)

Megalonyx jeffersoni: Variously reported as black bear to ox-sized, and consistently regarded, on the basis of the simplicityof its grinding teeth, as a browser.

Nothrotheriops shastensis: The smallest of the Californianground sloths, grizzly bear–sized at most. Dung deposits inUtah, Nevada, Arizona, and New Mexico have been attrib-uted to this animal. Studying deposits in Arizona, Hansen(1978) identified 72 genera of plants in Nothrotheriops dung.The most abundant taxa were Sphaeralcea ambigua (52%),Ephedra nevadensis (18%), Atriplex spp. (7%), Acacia greggii(6%), Cactaceae (3%), Phragmites communis (5%), and Yuccaspp. (2%). A similar study by Thompson et al. (1980) showeda greater percentage of Ephedra (51%), with Rosaceae andAgave the other prominent components.

Paramylodon harlani (Harlan’s Ground Sloth): Also knownas Glossotherium h., these were the largest edentates inCalifornia, ox-sized and weighing up to 3,500 pounds. Theywere capable of standing on their hind feet and manipulat-ing tall vegetation with massive forelimbs armed with largeclaws. Their simple, peglike grinding teeth have been inter-preted as useful for grazing grass as well as for browsing, butdefinitive research to elucidate their diet has not been done.

U RS I DAE (B EARS)

Arctodus simus (short-faced bear): These huge animals, out-sizing polar bears and having longer limbs, making themcapable of bursts of greater speed, were perhaps the mostpowerful predators of the Pleistocene world. Convergences inskull structure with felids and cheek teeth less modified foromnivory than those of other bears suggest that the largeungulates of the day may have been prime targets (Kurten1967; Shaw and Cox 1993).

Ursus americanus (black bear): These small bears (by Pleis-tocene standards) are omnivorous, with emphasis on plantsand insects.

Ursus arctos (grizzly bear): The diets of these animals arewell understood from studies in the northern United Statesand Canada. They are omnivores, feeding on everythingfrom limpets in the intertidal zone to glacier lily bulbs in thesubalpine, with copious supplementation from fish andmammals. In 1862 Brewer (1966) observed the extensiverototilling effects of these animals resulting from digging forbulbs in the south coast ranges. Whether in the Holocene(when it is likely that California’s geophyte flora attained itsgreatest prominence) or earlier, bulbs must always have beena major food source for grizzlies.

CAN I DAE (DOG S)

Canis dirus (dire wolf): This wolf was similar in size to the mod-ern gray or timber wolf but had a larger head, stronger jaws,more massive teeth, and a heavier build, but shorter lowerlimbs. It was presumably a pack-hunter like its modern relative,

P L E I S T O C E N E A N D P R E - E U R O P E A N G R A S S L A N D S 4 9

FIGURE 4.5. Grassy, semiarid woodland east of the Sierra Nevada crest.Today overgrazing and fire have decimated much of the grass cover ofthe Great Basin at lower elevations. At intermediate and higherelevations, however, precipitation is high enough for native grasses torecover from these impacts, and compete with invading plant species.This photo was taken in the Virginia Mountains just northeast ofReno, Nevada. The grasses are primarily bluebunch wheatgrass(Pseudoroegneria spicata, formerly Agropyron spicatum). Pollen recordsindicate that prior to the arrival of Euro-Americans most of thevegetation communities in the Great Basin, ranging from the lowersagebrush through the semiarid woodland and upper sagebrushcommunities, were much richer in grasses. Photograph by P. Wigand.

and thus capable of exciting and stampeding large ungulates,thus increasing the impacts of the latter on the landscape.

Canis latrans (coyote)Canis lupus (gray wolf)

FE LI DAE (CATS)

Felis concolor (puma): Pumas in California today will consumepractically any other mammal, but in terms of kill frequen-cies they are deer specialists, and that adaptation makes sensefor the Pleistocene, given the relatively gracile skeleton(though powerfully muscled) of these extremely nimble cats.

Homotherium serum (scimitar cat): These lion-sized cats hadsmaller sabers but longer limbs than their better-known rel-ative Smilodon and were probably more cursorial. Evidencefrom Friesenhahn Cave in Texas suggests that young mam-moths were a favorite prey (Turner 1997).

Lynx rufus (bobcat): Like pumas, bobcats consume a widerange of prey. They are fully capable of killing adult deer andcommonly do so.

