CALIFORNIA AGRICULTURE, NOVEMBER–DECEMBER 2001 47

Carefully timed burningcan control barb goatgrassJoseph M. DiTomaso ❏ Kerry L. Heise ❏ Guy B. Kyser

Adina M. Merenlender ❏ Robert J. Keiffer

Three species of goatgrass occur inCalifornia: jointed, ovate and barb

goatgrass. All three species are winterannual grasses introduced early in the20th century from Mediterranean Eu-rope and western Asia. They areclosely related to winter wheat (Triti-cum aestivum) and have been shown tohybridize with the cereal crop. In theWestern states, jointed goatgrass is themost widespread species within thegenus and is a serious problem for ce-real crops, particularly winter wheat.Unlike jointed goatgrass, ovate(Aegilops ovata) and barb goatgrass areinvasive primarily in disturbed andundisturbed grasslands and pastures.

Barb goatgrass (A. triuncialis) wasprobably first introduced to Californiaaround 1915. From early records, itwas reported to crowd out other valu-able range species, reduce forage qual-ity and quantity, and injure livestockwhen its barb awns (slender, bristle-like appendages with sharp “hooks”)became lodged in their noses, mouthsor eyes (Kennedy 1928). Once a grass-land became infested with barb

goatgrass, estimates indicated thatlivestock range capacity (the numberof cattle the acreage can support) wasreduced by 50% to 75% (Jacobsen1929). By the late 1920s it had spreadto thousands of acres, but the infesta-tions were local and restricted to twocounties, Calaveras and El Dorado(Talbot and Smith 1930). Despite itslimited distribution, state and countyofficials made an effort to eradicatebarb goatgrass. At that time, however,few options were available. Burningwas used as a control measure, butprescribed burns were generally con-ducted either too early, when con-trolled fires were not sufficiently hot,or too late, when seedheads were moreresistant to destruction. Consequently,burning as a control strategy was con-sidered unreliable unless it was com-bined with a previous mowing or oiltreatment (Talbot and Smith 1930).

The control efforts in the early partof the 20th century probably slowedthe spread of barb goatgrass, whichcan rapidly move through livestocktransfers and contaminate vehicles or

Barb goatgrass is a noxiousannual grass that is rapidlyinvading California’s grasslandecosystems. No effectivecontrol strategies for managingbarb goatgrass have beenavailable that do notsimultaneously injure othermore desirable grass andbroadleaf species. In our studyat the UC Hopland Researchand Extension Center, weconducted prescribed burningin late spring or early summerbefore barb goatgrass seedshad reached maturation. Oneyear of prescribed burning wasnot sufficient to control re-establishment the followingyear. However, 2 years ofcomplete burning gave effect-ive control of barb goatgrasswhile increasing nativeperennial grass cover andnative species richness, par-ticularly legumes. The successof the goatgrass control wasdirectly proportional to thecompleteness of the second-year burn.

Barb goatgrass, above, was introduced inCalifornia around 1915. This noxiousrangeland weed crowds out more valuablegrass species, reduces forage quality andcan cause mechanical injuries tolivestock. Right, researchers at the UCHopland Research and Extension Centertested the effectiveness of prescribedburns to control barb goatgrass.

48 CALIFORNIA AGRICULTURE, VOLUME 55, NUMBER 6

equipment. Nevertheless, by 1973 ithad expanded its range in the CentralValley and coastal foothills and wasreported to occur in about 13 counties.By 1995, it had spread to 21 Californiacounties (Peters et al. 1996). Today,barb goatgrass is a state-listed noxiousweed in California and has also beenlisted as a noxious weed in Oregon, al-though it has not been reported thereyet.

Despite 70 years of attempted man-agement of barb goatgrass, no new ef-fective control options were devel-oped. Although many new herbicidesare currently registered for use ingrasslands and pastures, none provideselective control of barb goatgrasswithout damaging other desirablegrasses (Peters et al. 1996). Burning re-mains potentially the most successfulstrategy, but little effort has beenmade to enhance its effectiveness withmore-specific timing recommenda-tions. Although not directed at barbgoatgrass control, other studies usingcarefully timed burns have provedsuccessful for native grassland restora-tion by reducing the cover of alien spe-cies while increasing the cover and di-versity of native species (Menke 1992;Meyer and Schiffman 1999).

