early post-fire plant establishment on a mojave desert burn

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofitpublishers, academic institutions, research libraries, and research funders in the common goal of maximizing access tocritical research.

Early Post-Fire Plant Establishment on a Mojave Desert BurnAuthor(s): Scott R. Abella, E. Cayenne Engel, Christina L. Lund, and Jessica E.SpencerSource: Madroño, 56(3):137-148. 2009.Published By: California Botanical SocietyDOI: http://dx.doi.org/10.3120/0024-9637-56.3.137URL: http://www.bioone.org/doi/full/10.3120/0024-9637-56.3.137

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in thebiological, ecological, and environmental sciences. BioOne provides a sustainable onlineplatform for over 170 journals and books published by nonprofit societies, associations,museums, institutions, and presses.

Your use of this PDF, the BioOne Web site, and all posted and associated contentindicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.

Usage of BioOne content is strictly limited to personal, educational, and non-commercialuse. Commercial inquiries or rights and permissions requests should be directed to theindividual publisher as copyright holder.

http://dx.doi.org/10.3120/0024-9637-56.3.137

http://www.bioone.org/doi/full/10.3120/0024-9637-56.3.137

http://www.bioone.org

http://www.bioone.org/page/terms_of_use

http://www.bioone.org/page/terms_of_use

EARLY POST-FIRE PLANT ESTABLISHMENT ON A MOJAVE DESERT BURN

SCOTT R. ABELLA

School of Environmental and Public Affairs, University of Nevada Las Vegas,4505 S. Maryland Parkway, Las Vegas, NV 89154-4030

[email protected]

E. CAYENNE ENGEL

School of Environment and Public Affairs, University of Nevada Las Vegas,4505 S. Maryland Parkway, Las Vegas, NV 89154-4030

CHRISTINA L. LUND1

Bureau of Land Management, Las Vegas Field Office, 4701 N. Torrey Pines Drive,Las Vegas, NV 89130

JESSICA E. SPENCER2

Public Lands Institute, University of Nevada Las Vegas, 4505 S. Maryland Parkway,Las Vegas, NV 89154-2040

ABSTRACT

Fire has become more extensive in recent decades in southwestern United States arid lands. Burnedareas pose management challenges and opportunities, and increasing our understanding of post-fireplant colonization may assist management decision-making. We examined plant communities, soils,and soil seed banks two years after the 2005 Loop Fire, located in a creosote-blackbrush communityin Red Rock Canyon National Conservation Area in southern Nevada’s Mojave Desert. Based on aspring sampling of 20, 0.01-ha plots, live + dead cover of the exotic annual Bromus rubens averagednine times lower on the burn than on a paired unburned area. Perennial species composition shiftedfrom dominance by late-successional native shrubs (e.g., Coleogyne ramosissima) on the unburnedarea, to dominance by native perennial forbs (e.g., Sphaeralcea ambigua, Baileya multiradiata) on theburn. Species richness of live plants averaged 26% (100 m2 scale) and 239% (1 m2 scale) greater on theburn compared to the unburned area. Only 5% of Larrea tridentata individuals resprouted, comparedto 64% of Yucca schidigera and baccata. Fire and microsite (interspace, below L. tridentata, or belowYucca) interacted to affect several 0–5 cm soil properties, with higher pH, conductivity, and total Pand K on burned Yucca microsites. Bromus rubens density in 0–5 cm soil seed banks was four timeslower on the burn, and its distribution among microsites reversed. Below-shrub microsites containedthe most B. rubens seeds on the unburned area, but the least on the burned area. Intense fire belowshrubs may have increased seed mortality, an idea supported by .3-fold decreases we found inemergence density after heating seed bank samples to 100uC. Our study occurred after a post-fireperiod of below-average precipitation, underscoring a need for longer term monitoring thatcharacterizes moister years.

Key Words: Colonization window, exotic species, grass-fire cycle, soil seed bank, wildfire.

Wildfire has become widespread in southwesternUSA deserts. In a record 2005 fire season in theMojave Desert, for example, more than 385,000hectares burned (Brooks and Matchett 2006). Thisarea represents approximately 3% of the entireMojave Desert. Fueled in large part by exoticannual grasses, these fires burned desert shrub-lands thought to have only burned infrequentlyhistorically (Brooks 1999; Esque and Schwalbe2002). Burns now occupy significant portions ofdesert landscapes, posing prominent management

challenges. In addition to such concerns as post-fire soil erosion, burns pose challenges andopportunities for meeting mandates of landmanagement agencies to manage for native speciesand communities in areas such as national parksand conservation areas. Fires in arid lands havebeen cited to initiate a ‘‘grass-fire cycle,’’ where firepromotes resurgence of the exotic grasses thatfueled the fire to create a frequent-fire regime(D’Antonio and Vitousek 1992). Such a fire cyclethreatens the sustainability of many fire-suscepti-ble, infrequently regenerating native species, suchas Coleogyne ramosissima Torr. (Rosaceae) orYucca brevifolia Engelm. (Agavaceae). However,just as in more mesic regions, deserts contain earlysuccessional native species that might benefit fromfire (Cave and Patten 1984).