Miracinonyx trumani (American cheetah): Fossils of thesegracile cats have been found in Nevada. These are the rarestof fossils in a family that is rare in fossil form to begin with.It is reasonable to suspect that fossils of Miracinonyx will showup in California, especially because prey animals capable ofspeeds far surpassing any other cats, namely pronghorns,were present in the California Rancholabrean.

Panthera leo atrox (American lion): This was the largest felidof the Rancholabrean. Males were about 25% larger thanAfrican lions. Grayson (1991) speculates that these wereanimals of open country, since they are absent from the“forested east,” though the eastern United States was not asforested then as it was historically (Guthrie 1990).

Panthera onca ( jaguar): Jaguars originated in Eurasia orNorth America and later spread south across Panamania toSouth America. Jaguars persisted in California into historictime, and “roamed the South Coast Ranges between SanFrancisco and Monterey up to at least 1826” (Jameson andPeeters 1988). In South America their preferred prey includepeccaries and tapirs, and since both were represented inRancholabrean California, one may speculate that jaguarshunted similar species in the north.

Smilodon fatalis (sabrecat): This saber-toothed cat was aboutthe size of a female African lion. Stock and Harris (1992) sug-gest on the basis of the skeleton that these animals were lesscapable than other big cats of the time in chasing down preyand hence would have depended more upon stalking andambush. One might speculate, then, that they did not spe-cialize in the fleeter ungulates.

TAYASS U I DAE (PECCAR I E S)

Platygonus compressus (flat-headed peccary): This animal is alittle larger than the extant peccary of the desert Southwest,with longer limbs. Although Grayson (1991) considered thisan animal of open habitats, its dentition is low-crowned andadapted for browsing. On analogy with living peccaries,

Platygonus may have been an opportunistic feeder, browsingbut also rooting for bulbs and taking small animals andcarrion. According to Simpson (1980), Platygonus is question-ably distinct from Catagonus, the living Chacoan peccary ofSouth America that was only discovered in 1975. Catagonuslives in dry thorn forest and feeds on cacti, bromeliad roots,fruits, and forbs.

CAM E LI DAE (CAM E LS)

Camelops hesternus (western camel): This was a large camel,with limbs up to 25% larger than those of the modern drom-edary (Webb 1965). Its cheek dentition is higher-crowned thanthat of modern tule elk, mixed grazer-browsers that consumeabout 50% grass. Camelops cheek teeth often preserve cemen-tum, which gives extra support for heavy mastication. Wearprofiles (mesowear) of cheek teeth are consistent with amixed-feeding strategy. Dompierre and Churcher (1996)concluded on the basis of comparisons of snout shapes inungulates that Camelops was a mixed grazer-browser. NorthAfrican dromedaries are similarly mixed grazer-browsers.Akersten et al. (1988) examined dental boli impacted inmolars from Rancho La Brea and found that they containedBouteloua, Bromus, Festuca, Hilaria, and Sporobolus, thesegrasses collectively amounting to 10.7% of the identifiableremains. The rest of the identifiable remains were dicotyle-donous, but overall, 80% of the material in the boli wasunidentifiable as to monocot vs. dicot.

Hemiauchenia macrocephala (large-headed llama): A slender-limbed, long-legged llama with relatively high-crowned cheekteeth, this species has often been interpreted as a swift, open-country grazer. The native living camelids of the Andes, vicuñasand guanacos, are open-country animals with diets focused onperennial bunchgrasses and forbs. Hemiauchenia is in or closeto their ancestry (Webb 1965). Dompierre and Churcher (1996)interpreted Hemiauchenia as a browser based on premaxillarymorphology, while MacFadden and Cerling (1996) decidedthis llama was a grazer, based on carbon isotopes in dentalenamel. More recent assessment of carbon isotopes by Feranec(2003) led to the characterization of Hemiauchenia as an inter-mediate feeder with a preference for browse.

CE RVI DAE (DE E R AN D E LK)

Cervus elaphus (elk): Observations of tule elk at Pt. ReyesNational Seashore reveal that these large deer are mixedgrazer-browsers, consuming about 50% grasses and 50%other, mostly forbs (Gogan and Barrett 1995). Their cheekteeth are less high-crowned than those of cattle, and theylack supporting cementum.

Odocoileus hemionus (mule deer): Deer are essentiallybrowsing animals that take very little grass, when the foliageis young and tender.

ANTI LOCAPR I DAE (PRONG HOR N S)

Antilocapra americana (pronghorn): Historically, pronghornsare browsers, and this is an important point, because in

5 0 H I S T O RY

popular literature they are regularly pictured as grazers thatimpact grasses directly and substantially. In fact they prefershrubs and forbs; like deer, they consume grasses only spar-ingly and usually when the foliage is young (Yoakum 1980).