Many invasive and undesirable an-nual grasses, including barb goatgrass,medusahead (Taeniatherum caput-medusae) and ripgut brome (Bromusdiandrus), have long barb awns thathelp facilitate their dispersal on cloth-ing or on the hair, fur or wool of ani-mals. To maximize the probability ofdispersal, seeds (caryopses) remain at-tached on the inflorescence (floweringstructures) much longer into the seasonthan most desirable annual grass andbroadleaf species. In medusahead,for example, Murphy and Lusk(1961) showed that seeds developedabout 20 days after soft brome(Bromus hordeaceus), a more desirableannual grass.

Exposed seeds remaining attachedto the culm (stem) are more vulnerableto grassland fires than are seeds on thesoil surface. Studies on jointedgoatgrass (A. cylindrica) demonstrated

that seeds directly exposed to flamefor more than 3 seconds did not germi-nate (Willis et al. 1988). By compari-son, the studies showed that jointedgoatgrass seed germinated followingexposure to air heated to 392°F fornearly 1 minute. Prescribed burns ingrasslands infested with yellowstarthistle (Centaurea solstitialis) pro-duced an average soil temperature of392°F, but the length of exposure tothis temperature was too short to dam-age most of the seeds on the soil sur-face (DiTomaso et al. 1999). To controlbarb goatgrass while favoring the es-tablishment of more desirable species,a prescribed burn must occur follow-ing seed dispersal and vegetative se-nescence (plant death) in desirablespecies, but before seed maturationand dispersal in barb goatgrass.

Site factors and fire season

In 1997 we initiated a barbgoatgrass control study using timelyprescribed burning at the UC HoplandResearch and Extension Center(HREC) in Mendocino County. In thisarea, barb goatgrass has expanded itsrange and threatens livestock grazingoperations and research activities onthe facility. The objectives of thisproject were to reduce the levels ofbarb goatgrass, improve forage qual-ity, increase total plant diversity on af-fected pastures and develop moreeffective control measures.

Two heavily infested barb goatgrasspastures were chosen for the study:Little Buck and South pastures. LittleBuck pasture is a mix of three soil se-ries: Climax, Laughlin and Sutherlin.Climax soils are fine-textured andmoderately deep, and they have devel-oped from partly metamorphosed ba-saltic rock. They are associated withthe medium-textured soils of theSutherlin and Laughlin series thathave developed in sandstone andshale. Climax series soils are associ-ated with good grass cover, while theSutherlin and Laughlin series are typi-cally associated with grasslands with asparse oak cover.

Two areas of the South pasture

From top to bottom: Barb goatgrassseedheads remain on the soil surface aftera prescribed burn; a native perennial grass(Hordeum brachyantherum) recoversrapidly; native forbs are prolific in thespring following the burn; 2 years later,barb goatgrass has been controlled on aHopland pasture (unburned check plot canbe seen in the background).

CALIFORNIA AGRICULTURE, NOVEMBER–DECEMBER 2001 49

were used in this study. The UpperSouth pasture is composed ofMontara series soils, which are de-rived from serpentine rock and areshallow, stony and sparsely grass-covered. The Lower South pasture iscomposed of soils of the Climax se-ries, similar to Little Buck pasture.

A 3.7-acre area of Little Buck pas-ture and a 15.6-acre area of South pas-ture (across both Upper and Lower)were burned in May 1997 and July1998. The relative humidity was 7% to57% (May 19, 1997) and 23% (July 9,1998) on burn days at Little Buck; and30% to 65% (May 7, 1997) and 23% to28% (July 9, 1998) at South pasture(Upper and Lower). A corresponding5.5 and 7.3 acres of Little Buck andSouth pastures, respectively, wereused as unburned controls. Timing ofthe burns corresponded to phenologyof the barb goatgrass population,which varied dramatically dependingon seasonal climatic conditions. Dur-ing the rainy season of 1996 and 1997,only 15% (17 cm; 6.7 inches) of the to-tal seasonal precipitation occurred af-ter January (fig. 1). In contrast, 65%(98 cm; 37.4 inches) of the seasonalrainfall occurred after January in the1997–1998 season. This late-seasonprecipitation slowed the development

of barb goatgrass and other desirablegrasses in 1998 compared with 1997,delaying the timing of the burn.