1 Present address: Bureau of Land Management,California State Office, 2800 Cottage Way, Sacramento,CA 95825.

2 Present address: U.S. Army Corps of Engineers, 701San Marco Blvd., Jacksonville, FL 32207.

MADRONO, Vol. 56, No. 3, pp. 137–148, 2009

Improving our understanding of plant coloni-zation on desert burns is important for evaluatingfuture fire hazard, whether natural colonizationwill meet management objectives for burned sites,or for planning active revegetation if this becomesa management goal. It is possible that opportu-nities exist for intervention in the grass-fire cycleimmediately following a fire if exotic grasscompetition is temporarily reduced while avail-able nutrients liberated by the fire increase. Thistime period may serve as a colonization window,which is often a key feature of plant succession inother regions. For example, periodic droughtsprovided windows for colonization by underrep-resented species during a 40-year permanent plotstudy of succession in New Jersey (Bartha et al.2003). The post-fire environment, together withthe frequent dry years characterizing desertclimates, may permit colonization by species notdominant prior to fires (Lei 2001; Brooks 2002).

In addition to precipitation, two of the manyfactors that may influence plant colonization aresoil seed banks and post-fire soil properties. Seedbanks can provide propagules of both native andexotic species if emergence requirements are metfollowing disturbance (Warr et al. 1993). Seedbank contributions to plant colonization afterwildfires hinge upon the density of seeds surviv-ing fire, their germinability, and post-fire seeddeposition and dispersal (Lei 2001). Soil pH oftenincreases after fire through combustion of organicacids or release of base cations. Correspondingly,availabilities of some soil nutrients also frequentlyincrease initially post-fire (Raison 1979). On onehand, pulses of nutrient availability, followingprecipitation events, are considered important tothe functioning of native desert plant communities(Noy-Meir 1973). On the other hand, exoticspecies often respond with vigorous growth tonutrient additions (James et al. 2006).

We measured plant communities, soil seedbanks, and soil properties during the second yearafter the 2005 Loop Fire in the eastern MojaveDesert. This burn occurred within a NationalConservation Area and encompasses a populartourist and recreation site. The fire burned withinand spread across a scenic loop road that receives900,000 visitors annually (Bureau of LandManagement, Las Vegas Field Office, Las Vegas,NV). The area is regarded as a symbol of the‘‘natural’’ desert landscape surrounding metro-politan Las Vegas, Nevada. Resource managersare concerned about post-fire succession inrelation to aesthetics, native community recovery,future fuel loads, and if or how active revegeta-tion should be a management goal. Our objec-tives were to (1) compare plant species richness,composition, and soil properties between theburn and a matched unburned area, (2) assessresprouting of burned dominant native shrubs,and (3) assay soil seed bank composition and

experimentally test responses of seed banksamples to the fire cues of heat and smoke.

METHODS

Study Site

This study occurred inside the Loop Drive areaof Red Rock Canyon National ConservationArea, managed by the Bureau of Land Manage-ment and located 15 km west of the western LasVegas suburbs in Clark Co., southern Nevada.Precipitation at the Spring Mountain Ranch StatePark weather station, 7 km south of the study site,has averaged 30 cm/yr (range 5 6–61 cm/yr,coefficient of variation [CV] 5 45%) during a1977–2008 period of record (Western RegionalClimate Center, Reno, NV). January daily mini-mum temperatures have averaged 21uC (range 526 to 2uC, CV 5 12%), and July maximumtemperatures have averaged 36uC (range 5 33–38uC, CV 5 2%). We studied the 348-ha LoopFire, which started from a lightning ignition onJuly 22, 2005. Maximum temperature on this daywas 38uC, and erratic winds gusted up to 80 km/hrduring the fire. Typical flame heights were $2 mand flame lengths were $6 m (Troy Phelps,Bureau of Land Management, personal commu-nication). Fire managers also noted that the fuelprovided by exotic annual grasses facilitatedignition of native shrubs and allowed the fire toburn continuously across the landscape. The fireleft fewer than five unburned islands, all ,1 ha insize, within the burn perimeter.

We matched an unburned area immediatelyadjacent to the west of the burned area that wassimilar in climate, soils, and vegetation. Based onplot sampling described below, mean elevation ofthe burned area (1274 m) differed by 54 m fromthe mean elevation of the unburned area (1220 m).Soils on the burned and unburned areas wereboth mapped in the county soil survey as .90% ofthe 731 (Purob-Irongold association) and 732(Purob extremely gravelly loam) types (Lato2006). These soils are classified as Typic and CalcicPetrocalcids, have alluvium parent material de-rived from limestone, and occur on fan piedmontlandforms. Based on identifying resprouts andburned shrub skeletons on the burned area, boththe burned and unburned areas were similarlydominated by Coleogyne ramosissima, Yuccaschidigera Roezl ex Ortgies (Agavaceae), Yuccabaccata Torr. (Agavaceae), and Larrea tridentata(DC.) Coville (Zygophyllaceae).