Capromeryx minor (dwarf pronghorn): This is a diminutivecreature, less than two feet (0.6 meter) tall at the shoulder.Anderson (1984: 76) reports it as a grazer, but this is unlikely.High surface-to-volume ratio would probably have necessi-tated a focus on higher protein values available in browse. Ifone can take Thompson’s gazelle, an animal of similar size, asan analogue, Capromeryx would have had no more interest ingrasses than Antilocapra does.

Bovidae (Cattle, Sheep, and Their Relatives)

Bison antiquus (Ice Age bison): These animals were morpho-logically very similar to living Bison bison and some author-ities prefer to classify antiquus as a subspecies. On the basisof plant tissues in dental boli from Rancho La Brea, Akerstenet al. (1988) considered this species a browser (only 13.4%monocot). However, the relatively wide muzzle is more char-acteristic of grazers; mesowear on cheek teeth is consistentwith a grazing emphasis; dung attributed to this animal fromCowboy Cave, Utah (Hansen 1980), is dominated by grasses;skeletal evidence suggests bison visitied La Brea only for onemonth in spring (Jefferson and Goldin 1989) when browsemay have been preferred locally; and isotopic analysesreported by Feranec (2004) and Feranec and MacFadden(2000) for Bison in Florida point to a diet ranging from graz-ing to mixed with an emphasis on grazing. The consensusamong those who have studied fossil Bison is that these werepredominantly grazing animals.

Bison latifrons (giant bison): The largest bison that everlived, with horn cores up to 7.5 feet (2.3 meters) across,became extinct perhaps as much as 12,000 years before thelast antiquus. McDonald (1981) studied the cranial morphol-ogy and concluded these were browser-grazers; that is, theypreferred browse but did some grazing. In terms of vegetationimpact, an analogy can be drawn with moose. McInnes et al.(1992) have shown that exclusion of these large ungulatescan lead to decline of herbaceous cover and invasion of clear-ings by shrubs and trees.

Euceratherium collinum (shrub ox): Some investigators haveclassified these animals, about elk- or cattle-sized, as grazers;but, according to Mead et al. (2003), dung pellets from sand-stone rock shelters in the Glen Canyon region of Arizonahave been attributed to Euceratherium and contain mostlybrowse species such as Quercus, Artemisia, and Chrysothamnus.Mesowear on cheek teeth is consistent with a mixed dietemphasizing browse.

Oreamnos americanus (mountain goat): Fossils of these ani-mals have been recovered only in the far north, in Lassen andShasta counties.

Ovis canadensis (bighorn sheep): Even in the Pleistocenethis species was probably focused in transmontaneCalifornia.

Symbos cavifrons (woodland musk ox): This species waswidespread, from Alaska to Texas, though so far in Californiait has been found only in Modoc County. Perhaps it is to beexpected at higher elevations in the mountains elsewhere.Symbos was bison-sized, but more slender. Its dietary adapta-tions have not been studied.

EQU I DAE (HORS E S)

Equus conversidens: This species was apparently restricted tothe desert counties east of the Transverse Ranges.

Equus cf. occidentalis (Western horse): According to Harrisand Jefferson (1985) the large sample at Rancho La Breaaffords a reconstruction about 4 .5 feet (1.4 meters) tall at theshoulder. The cheek teeth of these animals are as hypsodont asthose of any other mammals known. Although isotopic stud-ies suggest some browsing occurred, cheek teeth this hyp-sodont are intelligible only as adaptations for habitual grazingof grasses.

TAPI R I DAE (TAPI RS)

Tapirus (tapir): Jefferson (1989) reports two species of tapirin the Rancholabrean of California. Both occur along thesouth coast, but T. californicus also has been found at severallocalities in the central Sierra foothills, while T. merriami hasbeen recovered in Alameda and Contra Costa counties.Living tapirs are wetland/woodland/forest browsers, and thelow-crowned dentitions of the fossils resemble those ofliving species. According to Graham (2003), living tapirs ofthe New World are very selective browsers that show somepreference for colonizing plants that are low in toxic defensecompounds.

PROBOSCI DEA (E LE PHANTS AN D MASTODONTS)

Mammut americanum (American mastodon): These weremedium-sized elephant relatives, 6 to 9 feet (1.8–2.7 meters)high at the shoulder. Their bunodont molars, lacking sup-porting cementum, were adapted for browsing.