Five soil core samples (2 inches indiameter) were collected from eachsite in May 1997 and just after the burnin 1998. Analysis conducted at theHREC laboratory demonstrated no pHdifferences and little organic matterand nitrogen differences between thesites in 1997 before the burn and in1998 after a single year of prescribedburning (table 1). The only statisticaldifference between the South pasturesites was a higher percent of organicmatter and nitrogen in the lower burnarea compared with the unburnedcontrol site in 1997 prior to the burn.No significant year-to-year differenceswere measured in either the unburnedor burned areas.

Barb goatgrass control

We established five permanent100-foot line transects in each un-burned control and in the burned sitesof both pastures. We recorded our ob-servations of plant species interceptingeach of 50 points (at 2-foot intervals)along the transects. Vegetative cover(percent of groundcover) and speciesrichness (number of species inter-cepted) were calculated from the

point-intercept data. Point-intercepttransects were conducted in May orJune before the burns. We used thesame transects to determine the com-pleteness of the burns in the secondyear. This measure was defined as thepercent of the line transect that wasburned. For seed counts, two soil cores(2 inches in diameter and 2 inchesdeep) were randomly collected alongeach transect after the seeds dispersedin fall for a total of 10 samples pertreatment (1997 and 1999 only). Wecounted seeds following extractionwith a water/air elutriator, whichseparates seeds from soil to allow ac-curate evaluation of seed counts. Seed-ling counts were made in March orApril (1998 and 1999 only) by ran-domly tossing a 7.9-inch (20 centime-ter)-diameter ring within 3.28 feet(1 meter) of the permanent transects(5 samples per transect; 25 samplesper treatment). Both seed and seedlingdensities are presented as numbers persquare foot. We analyzed the data us-ing ANOVA followed by Fisher’s pro-tected LSD test. To estimate change inpercent frequency in individual spe-cies, four 2.7-square-foot quadrats

TABLE 1. Results of soil analysis before and after prescribed burnsfor barb goatgrass at HREC*

pH Organic matter (%) Nitrogen (%)

Pasture Treatment 1997 1998 1997 1998 1997 1998

South Unburned 7.06±0.21a 6.84±0.11a 12.3±1.5b 14.7±1.7a 0.11±0.07b 0.13±0.03aLower burn 7.18±0.23a 6.82±0.19a 14.3±1.3a 14.1±2.6a 0.18±0.03a 0.18±0.08aUpper burn 7.14±0.11a 6.72±0.13a 13.5±1.1ab 15.1±1.6a 0.11±0.05ab 0.16±0.07a

Little Unburned 6.16±0.49a 6.04±0.45a 8.9±3.0a 9.8±3.7a 0.18±0.04a 0.18±0.08aBuck Burned 6.62±0.47a 6.30±0.60a 11.4±1.4a 13.4±4.7a 0.19±0.04a 0.24±0.13a

* Mean values ± standard deviation. Within each site and year, values followed by the same letters are not different(P = 0.05). Means compared using ANOVA followed by Fisher’s protected LSD test.

Fig. 1. Precipitation at HREC from 1996to 1999.

A single year of burning did not provide effective barb goatgrass control, but 2 years ofcarefully timed burns worked well. Botanist Kerry Heise counts barb goatgrassseedlings at Hopland.

50 CALIFORNIA AGRICULTURE, VOLUME 55, NUMBER 6

(small rectangular plots) were ran-domly placed within 3.28 feet on ei-ther side of each transect for a total of20 quadrats per treatment. We re-corded each species and used percentfrequency to determine the change inspecies composition. Indicator speciesanalysis was used to statisticallyevaluate species frequency data ob-tained from the quadrat analysis. Indi-cator species analysis is a MonteCarlo–type nonparametric statisticaltechnique that can be used to deter-mine significance of species shifts be-tween treatments (Ludwig andReynolds 1988).

In May 1997, the dried litter layerfrom plants that had senesced the cur-rent or previous year in Little Buckand Upper and Lower South pastureprovided ample fuels for completeburns at all sites. A complete burn oc-curred at Little Buck pasture in July1998, but the fuel load was not greatenough at the Upper or Lower Southpastures for complete burns (table 2).