Plot Sampling

We overlaid a square grid of 315 ha encom-passing 91% of the irregularly shaped burn andrandomly located ten 10 m 3 10 m (0.01 ha) plotswithin the grid using randomly selected coordi-

138 MADRONO [Vol. 56

nates in a geographic information system. Welocated 10 plots in the adjacent unburned areausing the same method but in a square grid of130 ha that the site would accommodate. Eachplot contained six, 1 m 3 1 m subplots located atthe four plot corners and centered at 5 m alongthe southern and northern plot boundaries. Wevisually categorized areal percent cover of eachlive perennial species and of live and dead annualspecies rooted in each subplot using Peet et al.’s(1998) cover classes: 1 5 trace (assigned 0.1%), 25 0–1%, 3 5 1–2%, 4 5 2–5%, 5 5 5–10%, 6 510–25%, 7 5 25–50%, 8 5 50–75%, 9 5 75–95%,and 10 5 95–100%. We measured both live anddead annuals to provide a measure of the recentannual communities on plots. Dead annuals(especially the exotic grasses) also provide thefuel that allows fire to spread between perennialplants. We quantified measurement reproducibil-ity in subplot data by remeasuring a total of threerandomly selected subplots on three differentplots. Species richness/m2 was identical in originaland repeated measurements. The maximumdifference in cover class between original andrepeated measurements was one class. We inven-toried whole plots for species not alreadyoccurring in subplots and categorized cover ofthese species on a whole-plot basis. Plants notable to be readily identified to species in the fieldwere collected, pressed, and identified to specieswhen possible. Six vegetative specimens, whichoccurred on only one plot each, could not beidentified to species or genus and were deletedfrom the data set. Nomenclature, classification ofspecies life forms (e.g., shrub, forb), longevity(e.g., annual, perennial), and native/exotic statusin North America followed the PLANTS data-base (USDA 2007).

We counted the number of individuals of eachshrub species on each plot. On a 10 m 3 40 m(0.04 ha) plot originating at each 10 m 3 10 mplot, we further counted the number of Larreatridentata, Yucca baccata, and Yucca schidigeraindividuals and classified them into one of fourcategories: unburned alive, burned but crownsurvived, resprout, or dead (Rogers and Steele1980).

We collected soil and soil seed bank samplesfrom three interspaces (.1 m from any shrub)and from below the canopies of three each of thelargest Larrea tridentata and Yucca spp. (Y.baccata or Y. schidigera) adjacent to each plot. Inthe shrub microsites, we collected a 200-cm3 soiland a 200-cm3 seed bank sample (cores 7 cm indiameter) halfway between the main central stemand the outer canopy edge on both the north andsouth sides of each shrub. Samples were collectedfrom a 0–5 cm depth, which could include a litterlayer (only the below-shrub microsites containedappreciable litter) only for the seed bank samples.We included litter as part of the 0–5 cm seed bank

sampling depth because litter can trap seeds(Warr et al. 1993). We chose this depth for bothsoil and seed bank sampling because upper soilsbecome hottest during fire and may morestrongly reflect post-fire differences than deeperlayers (Raison 1979; Patten and Cave 1984). Thisdepth also includes the highest density of viableseeds in desert seed banks and is a maximumdepth from which most seeds can emerge or reachemergence depth (Guo et al. 1998). We compos-ited soil and seed bank samples on a plot basis foreach of the three microsite types. This resulted ina soil volume of 1200 cm3 each for soil and seedbanks for each microsite type on each plot. Wealso collected separate 200-cm3 samples fromeach microsite for measuring soil bulk density (2-mm sieving followed by drying at 105uC for24 hr).

Sampling occurred in 2007 (two years post-fire)in spring (April-May), the time of peak annualplant community growth in the Mojave Desert(Beatley 1966). We did not collect soil seed banksamples during what is considered the optimaltime (fall or early winter just prior to germinationof winter annuals) in this region. However, asonly 6 cm of precipitation fell (as opposed to thelong-term average of 23 cm) from September2006 through March 2007 prior to sampling,emergence of annual plants was limited. There-fore, depletion of soil seed banks through fieldgermination should not have been a major factoraffecting the detection of seeds in the greenhousesoil seed bank assay.

Soil Laboratory Analysis

The air dried, ,2-mm fraction of soil sampleswas analyzed for texture (hydrometer method),electrical conductivity and pH (1:1 soil:water),total C and N (Leco C/N analyzer), andextractable P and K (Olsen NaHCO3 method)following methods in Amacher et al. (2003).Measurement error, based on analyzing a dupli-cate sample every 10 samples and comparingoriginal and duplicate means, ranged from 0.2%(electrical conductivity) to 6.8% (total N).

Soil Seed Bank Experiment

We assessed responses of seed bank samples tosimulated fire exposure in a factorial experimentconsisting of heat (two levels: presence or absenceof 100uC heating) and liquid smoke (two levels:presence or absence of 10% smoke). Treatmentswere performed on 240 cm3 soil volumes extract-ed from microsite/plot composite samples. Weconducted heating treatments by placing samplesin tins in an electric oven. Samples remained inthe oven until the soil reached 100uC for oneminute, which required about 10 min, at whichtime samples were removed from the oven. We

2009] ABELLA ET AL.: POST-FIRE PLANT ESTABLISHMENT 139

monitored temperature using a probe accurate to0.1uC and placed directly in the soil. The soilcooled to 65uC within five minutes of beingremoved from the oven and to 45uC within10 min. We chose the 100uC treatment to bewithin the temperature range previously reportedto occur in upper soil layers during burns in theSonoran and Mojave Deserts (Patten and Cave1984; Brooks 2002). For smoke only samples andafter heating for heat 3 smoke samples, weapplied 120 ml (diluted to 10% smoke by volumeusing tap water) of commercially available liquidsmoke (Wright’s Brand, Roseland, NJ) to eachsample. Liquid smoke contains the same butanolcompound contained in air smoke that can affectgermination in some species (Flematti et al.2004). The non-smoke samples received 120 mlof tap water not containing liquid smoke.