Mammuthus columbi (Columbian mammoth): As large asAfrican elephants, these massive animals had high-crownedcheek teeth with closely packed lamellae intermediate innumber between those of African and Indian elephants, bothof which are browser-grazers. Dung ascribed to mammothsfrom Bechan Cave in Utah (Haynes 1991) contained over95% by weight grasses, sedges, and rushes. Dung from Cow-boy Cave, also in Utah (Hansen 1980), contained more than95% grasses, mostly Sporobolus. At Grobot Grotto the dungcontained mostly Phragmites. These mammoths have tradi-tionally been interpreted as open-country animals. Theimmense size and length of tusks on some males, nearlydoubling the overall length of the animal, surely would havelimited their mobility in a forested environment. Judgingfrom behavior of modern African elephants, California mam-moths may have opened up vegetation by trampling andtree felling.

P L E I S T O C E N E A N D P R E - E U R O P E A N G R A S S L A N D S 5 1

Effects of Megafauna on California Grasslands

Grazing, browsing, and trampling are different but overlap-ping activities, and the habits of herbivorous mammals sortout on a continuum. Horses, rabbits, and elephants graze,browse, and trample with different foci and intensities. Thetrampling element is increased as animals are harried bypredators, and Rancholabrean California had one of the mostawesome arrays of predators known since the days of Tyran-nosaurus rex. Grazing, browsing, and trampling have all beendemonstrated by contemporary studies to contribute to pro-tection of grassland from encroachment by woody plants. Itis likely that such impacts by Rancholabrean megafaunahelped to maintain large areas of grassland even during glacialperiods when cooler, more mesic conditions favored forests.

The origins of California grasslands ultimately cannot befully understood without considering the fauna that evolvedwith them. Five million years and more of that historyinvolved a diverse megafauna, a fauna growing larger andmore diverse with time. Only 10,000 years involved theanomalously depleted faunal remnants of the Holocene.

Study of Pleistocene megafauna is no idle pursuit. Parkman(2002) has made a convincing case that brilliantly polishedsurfaces high on raised seastacks along the Sonoma coastwere mammoth rubbing stations. He has also suggested thatsome of California’s extant vernal pools may have originatedas wallowing basins of Rancholabrean herbivores, and thishypothesis should be investigated. As for the native plants ofCalifornia’s grasslands, it is likely that they preserve a sub-stantial genetic legacy of their relations with magnificentanimals that grazed, browsed, and trampled them not reallyso long ago.

Species Composition at the Time of FirstEuropean SettlementPaula M. Schiffman

A very basic question nags at ecologists and habitat man-agers: What was the species composition of California’s grass-lands like at the time of European contact? More specifically,which species were dominant? This question exists becausethe grasslands were colonized by several invasive plantspecies soon after European contact (Hendry 1931; Spira andWagner 1983; Sauer 1988; Blumler 1995; Mensing and Byrne1998, 1999), and these species rapidly became incorporatedinto natural landscapes. There are no descriptions of grass-land species composition from that early time period, and,amazingly, the invasion went unnoticed. When people finallybegan to record detailed vegetation accounts in the mid-1800s(e.g., Cronise 1868), invasive plant species were already geo-graphically widespread and ecologically dominant.

Early Records

The historical spatial extent of California’s grassland area wasenormous (5.29 million hectares; Barbour and Major 1988:3–10). That such a massive invasion could have occurred,

without anyone documenting it, is remarkable and perplex-ing. The Native Americans who lived in this ecosystem formillennia used oral communication to share information.When European diseases and brutality decimated theirpopulations (Preston 2002b), their in-depth knowledge ofhistorical grassland species compositions and communitydynamics was largely lost. The first European settlers were notnaturalists, and from the very start they tried to dominate,rather than describe, their vast new environment. Theysimply viewed it as land where opportunities for livestockgrazing and cultivation abounded. Moreover, California’sgrassland-covered plains and valleys were subtle landscapesthat, except in the spring when expanses of colorfulwildflowers appeared, were dry, stark places that lacked thevisual drama of California’s wave-crashed coastline or impos-ing mountain ranges. Perhaps the minutiae of grasslandspecies composition seemed trivial when juxtaposed withsuch huge, wide-open landscapes. Harrison (1982) has notedthat as a group, early explorers and settlers were unusuallyreticent to record their observations and impressions of theNorth American prairie landscapes that they encountered. Hedescribes this odd phenomenon as “verbal blindness.”