Soil seed counts were very highprior to the burn in 1997 (table 2),ranging from between 59 and 226seeds per square foot. In 1999, barbgoatgrass seedbanks were dramati-cally reduced after 2 consecutive yearsof prescribed burning. This corre-sponded to a significant reduction in

seedling counts. After 2 years of com-plete burning, no seedlings werefound in Little Buck pasture. Seedlingreductions in the two areas of Southpasture correlated exactly with thecompleteness of the second-year burn.Seedlings were reduced by 75% in thelower burn site and 98% in the upperburn site.

Vegetative cover analysis of matureplants demonstrated that a single yearof prescribed burning did not controlbarb goatgrass in the following year(fig. 2). In a similar study, Hopkinsonet al. (1999) conducted a prescribedburn for a single year in two sites nearEl Sobrante in Alameda County. Theirresults indicated no significant reduc-tion in one site and a 45% reduction inbarb goatgrass cover the followingyear in the second site. We concludethat a single year of control is not suf-ficient to provide long-term manage-ment of barb goatgrass. Although nofield studies have been conductedwith barb goatgrass, studies of theclosely related jointed goatgrass indi-cated that less than 8% of the seeds lefton the surface of undisturbed soil sur-vived more than 2 years (Donald1991). In another study, 90% or moreof the jointed goatgrass seeds in spike-lets located on the soil surface weredestroyed by burning wheat andjointed goatgrass stubble after thewheat crop was harvested (Young etal. 1990).

In the year following the secondburn of our study, control of barbgoatgrass closely corresponded to thecompleteness of the burn (fig. 3). Acomplete burn was achieved in Little

Fig. 2. Barb goatgrass vegetative cover inburned and unburned sites of Little Buckand South pastures in 1997, 1998 and1999. Lines above bars represent onestandard deviation from mean values.

Fig. 3. Percent of transect burned insecond year plotted against percent barbgoatgrass control along each transect.Percent control was for entire transect,including areas not completely burned.Data combined from all prescribed burntreatment areas. Line best fitted to linearregression with r2 = 0.87.

TABLE 2. Barb goatgrass seed and seedling densities, and percent of area burnedin second year of prescribed burn at HREC*

Seeds/ft2 Seedlings/ft2

1997 1999 (after 1998 (after 1999 (after Area burnedPasture Treatment (before burn) 2 yrs. burn) 1 yr. burn) 2 yrs. burn) second yr. (%)

South Unburned 137±209a 553±436a 60±44a 47±24a 0Lower burn 212±325a 58±55bc 21±14ab 12±14b 75Upper burn 226±211a 0c 10±4b 1±1bc 98

Little Buck Unburned 59±88a 175±270† 38±38a 22±23a 0Burned 146±87a 15±33† 6±5a 0b 100

* No data presented for seeds in 1998, and seedlings in 1997. Mean values ± standard deviation. Within each site andyear, values followed by the same letters are not different (P = 0.05). Means compared using ANOVA followed byFisher’s protected LSD test.

† Different at P = 0.1.

Buck pasture in the second year, andthis resulted in 100% control of barbgoatgrass in 1999 (fig. 2). In the LowerSouth pasture, where the burn waspatchy and calculated to be only 75%complete, barb goatgrass cover in 1999was reduced by only 46% comparedwith the unburned area. By compari-son, the second-year burn in the Up-per South pasture was 98% complete,and this correlated to an 88% reduc-tion in barb goatgrass. When asecond-year burn was complete(Little Buck pasture), no seed germi-nated in the subsequent season. Atthe Mendocino County site, barbgoatgrass seed longevity was greaterthan 1 year, but the lack of new seed-

CALIFORNIA AGRICULTURE, NOVEMBER–DECEMBER 2001 51

TABLE 3a. Vegetative cover (%) at HREC upper and lower South pasture, before and after prescribed burns*

Unburned Lower burn Upper burn

Vegetation type 1997 1998 1999 1997 1998 1999 1997 1998 1999

Grasses (introduced annuals) 36±15b 90±26a 89±30a 87±13a 85±20a 81±8a 78±13a 54±14ab 22±7b

(native perennials)† 0 1±2 2±3 0 0 0 0 0 0

Total grasses 43±24b 91±24a 91±27a 93±19a 85±20a 81±8a 82±14a 57±15ab 23±9b

Total plant cover 59±9c 103±23abc 104±24abc 109±22ab 146±21a 123±17ab 121±32ab 86±29bc 58±21c

* Mean values ± standard deviation. Across each row, values followed by the same letters are not different (P = 0.05). Means compared using ANOVA followed byFisher’s protected LSD test.