After applying treatments, we placed eachsample in a 2 cm thick layer on top of 300 cm3

of sterile potting soil in 700-cm3 square plasticpots. We randomly arranged these pots, includingsix pots containing only potting soil to check forseed contamination (none was detected), on abench in a greenhouse. This greenhouse did notreceive supplemental lighting and was maintainedat approximately 28uC during the first threemonths (summer) of the emergence period and17uC during the last three months (fall). Sampleswere watered by an automated misting systemthat delivered approximately 1.5 cm of water/day.We conducted periodic (at least monthly) inven-tories during a six-month emergence period; 93%of the total number of emerging seedlingsemerged within the first 10 d. We also checkedsamples every 1–2 d to monitor for potential massmortality between inventories, which we did notdetect.

Data Analysis

Statistical inference is limited to the particularfire that we studied (van Mantgem et al. 2001).We compared live, dead, and total mean cover ofBromus rubens L. (Poaceae), a fuel-producingexotic annual grass, and community speciesrichness (per m2 and per 100 m2) between theburned and unburned area using two-tailed t-tests. Findings can change based on the scale ofanalysis for species richness, which is why weexamined effects of fire at two different scales.Based on species importance values for livevegetation for each plot, we ordinated speciescomposition using non-metric multidimensionalscaling (Sørensen distance, thorough mode) inPC-ORD (McCune and Mefford 1999). Wecalculated importance values as the average ofrelative frequency (based on six subplots per plot)and relative cover (with percent cover calculatedas the midpoint of a cover class). We tested thehypothesis of no difference in live plant species

composition between burned and unburned areasusing multi-response permutation procedures(Zimmerman et al. 1985). We based this analysison Sørensen distance and performed it in PC-ORD. To examine the fidelity of live species withburn status, we performed indicator speciesanalysis (Dufrene and Legendre 1997) usingspecies importance values and a Monte Carlotest of significance (1000 permutations). Indicatorspecies analysis combines the relative abundanceand frequency of a species within a group toproduce an indicator value that ranges from zero(no fidelity) to 100 (maximum fidelity). Weanalyzed univariate soils data as a split-plotanalysis of variance (ANOVA) with burn statusas the whole-plot factor and microsite (inter-space, below Larrea, or below Yucca) as thesubplot factor using JMP (SAS Institute 2004).Tukey’s test was used for mean separation.Bromus rubens comprised 92% of seeds in theseed bank experiment, so we focused the statis-tical analysis on this species. Log10 transformedB. rubens seed counts were modeled using amixed-model ANOVA (SAS PROC MIXED;REML) with burn status, microsite, heat, smoke,and all two-, three-, and four-way interactions asfixed effects (SAS Institute 1999). This was apartially nested design, so plot nested within burnstatus and its two- and three-way interactionswith microsite, heat, and smoke were treated asrandom effects.

RESULTS

Relative to longer lived species, live annualcover was low in the unburned area and both liveand dead annual cover was low in the burnedarea. Live annuals averaged only 0.05% (SD 50.09%) cover in the unburned area (0.4% of the15% total mean live plant cover), and 0.6% (SD5 0.6%) in the burned area (27% of the 2% totalmean live plant cover). Total mean cover ofBromus rubens was nine times lower in the burnedthan in the unburned area, a difference driven bysignificantly greater (t 5 25.32, P , 0.001) deadcover in the unburned area (Fig. 1). Althoughlive cover was sharply lower than dead cover inboth areas, live B. rubens cover was significantlygreater (t 5 3.09, P 5 0.006) in the burned arearelative to the unburned area.

Species richness of live plants was 3.4 timesgreater at a 1-m2 scale and 26% greater at a 100-m2 scale in the burned area compared to theunburned area (Fig. 2). Exotic annuals in theburn comprised 50% of richness/m2, but thispercentage declined to 18% of richness/100m2.Only 7% of richness at both scales in both theburned and unburned areas consisted of nativeannuals. Native perennial forbs constituted 34–35% of richness at the two scales on the burn,compared to 2–13% on the unburned area.


Conversely, native perennial shrubs composed76–79% of richness in the unburned area andonly 5–30% in the burned area.

Species composition differed sharply between theburned and unburned areas (Fig. 3). Importance ofColeogyne ramosissima and Yucca baccata waspositively correlated with unburned plots in theordination. In contrast, species associated withburned plots included the native perennial forbsSphaeralcea ambigua A. Gray (Malvaceae), Erio-gonum inflatum Torr. & Frem. (Polygonaceae),and Baileya multiradiata Harv. & A. Gray ex A.Gray (Asteraceae), the perennial grass Dasyochloapulchella (Kunth) Willd. ex Rydb. (Poaceae), andthe exotic annuals Erodium cicutarium (L.) L’Her.ex Aiton (Geraniaceae) and Bromus rubens.Consistent with the ordination results, averageSørensen similarities among burned (53%) andunburned plots (47%) were about four timesgreater than the average similarity between burnedand unburned plots (12%). This difference inspecies composition between the burned andunburned areas was significant based on multi-response permutation procedures (T statistic 5212.1, chance-corrected within-group agreementA statistic 5 0.27, P , 0.001).