Although detailed ecological accounts do not exist, a fewearly observers did record general descriptions of Californiagrasslands. The writings of Juan Crespí, a Spanish priest whojourneyed from Baja California to San Francisco Bay in1769–1770 and then from San Diego to Monterey in 1770,were full of descriptions of places with “everything verygrass-grown” (Crespí 2001: 309). Spanish mission periodjournals of other early Europeans such as Francisco Garcés,Pedro Fages, Juan Bautista de Anza, Pedro Font, Josef JoaquinMoraga, Francisco Palou, George Vancouver, Georg vonLangsdorff, and others also commented on the productivepastoral environments that they encountered (Coues 1900;Priestley 1937; Bolton 1930, 1931, 1966; Paddison 1999).These observers’ accounts of the vegetation were extremelygeneral and it is clear that descriptions such as “good grass,”“much grass,” and “level and grassy” terrain were not usedin a strict taxonomic sense. Rather, they were general por-trayals of low green vegetation that could be exploited forlivestock grazing. In this context, simplifying a diverse assem-blage of species—which would have included manygraminoids, forbs, geophytes, and even subshrubs—as“grass” made sense. Even if some of the plants were not allgrasses, they grew alongside grasses and they were consumedby grazing livestock as well as herds of native elk and prong-horn antelope. References to California’s grass-covered land-scapes continued well into the 1800s, as did the botanicalimprecision. For example, in an 1847 observation, EdwinBryant, an American journalist, noted that “[T]he varieties ofgrass are greater than on the Atlantic side of the continent,and far more nutritious. I have seen seven different kinds ofclover, several of them in a dry state, depositing seed uponthe ground so abundant as to cover it, which is lapped up bythe cattle and horses and other animals” (Bryant 1848: 448).Native clovers (Trifolium spp.) and grasses co-occurred, and

5 2 H I S T O RY

Bryant lumped these two completely different and unrelatedtaxa together into a single functional group.

However, early observers were not so unobservant or naïveto think that grasslands consisted only of grasses. It is evidentfrom their journal accounts that native forbs were abundantin the grassland landscapes through which they passed. Forexample, on May 7, 1770, when traveling near the SantaYnez River in what is now Santa Barbara County, Crespídescribed “a great plenty of white, yellow, red, purple andblue blossoms: a great many yellow violets or gillyflowerssuch as are planted in gardens, a great deal of larkspur, agreat deal of prickly poppy in bloom, a great deal of sage inbloom; but seeing all the different sorts of colors together waswhat beautified the fields the most” (Crespí 2001: 711). Earlydescriptions like this one did not include nearly enoughinformation for us to reconstruct the species composition ofthese landscapes accurately today. Still, it is quite clear thatspring-flowering forbs were important, though ephemeral,ecosystem constituents. A little more than a century afterCrespí, naturalist John Muir’s writings included reminiscencesof great profusions of annual wildflowers in the mid-1800s.He wrote, “The Great Central Plain of California, during themonths of March, April, and May, was one smooth continu-ous bed of honeybloom, so marvelously rich that, in walkingfrom one end of it to the other, a distance of more than400 miles, your foot would press about a hundred flowers atevery step. Mints, Gilias, Nemophilas, Castillejas, and innu-merable Compositae were so crowded together that, hadninety-nine percent of them been taken away, the plainwould still have seemed to any but Californians extravagantlyflowery” (Muir 1894: 339). Although Muir mentioned severalannual taxa, his descriptions primarily conveyed a vividsense of biodiversity rather than an ecologically meaningfulaccounting of community composition. However, he didremark that “all of the ground was covered, not with grassand green leaves, but with radiant corollas” (Muir 1894: 342).

Clements’ Influence and Recent Interpretations

Because of their ephemeral nature, the ecological importanceof these annual and perennial forbs was not widely recog-nized. Frederic E. Clements’ (1934) relict analysis indicatedthat the perennial bunchgrass, Nassella pulchra (Stipa pulchraand S. setigera; Hamilton 1997a), had been the historicaldominant in California’s grasslands. He interpreted theprominence of N. pulchra in some relict grassland fragmentsas key. Clements’ reputation as a leading twentieth-centuryecologist led to the acceptance of his hypothesis amongCalifornia biologists (e.g., Piemeisel and Lawson 1937; Munzand Keck 1959; Burcham 1961; Heady 1988). However, therelatively mesic and periodically burned fragments that wereClements’ exemplars did not constitute a good representa-tion of the wide range of habitats that supported grasslandvegetation in California. In addition, as Hamilton (1997a)convincingly explains, the scientific basis for Clements’hypothesis was shaky because it relied upon little real data

and several erroneous assumptions. Nevertheless, relativelyrecent references that discuss California grassland compositionand ecology in detail still usually identify N. pulchra as thelikely historically dominant species (Heady 1988; Schoenherr1992; Holland and Keil 1995), and field studies, particularlythose focused on conservation and restoration, have contin-ued to give more attention to N. pulchra than to any othernative grassland species. However, it has also been suggestedthat several other perennial grasses (e.g., Poa secunda,Leymus triticoides, Melica spp., Muhlenbergia rigens) were his-torically more important community constituents in someenvironments (Keeley 1990; Heady et al. 1992; Holland andKeil 1995; Holstein 2001).