† No significant differences.

Fig. 4. Transect counts of native plantspecies in burned and unburned sites atLittle Buck and Upper South pastures in1997 (preburn), 1998 and 1999. Valuesrepresent average number of nativespecies contacted using point methodalong line transects (50 points pertransect). Lines above bars represent onestandard deviation from mean values.

seedbanks. In addition, fire has beenshown to be beneficial to the mainte-nance of perennial bunchgrass popula-tions and other native broadleaf spe-cies, particularly native legumes suchas clovers (Trifolium spp.), Gambel’slocoweed (Astragalus gambelianus) andWrangel’s trefoil (Lotus wrangelianus)(DiTomaso et al. 1999).

In general, the cover of introduced,native and total forbs did not changesignificantly over the 3-year period(data not shown). However, vegetativecover analysis indicated a consistentreduction in introduced annualgrasses following the second burn inboth Little Buck and Upper South pas-tures (tables 3a and 3b). As a result, to-tal plant cover was also lower in thesetwo sites compared with the adjacentunburned areas. The reduction in in-troduced grasses and total vegetativecover was primarily due to the de-crease in barb goatgrass. Very few na-tive perennial grasses were present ineither the unburned or burned areas ofthe South pasture.

When data from the unburned siteswere combined, the average numberof native plant species encountered inthe line transects gradually declinedfrom 1997 to 1999 (fig. 4). When the

data from the burned sites in UpperSouth and Little Buck pastures werecombined, the average number of na-tive species per line transect before theburn was slightly, but not significantlygreater (28%) than the unburned site.The native species values were signifi-cantly greater in the burned sites com-pared with the unburned areas afterboth the first (113% greater) and sec-ond (139% greater) year of prescribedburning (table 4).

From the indicator species analysis,the frequency of the introduced spe-cies barb goatgrass, hedgehogdogtailgrass (Cynosurus echinatus), andMalta starthistle or tocalote (Centaureamelitensis) declined with the pre-scribed burn regime, although in thecase of hedgehog dogtailgrass andMalta starthistle population differ-ences were not significant. Interest-ingly, soft brome significantly de-creased in both the burned andunburned sites at Upper South andLittle Buck pastures. By comparison,several native species increased sig-nificantly, including a native sandwort(Minuartia californica) and three nativelegume species (Trifolium bifidum, As-tragalus gambelianus and Lotushumistratus). Similar increases in le-

TABLE 3b. Vegetative cover (%) at HREC Little Buck pasture, before and after prescribed burns*

Unburned Burned

Vegetation type 1997 1998 1999 1997 1998 1999

Grasses (introduced annuals) 126±18a 152±36a 149±35a 115±27a 164±19a 58±23b

(native perennials)† 0 0 1±2 1±2 9±18 10±18

Total grasses 127±17ab 156±35ab 152±34ab 116±27bc 173±28a 68±11ca

Total plant cover 35±17bcd 202±46a 173±43a 126±28cd 198±39ab 79±15d

* Mean values ± standard deviation. Across each row, values followed by the same letters are not different (P = 0.05).Means compared using ANOVA followed by Fisher’s protected LSD test.

† No significant differences.

lings after a second year of controlsuggests that seeds do not survivemore than 2 years in the soil.

Vegetation changes

In California, suppression of peri-odic fire has dramatically changed thecomposition of rangelands. For nox-ious annuals such as barb goatgrass,medusahead, ripgut brome and yellowstarthistle, fire suppression can lead totheir dominance in grasslands. Reintro-duction of timed burns to rangelandscan remove litter, recycle nutrients,stimulate tillering (the emergence ofshoots from the base of the plant) inperennial grasses and reduce noxiousannual grass or broadleaf weed

52 CALIFORNIA AGRICULTURE, VOLUME 55, NUMBER 6

J.M. DiTomaso is Extension Weed Spe-cialist, UC Davis; K.L. Heise is Botanistand Staff Research Associate, UC Inte-grated Hardwood Range ManagementProgram; G.B. Kyser is Staff Research As-sociate, Department of Vegetable Crops,UC Davis; A.M. Merenlender is Exten-sion Specialist, Department of Environ-mental Science, Policy, and Management,UC Berkeley; and R.J. Keiffer is PrincipalSuperintendent of Agriculture, HREC.We thank Charles Vaughn for conductingthe soil analyses, the UC Integrated Hard-wood Range Management Program for itssupport and the HREC Research AdvisoryCommittee for appropriating necessaryland and staff support.