Distributions of individual species also reflect-ed differences between burned and unburnedspecies composition (Fig. 4). Twelve speciesconstituted 71–78% of all relative cover. Basedon indicator species analysis, six of these specieswere significant (P , 0.05) indicators for the burnand three for the unburned area.

Resprouting varied between measured shrubspecies, with only 5% of burned Larrea tridentata

resprouting compared to 64% of burned Yucca(baccata and schidigera; Fig. 5). No measured L.tridentata and only 3% of Yucca in the burnedarea had crowns that survived the fire. Althoughwe did not measure dead Coleogyne ramosissimain the burned area, we found live individuals(three mature individuals presumably avoided bythe fire) in only one plot on the burned area andno resprouts. In comparison, live C. ramosissimadensity averaged 4,560/ha (SD 5 3,041) in theunburned area.

Several soil properties differed significantlyamong microsites and between the burned andunburned areas (Table 1). There was a burn 3microsite interaction for four properties (pH,conductivity, P, and K) because the Yuccamicrosite contained higher levels in the burnedthan in the unburned area compared to the othermicrosites. Total C varied both with burn statusand microsite, being greater on the burn. Total Nwas higher in Yucca microsites than in interspacesbut did not differ overall between burned andunburned areas. Soil texture did not differsignificantly among microsites or between burnedand unburned areas.

In the seed bank experiment, seed density ofBromus rubens reversed among microsites withburn status, resulting in a significant burn 3microsite interaction (F2,36 5 47.6, P , 0.001).Seed density was greatest below shrubs on theunburned area, but interspaces contained themost seeds on the burn. Heat also was asignificant effect (F1,18 5 101, P , 0.001) thatsharply reduced the density of emerging seeds.Therefore, results for the three-way interaction of

FIG. 1. Areal cover of Bromus rubens on a burned and unburned area, Red Rock Canyon National ConservationArea, Mojave Desert. Error bars are one standard deviation for total cover. Total mean cover was comparedbetween the burned and unburned areas using a two-tailed t-test.


burn status 3 microsite 3 heat (F2,36 5 11.5, P ,0.001) are presented in Fig. 6.

DISCUSSION

Study Assumption and Context

An assumption of this study is that differencesbetween burned and unburned areas resultprimarily from the fire and not from pre-existingvegetation or environmental differences. Theadjacent burned and unburned areas demonstrat-ed a priori similarity in mapped soil type (Lato2006) and dominant vegetation type. Plot sam-

pling corroborated this similarity, with meanelevation differing by only 54 m, soil texturenearly identical differing by only 1–2% in sand,silt, or clay (Table 1), and similar dominantshrub composition. Pre-existing Larrea tridentatadensity was probably higher in the burned area(Fig. 5). Because of this, we did not statisticallycompare post-fire L. tridentata density betweenburned and unburned areas, as instead wefocused on the proportion of burned individualsthat resprouted.

Annual precipitation was 134% of average in2004 preceding the fire and 169% of average inthe fire year of 2005. These high-precipitation

FIG. 2. Plant species richness on a burned and unburned area, Red Rock Canyon National Conservation Area,Mojave Desert. Error bars are one standard deviation for total mean richness. Total mean richness was comparedbetween the burned and unburned areas using a two-tailed t-test.


years resulted in accumulation of annual plantbiomass flammable after senescence (Brown andMinnich 1986). Precipitation has been belowaverage since the fire, averaging 65% of averagein 2006 and only 14% of the January-May meanin 2007 up to the sampling time for this study(Western Regional Climate Center, Reno, NV).Annual plant production has been negligiblesince the fire, and our results may be differenthad different weather patterns occurred after thefire. This observation underscores a need forlonger term monitoring, particularly given thehigh interannual variability characterizing Mo-jave Desert climates (Beatley 1966).

Bromus rubens

Total mean cover (live + dead) of Bromusrubens was sharply lower on the burned than theunburned area (Fig. 1). Furthermore, B. rubensdensity in the soil seed bank of the burn was only28% of the density on the unburned area (Fig. 6).

Seed density also reversed among microsites, withgreater density in interspaces in the burn butgreater density below shrubs in the unburnedarea. Below-shrub microsites, where B. rubens ismost abundant aboveground in unburned areas,likely burned more intensively due to greater fuelaccumulation than in interspaces (Brooks 2002).This intense burning probably killed seeds, anidea supported by results of our experimental100uC heating that sharply reduced emergencefrom samples. It is important, however, todetermine temperature thresholds for seed mor-tality of B. rubens in comparison with otherspecies. If B. rubens has a higher or lowerthreshold, this could influence its ability toexclude other species after fire. We detected 12species other than B. rubens in seed bank samples,but the low abundance of these species (only 8%of the total seeds detected) precluded makingthese comparisons.