But what about the historical importance of forbs? Histor-ical accounts, though limited in ecological detail, did clearlypoint to an impressive diversity and cover of colorful springwildflowers. Even Clements recognized perennial forbs as“subdominants” and stated that “even more typical are thegreat masses of annuals, representing more than 50 generaand several hundred species” (Clements and Shelford 1939:288). In fact, his description of the springtime vegetation of1935 bore considerable resemblance to the much earlierdescriptions of Crespí and Muir: “the carpet of brilliant blues,oranges, and yellows covered an area approximately 50 mileswide and 100 miles long” (Clements and Shelford 1939: 288).Like other observers, Clements noted the abundance ofnative annuals and then glossed over their identities as ifthey were unimportant. Despite his clear acknowledgementof their tremendous percent cover, these plants’ transientnature indicated to him that they had little real ecologicalvalue. Clements’ endeavor to draw ecological linkages betweenCalifornia’s grasslands and those of the midwestern UnitedStates demanded that he emphasize perennials, especiallygrasses (Hamilton 1997a), despite the ubiquity of so manyannual forbs.

The ruderal nature of annual plants (Grime 1979a) wasanother feature of California’s native forbs that precludedClements from considering them to be ecologically impor-tant. By definition, he viewed climax communities as gen-erally stable associations of species that developed throughsuccession (Hamilton 1997a). So, although vegetation madeup of weedy, invasive, non-native annuals including Avena,Bromus, Hordeum, Festuca (Vulpia), and Erodium was consid-ered a “proclimax” community, a stable community domi-nated by an association of disturbance-adapted nativeannual plants completely violated his theoretical frameworkand, therefore, went unrecognized. Today, it is well knownthat native forbs repeatedly reappear on the same sites fordecades, though their covers vary with annual rainfallamounts. In addition, soil disturbances by small burrowingmammals, herbivory, periodic fires, and environmentalmanagement by Native people were integral ecosystemprocesses that had compositional consequences includingthe promotion of annuals (Blumler 1992; Hobbs andMooney 1995; Painter 1995; Schiffman 2000; Reichman andSeabloom 2002; Keeley 1990, 2002; Anderson 2005). Surely,

P L E I S T O C E N E A N D P R E - E U R O P E A N G R A S S L A N D S 5 3

TA

BL

E4

.1C

har

acte

rist

ics

of t

he

13 R

elic

t G

rass

lan

ds I

ncl

uded

in t

he

Ord

inat

ion

Perc

ent

Perc

ent

Mea

n ra

infa

llTo

tal n

ativ

epe

renn

ial

annu

alG

rass

land

Cod

e*C

ount

yLa

titu

de(c

m/y

r)A

rea

(ha)

spec

ies

gras

ses

forb

s

Hop

lan

d R

esea

rch

an

d Ex

ten

sion

Cen

ter

HO

PM

endo

cin

o39

o00

’94

.02,

165

396

8.6

43.7

C

osum

nes

Riv

er P

rese

rve

CO

SSa

cram

ento

38o25

’44

.216

,160

216

5.6

50.9

Jeps

on P

rair

ie R

eser

veJE

PSo

lan

o38

o15

’47

.563

422

84.

052

.2Po

int

Rey

es N

atio

nal

Sea

shor

eR

EYM

arin

38o05

’16

5.5

25,9

0748

39.

729

.8La

wre

nce

Liv

erm

ore

Nat

ion

al

Labo

rato

ry S

ite

300

LAW

Ala

med

a37

o42

’36

.82,

828

262

4.2

66.8

Jasp

er R

idge

Bio

logi

cal P

rese

rve

JAS

San

Mat

eo37

o24

’65

.248

143

96.

444

.0Sa

n J

oaqu

in E

xper

imen

tal R

ange

JOA

Mad

era

37o05

’48

.61,

806

248

3.6

65.7

Elkh

orn

Slo

ugh

Nat

ion

al E

stua

rin

e R

esea

rch

Res

erve

ELK

Mon

tere

y36

o48

’55

.256

614

79.