TABLE 4. Frequency (%) of common species in Little Buckand Upper South pastures in 1997, 1998 and 1999*

Unburned Burn

Species† 1997 1998 1999 1997 1998 1999Preburn 1 yr. burn 2 yr. burn

Little Buck pastureBarb goatgrass(Aegilops triuncialis) (I) 80 85 85 100a 95a 0b¶

Clover(Trifolium bifidum) (N) 5 5 0 15a‡ 25b 35b

Hedgehog dogtailgrass(Cynosurus echinatus) (I) 35 50 70 20 20 0

Meadow barley(Hordeum brachyantherum) (N) 0 0 0 5 10 20

Soft brome(Bromus hordeaceus) (I) 60a 55ab 10b¶ 100a 20a 5b¶

Subterranean clover(Trifolium subterraneum) (I) 10 25 0 45a§ 50b 55b

Upper South pastureBarb goatgrass(Aegilops triuncialis) (I) 100 100 100 100a 95a 25b¶

California sandwort(Minuartia californica) (N) 50 60 45 22a¶ 55b 65b

Gambel’s locoweed(Astragalus gambelianus) (N) 0 0 0 11a 40ab 70b†

Malta starthistle(Centaurea melitensis) (I) 25 65 50 45 55 5

Soft brome(Bromus hordeaceus) (I) 75a§ 45b 15b 78a¶ 10b 40b

Trefoil(Lotus humistratus) (N) 100 30 25 11a‡ 55b 65b

* Across each row within the burn or no-burn site, values followed by different letters are significantly different. Speciesfrequency values without letters for each treatment are not statistically significant (P > 0.2). Values compared usingindicator species analysis.

† I = introduced; N = native.

‡ Statistically significant at P = 0.20.

§ Statistically significant at P = 0.10.

¶ Statistically significant at P = 0.05.

Control maintained3 years later

Vegetative cover of barb goatgrassand other species in both LittleBuck and South pastures wasrevisited and evaluated in June2001, 3 years after the last burn inthis study. In Little Buck pasture,which burned completely in both1997 and 1998, barb goatgrassrepresents only 5% of the totalcover compared with 35% in theadjacent unburned site. Withoutfurther management efforts, thelevel of control is still 85%. Thenative perennial grass populationthat increased after the burn inLittle Buck pasture continues to beconsiderably higher (8% cover)than in the unburned area (1%). InLower South pasture, where thesecond-year burn was incomplete,barb goatgrass was nearly thesame (42% cover) as that in theadjacent unburned area (57%cover). In Upper South pasture,which burned more completely inthe second year, barb goatgrasscover in 2001 (26%) was about55% less than in the unburnedsite. These updated resultsindicate that two consecutiveyears of successful prescribedburning can provide continuedcontrol of barb goatgrass for atleast a couple of years aftertreatment. — J.D. et al.

gumes were also measured when pre-scribed burning was used to controlyellow starthistle (DiTomaso et al.1999).

Two good burns

Based on these results, two con-secutive years of prescribed burningcan be an effective tool for barbgoatgrass control. In addition, timely

prescribed burning can also enhancethe population of several native plantspecies, particularly perennial grassesand legumes. It is important to note,however, that unless two consecutiveyears of complete burning can beachieved at the appropriate times, thisinvasive annual grass will not be com-pletely controlled.

CALIFORNIA AGRICULTURE, NOVEMBER–DECEMBER 2001 53

Animals and fungi can affectgoatgrass establishment

The establish-ment of barb

goatgrass is greatlyaffected by its inter-actions with otherspecies in Califor-nia grassland. Forexample, goatgrassestablishment is en-hanced by a fungus, Ulocladiumatrum. This fungus is visible asblack lesions on the surface of theseedhead; it helps break down thetough, woody seedhead sur-rounding goatgrass seeds. Thisspeeds up germination and leadsto a 65% increase in goatgrass bio-mass at the end of the growingseason.