Our finding that Bromus rubens was reducedafter fire concurs with Brooks (2002), who found

FIG. 3. Non-metric multidimensional scaling ordination of plant species composition on a burned and unburnedarea, Red Rock Canyon National Conservation Area, Mojave Desert. The lengths of vectors are proportional tocorrelations with ordination axes, and only vectors with r2 values $ 0.3 are shown. Vectors for species areabbreviated as the first three letters of the genus and species: BAIMUL 5 Baileya multiradiata, BRORUB 5Bromus rubens, COLRAM 5 Coleogyne ramosissima, DASPUL 5 Dasyochloa pulchella, EPHNEV 5 Ephedranevadensis, ERIINF 5 Eriogonum inflatum, EROCIC 5 Erodium cicutarium, KRAERE 5 Krameria erecta,SPHAMB 5 Sphaeralcea ambigua, and YUCBAC 5 Yucca baccata. Vectors for soils are shown as microsite(int 5 interspace, lt 5 Larrea tridentata, yuc 5 Yucca spp.) and variable (C 5 total carbon, EC 5 electricalconductivity, K 5 potassium, and P 5 phosphorus).


that B. rubens biomass declined after prescribedfires at three Mojave Desert sites. In that study,B. rubens was significantly decreased for fouryears after fire in shrub microsites and for twoyears in interspaces. Other studies have reportedvarying patterns of B. rubens abundance afterfire. For example, no clear trend in cover of thisspecies was apparent in a 1–37 yr time-since-firechronosequence in the northeastern MojaveDesert (Callison et al. 1985). Similarly, Caveand Patten (1984) found that post-fire B. rubensabundance varied between wildfire and pre-scribed fire in the upper Sonoran Desert. AsBrooks and Matchett (2003) note, longer termpatterns of post-fire B. rubens abundance canhinge upon climate, disturbance (e.g., grazing),

and other factors. It is unclear how much time isneeded after wildfire for B. rubens to return topre-fire levels, or which combinations of factors(e.g., native community composition, soil prop-erties) could result in higher or lower post-fire B.rubens abundance.

Native Perennial Species Composition

Although resprouts of perennial shrubs cur-rently contribute little to plant cover on the burn,this contribution may increase over time asexisting resprouts grow, particularly for Yuccabaccata and schidigera. Our finding that thesetwo Yucca species vigorously resprout concurswith Minnich’s (1995) findings in Joshua Tree

FIG. 4. Mean relative cover and indicator values (top of bars and only given for species that were significantindicators at P , 0.05) of the 12 most dominant species on a burned and unburned area, Red Rock CanyonNational Conservation Area, Mojave Desert. ASTSPP 5 Astragalus spp., BAIMUL 5 Baileya multiradiata,BRORUB 5 Bromus rubens, COLRAM 5 Coleogyne ramosissima, DASPUL 5 Dasyochloa pulchella, ENCVIR 5Encelia virginensis, EPHNEV 5 Ephedra nevadensis, ERIINF 5 Eriogonum inflatum, EROCIC 5 Erodiumcicutarium, SPHAMB 5 Sphaeralcea ambigua, YUCBAC 5 Yucca baccata, and YUCSCH 5 Yucca schidigera.

FIG. 5. Density of Larrea tridentata and Yucca spp. (schidigera and baccata) on a burned and unburned area, RedRock Canyon National Conservation Area, Mojave Desert. Error bars are one standard deviation for totalmean density.


TA

BL

E1.

SO

ILP

RO

PE

RT

IES

AM

ON

GM

ICR

OS

ITE

SO

NA

BU

RN

ED

AN

DU

NB

UR

NE

DA

RE

A,

RE

DR

OC

KC

AN

YO

NN

AT

ION

AL

CO

NS

ER

VA

TIO

NA

RE

A,

MO

JAV

ED

ES

ER

T.

Mea

ns

wit

ho

ut

share

dle

tter

sd

iffe

rat

P,

0.0

5(o

ne-

way

AN

OV

A,

Tu

key

’sT

est)

.M

icro

site

ab

bre

via

tio

ns:

INT

5in

ters

pace

,L

T5

bel

ow

Larr

ea

trid

enta

taca

no

py,

an

dY

UC

5b

elo

wY

ucca

(bacca

tao

rsc

hid

iger

a)

can

op

y.

1T

he

mea

ns

for

all

bu

rn3

mic

rosi

tele

vel

sare

pre

sen

ted

,ev

enth

ou

gh

som

eo

fth

evari

ab

les

did

no

th

ave

ast

ati

stic

all

ysi

gn

ific

an

tin

tera

ctio

n.

We

pre

sen

tall

of

the

mea

ns

bec

au

seth

eef

fect

size

sm

ay

have

bio

logic

al

sign

ific

an

ce,

alt

ho

ugh

we

ack

no

wle

dge

that

the

po

st-h

oc

test

was

mad

em

ore

con

serv

ati

ve

by

the

incl

usi

on

of

the

ad

dit

ion

al

mea

ns.

2E

lect

rica

lco

nd

uct

ivit

y.

3B

ulk

den

sity

,ca

lcu

late

das

the

oven

-dry

mass

of

soil

(,2

mm

)co

nta

ined

ina

core

of

kn

ow

nvo

lum

e,w

hic

hin

clu

ded

vo

lum

eo

fth

e.

2m

mfr

act

ion

.4B

ulk

den

sity

,ca

lcu

late

dth

esa

me

as

ab

ove

exce

pt

that

vo

lum

eo

fth

e.

2m

mfr

act

ion

was

no

tin

clu

ded

.