526

.5H

asti

ngs

Nat

ural

His

tory

Res

erva

tion

HA

SM

onte

rey

36o22

’53

.093

239

96.

057

.4C

arri

zo P

lain

Nat

ion

al M

onum

ent

CA

RSa

n L

uis

Obi

spo

35o10

’14

.510

1,00

034

62.

958

.6Se

dgw

ick

Res

erve

SED

San

ta B

arba

ra34

o45

’38

.02,

358

201

6.0

53.2

Star

r R

anch

San

ctua

ryST

AO

ran

ge33

o37

’38

.11,

616

266

4.5

41.0

San

ta R

osa

Plat

eau

Ecol

ogic

al R

eser

veR

OS

Riv

ersi

de33

o31

’48

.03,

434

395

6.3

48.4

NO

TE:T

he

gras

slan

ds a

re p

lott

ed b

y co

de in

Fig

ure

4.7.

P L E I S T O C E N E A N D P R E - E U R O P E A N G R A S S L A N D S 5 5

the endurance of native annual forbs in California’s grass-lands and their apparently adaptive interactions with otherorganisms and processes reflects their historical ecologicalsignificance.

In recent years, researchers have used evaluations of his-torical accounts, floristic surveys, relict analyses, and modernexperimental and comparative findings to propose alterna-tives to Clements’ vision of California’s grassland speciescomposition. Several of these reconstructions have suggestedthat annual plant species, rather than N. pulchra or otherperennial grasses, had been the most ecologically importantspecies in much of southern California and relatively aridinland environments including the Central Valley (Talbotet al. 1939; Twisselmann 1967; Wester 1981; Blumler 1995;Holstein 2000, 2001; Schiffman 2000, 2005). In more mesicareas, annual forbs still constituted a diverse group of plants.Sadly, it is now impossible to truly understand the ecologi-cal roles of individual plant species at the time of Europeancontact. Clues to the historical past have been blurred bymassive changes caused by the contamination of California’sgrasslands by invasive non-native annuals and a wide rangeof human activities including cultivation, livestock grazing,fire suppression, eradication of the grizzly bear (a keystonespecies), and habitat fragmentation. So, the degree to whichthe ecological dynamics in relict grasslands resemble those ofhistorical ecosystems is somewhat unclear. One thing is quitecertain, however. These habitats continue to support verylarge numbers of native species, particularly forbs, just asthey did when Europeans first encountered them.

A Relict Analysis

Relict grassland floras typically include hundreds of nativespecies in addition to grasses. Therefore, a study of relict flo-ras that focuses on grassland plants of all forms should yieldhistorically meaningful results. For example, this approachcan be used to estimate the degree to which the native speciescompositions of historical grasslands in California resembledeach other. Did regional differences in latitude, proximity tothe Pacific coast, and rain shadow–producing hills mean thatthe grasslands of the northern coast or Sacramento Deltabore little resemblance to those of the San Joaquin Valley orsouthern California? They undoubtedly had some species incommon, but how similar were these floras? Were they asmonolithic as much of the literature has implied?

To address these questions I have compared the nativefloras of 13 different relict grassland preserves in California(Table 4.1). Comprehensive plant species lists available foreach of the preserves were the data sources for the study. Theboundaries of these preserves encompass other vegetations inaddition to grasslands (e.g., wetlands, chaparral, oak wood-lands, riparian forests, and coniferous forests), and they fre-quently intergrade. Therefore, grasslands typically share somespecies with adjacent communities, and the communitiesthemselves can be difficult to differentiate. Because there isambiguity about the definition of “grassland,” my relict

analysis was limited to low-stature native plants that could beconsidered to be grassland species, at least in a broad sense(grasses, graminoids, annual, biennial, and perennial forbs,geophytes, and subshrubs as indicated by species descriptionsin Hickman 1993). Trees and shrubs were excluded from theanalysis, as were their parasites and nonwoody plants that,according to Hickman (1993), occur primarily in forests. Mul-tiple taxa differentiated below the species level (subspeciesand varieties) were also excluded from the analysis.