In contrast, small mammals helpto slow goatgrass invasion. In twoseparate experiments, more than3,000 individually markedgoatgrass seedheads were laid outin plots at the UC Hopland Re-search and Extension Center(HREC). Within days, new rodentholes had formed in the areaswhere the seeds were planted, andevery single seedhead was missing.It was not determined if this preda-tion was due to mice or voles.

Gophers have a dramatic effecton goatgrass establishment. Weestablished 10.76-square-foot plotsof eight different plant speciescommonly found in California an-nual grassland at HREC. Therewere 18 replicate plots of eachspecies randomly distributedthroughout the study area. Go-phers selectively built mounds ingoatgrass plots, completely bury-ing the goatgrass and killing mostof the seedlings. Gophers buriedevery goatgrass plot, while only a

Valerie T. Eviner

F. Stuart Chapin III

handful of other plots had signifi-cant levels of gopher disturbance.Goatgrass biomass was 3.5 timeshigher in gopher exclosures than inthe presence of gophers. A prelimi-nary survey of the pastures at HRECindicated that where gophers werepresent, goatgrasscoexists with other plant species,while in areas without gopher activ-ity, goatgrass formed a dense mono-culture stand.

While much more research isneeded to determine the relative im-portance of the fungus Ulocladium,mice and voles, gophers and compe-tition with plants on goatgrass es-tablishment, it is clear that theseinteractions can play a significantrole in the establishment of this in-vasive species.

V.T. Eviner is Ph.D. Candidate, De-partment of Integrative Biology, UCBerkeley; and F.S. Chapin III is Profes-sor, Institute of Arctic Biology, Univer-sity of Alaska, Fairbanks.

ReferencesDiTomaso JM, Kyser GB, Hastings MS.

1999. Prescribed burning for control of yel-low starthistle (Centaurea solstitialis) andenhanced native plant diversity. Weed Sci47(2):233–42.

Donald WW. 1991. Seed survival, germi-nation ability, and emergence of jointedgoatgrass (Aegilops cylindrica). Weed Sci39(2):210–6.

Hopkinson P, Fehmi JS, Bartolome JW,et al. 1999. Adaptive management and firecontrol of barb goatgrass. Res ManageNotes 17(3):168–9.

Jacobsen WC. 1929. Goatgrass — Aweed pest of the range. Monthly Bull, CalifDept Agr 18(1):37–41.

Kennedy PB. 1928. Goatgrass or wildwheat (Aegilops triuncialis). J Am Soc Agr20(12):1292–6.

Ludwig JA, Reynolds JF. 1988. StatisticalEcology. New York: J Wiley. 337 p.

Menke JW. 1992. Grazing and fire man-agement for native perennial grass restora-tion in California grasslands. Fremontia20(12):22–5.

Meyer MD, Schiffman PM. 1999. Fireseason and mulch reduction in a Californiagrassland: A comparison of restoration strat-egies. Madroño 46(1):25–7.

Murphy AH, Lusk WC. 1961. Timingmedusahead burns to destroy more seedand save good grasses. Cal Ag 15(11):6–7.

Peters A, Johnson DE, George MR.1996. Barb goatgrass: A threat to Californiarangelands. Rangelands 18(1):8–10.

Talbot MW, Smith LS. 1930. Thegoatgrass situation in California. MonthlyBull, Calif Dept Agr 19(1):40–6.

Willis BD, Evans JO, Dewey SA. 1988.Effects of temperature and flaming on ger-minability of jointed goatgrass (Aegilopscylindrica Host) seed. Proc West Soc WeedSci 41:49–55.

Young FL, Ogg AG Jr, Dotray PA. 1990.Effect of postharvest field burning on jointedgoatgrass (Aegilops cylindrica) germination.Weed Tech 4(1):123–7.

Controlling gophers on rangeland canhave an unintended side effect:dramatically increased establishmentof barb goatgrass. In experiments atHopland, gophers selectively disturbedgoatgrass plots, burying almost all theseedlings, left, while goatgrass plotsprotected from gophers have denseseedling establishment.

In addition to controlling noxious annualweeds, prescribed burns on rangeland canremove litter, recycle nutrients andstimulate the emergence of desirableperennial grasses.