Vari

ab

le

Bu

rned

Un

bu

rned

Fact

or

effe

cts

INT

LT

YU

CIN

TL

TY

UC

Bu

rnM

icro

site

B3

M1

Mea

n(c

oef

fici

ent

of

vari

ati

on

%)

Pro

b.

.F

pH

7.5

b(3

)7.5

b(2

)7.9

a(3

)7.6

b(1

)7.6

b(1

)7.7

b(1

)0.7

219

,0.0

001

0.0

065

EC

(dS

/m)2

0.5

c(1

8)

1.5

b(5

3)

2.9

a(2

6)

0.4

c(2

7)

0.7

c(1

7)

0.7

c(2

3)

,0.0

001

,0.0

001

,0.0

001

P(m

g/k

g)

15c

(30)

32b

c(2

1)

94a

(21)

16c

(37)

21b

c(2

1)

36b

(70)

,0.0

001

,0.0

001

,0.0

001

K(m

g/k

g)

190d

(11)

426b

(22)

754a

(21)

182d

(37)

411b

c(3

0)

324c

(32)

0.0

007

,0.0

001

,0.0

001

C(%

wt.

)4.4

bc

(13)

5.1

ab

(14)

5.9

a(1

3)

2.0

d(7

1)

3.4

c(4

9)

3.7

c(6

5)

0.0

017

,0.0

001

0.3

336

N(%

wt.

)0.0

9b

c(6

4)

0.1

5ab

(24)

0.1

8a

(18)

0.0

6c

(21)

0.1

5ab

(35)

0.1

7a

(45)

0.3

287

,0.0

001

0.5

761

San

d(%

wt.

)53

(6)

55

(5)

53

(8)

56

(13)

56

(16)

55

(12)

0.3

716

0.3

591

0.8

451

Sil

t(%

wt.

)34

(6)

33

(9)

34

(10)

32

(20)

32

(23)

33

(20)

0.5

415

0.4

925

0.8

319

Cla

y(%

wt.

)13

(18)

13

(10)

14

(12)

13

(21)

12

(18)

12

(12)

0.2

334

0.5

162

0.6

243

BD

(g/c

m3)3

0.7

0ab

(15)

0.5

8b

(13)

0.6

4b

(8)

0.7

8a

(11)

0.6

2ab

(7)

0.6

7ab

(10)

0.0

103

0.0

028

0.6

462

BD

(g/c

m3)4

0.8

3ab

(16)

0.6

6c

(11)

0.7

3b

c(8

)0.8

9a

(7)

0.7

0b

c(8

)0.7

7ab

(9)

0.0

151

0.0

004

0.9

314


National Park in the southwestern MojaveDesert, where 80–84% of Yucca schidigeraresprouted after fire. The relatively low (5%)proportion of resprouting for Larrea tridentatathat we recorded is lower than Minnich (1995:17%) and two studies in the Sonoran Desert(Rogers and Steele 1980: 23%; McLaughlin andBowers 1982: 37%). This proportion is similar,however, to Brown and Minnich’s (1986) 3%resprout frequency at another site in the SonoranDesert. Differences in resprouting frequencyamong studies could be related to differences infire intensity, site factors (e.g., soils, prevalence ofherbivory), or climate (O’Leary and Minnich1981; Gibson et al. 2004). Our study site also wassituated near the upper elevational limit for L.tridentata, possibly limiting resprouting com-pared to other studies.

Relative to the unburned area, fire converted alate-successional shrub community dominated byhigh cover of Coleogyne ramosissima to acommunity with the perennial cover componentdominated by forbs. Dominant native perennialscolonizing the burn, such as Baileya multiradiata,Sphaeralcea ambigua, Eriogonum inflatum, andDasyochloa pulchella, have previously been re-ported as colonizers after fire or other distur-bances (e.g., road abandonment, land clearing).For example, B. multiradiata exhibited a density1.7 times greater than the next highest species(among 35 species) one year after fire in the upperSonoran Desert (Wilson et al. 1995). Similarly, S.ambigua, the most abundant perennial on theburn in our study, had the third highest densityamong all burn colonizers in Wilson et al.’s(1995) study. Brooks and Matchett (2003) foundthat these two species also had greater cover onburned than unburned areas in the MojaveDesert. Fifty to seventy years after agriculturalfield abandonment at 20 sites in the eastern

Mojave Desert, Carpenter et al. (1986) found thatcover of S. ambigua was three times greater onabandoned fields than off field. Dasyochloapulchella and E. inflatum also had greater coveron disturbed than undisturbed areas in MojaveDesert studies of succession on a cleared borrowpit (Vasek 1983) and a pipeline corridor (Abellaet al. 2007). These observations suggest that themajor native perennial colonizers of the burn inour study are early colonizers of several distur-bance types.