The analysis encompassed a remarkable number of plantspecies and indicated that California’s extant grasslands areextremely important reservoirs of biodiversity. A total of1,348 native grassland species occurred at the 13 sites sur-veyed. This means that these relict grasslands collectivelysupport about 40% of the state’s total native plant speciesrichness (Hickman 1993). Many of the species in this studyoccurred at only one or two of the sites, and most of thesespecies were annuals (Figure 4.6). Surprisingly, just 1% of thespecies were present in all of the study’s grasslands. This smallgroup of ubiquitous species consisted of a perennial herb(Achillea millefolium), 10 annual forbs (Amsinckia menziesii,Calandrinia ciliata, Claytonia perfoliata, Crassula connata,Eschscholzia californica, Lasthenia californica, Lotus wrangelianus,Lupinus bicolor, Mimulus guttatus, and Trifolium willdenovii), anannual graminoid (Juncus bufonius), and just one perennialgrass (Nassella pulchra). These findings strongly indicate that,historically, California’s grasslands were habitat for an enor-mous number of different plant species and that the vastmajority of them were not perennial grasses.

PC-ORD (MJM Software Design, Gleneden, OR) was used tocompute Jaccard distances (Magurran 1988) for the 13 grass-lands and to ordinate them in two-dimensional space(Figure 4.7). Separation of the grasslands along the horizontalaxis (axis 1) was strongly correlated with percentages ofannual forbs and perennial grasses as well as with meanannual precipitation (Table 4.2). Latitude was most highlycorrelated with the distribution of grasslands along the

F IG U R E 4.6. Frequency distribution of annual forbs, perennialgrasses, and other species in the 13 relict grasslands.

distributed as a generally latitudinal group with moderateproportions of annual forbs and perennial grasses. Finally, afloristically distinctive grassland type occurred in theSacramento Delta (Cosumnes River Preserve and JepsonPrairie). These northerly grasslands also had moderate levelsof annual forbs and perennial grasses.

Unfortunately, these relict grasslands now also includemany non-native plant species, and they no longer experi-ence the disturbance regimes of the pre-European settlementenvironment. So it is impossible to estimate the importance(e.g., percent cover) of particular native species at the time offirst European settlement. Moreover, historical percents coverof native plants, particularly annuals, would have varied withthe year-to-year variation in winter rainfall amounts andother environmental factors. Despite the information limi-tations caused by such realities, this study’s comparativeapproach to species presence/absence likely provides an accu-rate perspective on the historical species compositions ofCalifornia’s grasslands. If, however, composition is viewedmore narrowly (for example, in terms of the presence/absence of perennial grasses or the presence/absence ofinvasive non-native species), the relict grassland sites thatthis study found to be floristically different would seemmuch more homogeneous. It is clear that by fixating on a fewperennial grasses and invasive species, California biologistshave been distracted from what was actually an array of com-positionally diverse and regionally distinctive historicalgrasslands.

5 6 H I S T O RY

TABLE 4.2Pearson Correlation Coefficients for the

Two-dimensional Ordinationof 13 Relict California Grasslands

Correlation coefficients (r)

Variable Axis 1 Axis 2

Percent perennial grasses "0.918 0.198Percent annual forbs 0.905 "0.114Mean annual precipitation "0.670 0.218Latitude "0.233 0.819Area 0.372 "0.259Total number grassland species "0.138 "0.245

vertical axis (axis 2). These correlation relationships wereplotted as vectors (Figure 4.7).

Although Nassella pulchra did occur in all of the grasslandsincluded in this study, this very simple relict analysis ofspecies presence/absence data strongly suggested that, his-torically, grasslands located in different regions of Californiahad broadly differing species compositions. The ordinationshowed four geographically distinctive grassland groupings(Figure 4.7). San Joaquin Valley grasslands (represented byCarrizo Plain National Monument, San Joaquin Experimen-tal Range, and Lawrence Livermore National Laboratory Site300), were characterized by high proportions of annual forbsand relatively few perennial grasses. In contrast, the moremesic coastal prairies at Elkhorn Slough National EstuarineResearch Reserve and Point Reyes National Seashore had highpercentages of perennial grasses and fewer annual forbs.Latitude is associated with environmental and floristicgradients, and the grasslands of the southern, central, andnorthern coastal mountains (Starr Ranch Sanctuary,Sedgwick Reserve, Santa Rosa Plateau Ecological Reserve,Hastings Natural History Reservation, Jasper Ridge BiologicalPreserve, and Hopland Research and Extension Center) were

F IG U R E 4.7. Ordination of 13 relict California grasslands producedusing Jaccard distances. Each grassland site is indicated by a three-letter code (Table 4.1). Variables correlated with most of theseparation of grasslands along the two axes are plotted as vectors(percent annual forbs, percent perennial grasses, mean annualprecipitation, and latitude).