Colonization Windows andRevegetation Approaches

Brooks (2002) proposed that reduced Bromusrubens immediately following fire may offer acolonization window for actively revegetatingnative species on desert burns. Although theeffectiveness of this window may depend onprecipitation, we found that several nativeperennials colonized the burn even during theperiod of below-average precipitation character-izing our study. Despite this colonization, totallive native plant cover remained more than seventimes lower on the burned than the unburnedarea. Although active revegetation (e.g., seeding)can be expensive and prone to failure (Lovich andBainbridge 1999), managers may wish to attemptrevegetation of desert burns for several reasons.Revegetation may improve aesthetics, reducefugitive dust, increase animal forage, and providecompetition with exotic plants (DeFalco et al.2007). Based on several studies, natural reestab-lishment of late-successional species such asColeogyne ramosissima or Larrea tridentata willrequire long time periods, often longer than 30 yrand possibly longer than 100 yr (Callison et al.1985; Lei 1999; Lovich and Bainbridge 1999). Apotential concern about reestablishing these late-

FIG. 6. Emergence density of Bromus rubens among treatments performed on 0–5 cm soil seed bank samples froma burned and unburned area, Red Rock Canyon National Conservation Area, Mojave Desert. Error bars are onestandard deviation. Means without shared letters differ at P , 0.05 and were compared using Tukey’s test.


successional shrubs is that B. rubens is mostabundant below their canopies, facilitating fuelaccumulation (Brooks 2002). Therefore, reestab-lishing these shrubs may need to occur incombination with B. rubens control treatments.A complementary approach to test, in addition torevegetating with late-successional species, couldbe augmenting establishment of early colonizingnative species (e.g., Baileya multiradiata andSphaeralcea ambigua). In a Mojave Desertseeding study, B. multiradiata established at adensity of 3 plants/m2, second most among 12seeded species (Walker and Powell 1999). Ifnative species could be established in a post-firecolonization window, this early successionalvegetation type can persist for more than 30 yr(Carpenter et al. 1986; Abella et al. 2007).Although these species are not thought to formthe fertile islands most conducive to B. rubensestablishment, it is unclear whether this vegeta-tion type can depress B. rubens resurgence. Incombination with experiments, longer term mon-itoring of natural post-fire colonization patternsmay provide insight into these types of practicalquestions. However, short-term data on initialcolonization patterns such as provided by thisstudy also are crucial, as managers often mustplan revegetation activities soon after a fire.

ACKNOWLEDGMENTS

We thank the Bureau of Land Management (South-ern Nevada District) for funding the study through acooperative agreement with the University of NevadaLas Vegas (UNLV); Jill Craig (UNLV) for help withfieldwork; the UNLV research greenhouse for provid-ing space for the seed bank experiment; Cheryl Vanier(UNLV) for performing the statistical analysis of theseed bank experiment; and Sharon Altman (UNLV)and two anonymous reviewers for providing helpfulcomments on the manuscript.

LITERATURE CITED

ABELLA, S. R., A. C. NEWTON, AND D. N. BANGLE.2007. Plant succession in the eastern MojaveDesert: an example from Lake Mead NationalRecreation Area, southern Nevada. Crossosoma33:45–54.

AMACHER, M. C., K. P. O’NEILL, R. DRESBACH, AND

C. PALMER. 2003. Forest inventory and analysismanual of soil analysis methods. U.S. Departmentof Agriculture, Forest Service. Website http://nrs.fs.fed.us/fia/topics/soils/documents/FIA.P3.Soils.Lab.Manual.2003.pdf. [accessed 23 November2009].

BARTHA, S., S. J. MEINERS, S. T. A. PICKETT, AND

M. L. CADENASSO. 2003. Plant colonizationwindows in a mesic old field succession. AppliedVegetation Science 6:205–212.

BEATLEY, J. C. 1966. Ecological status of introducedbrome grasses (Bromus spp.) in desert vegetation ofsouthern Nevada. Ecology 47:548–554.

BROOKS, M. L. 1999. Alien annual grasses and fire inthe Mojave Desert. Madrono 46:13–19.

———. 2002. Peak fire temperatures and effects onannual plants in the Mojave Desert. EcologicalApplications 12:1088–1102.

——— AND J. R. MATCHETT. 2003. Plant communitypatterns in unburned and burned blackbrush(Coleogyne ramosissima Torr.) shrublands in theMojave Desert. Western North American Natural-ist 63:283–298.

——— AND ———. 2006. Spatial and temporalpatterns of wildfires in the Mojave Desert, 1980–2004. Journal of Arid Environments 67:148–164.

BROWN, D. E. AND R. A. MINNICH. 1986. Fire andchanges in creosote bush scrub of the westernSonoran Desert, California. American MidlandNaturalist 116:411–422.

CALLISON, J., J. D. BROTHERSON, AND J. E. BOWNS.1985. The effects of fire on the blackbrush(Coleogyne ramosissima) community of southwest-ern Utah. Journal of Range Management38:535–538.

CARPENTER, D. E., M. G. BARBOUR, AND C. J. BAHRE.1986. Old field succession in Mojave Desert scrub.Madrono 33:111–122.

CAVE, G. H. AND D. T. PATTEN. 1984. Short-termvegetation responses to fire in the upper SonoranDesert. Journal of Range Management 37:491–496.

D’ANTONIO, C. M. AND P. M. VITOUSEK. 1992.Biological invasions by exotic grasses, the grass/fire cycle, and global change. Annual Review ofEcology and Systematics 3:63–87.

DEFALCO, L. A., G. C. J. FERNANDEZ, AND R. S.NOWAK. 2007. Variation in the establishment of anon-native annual grass influences competitiveinteractions with Mojave Desert perennials. Bio-logical Invasions 9:293–307.

DUFRENE, M. AND P. LEGENDRE. 1997. Speciesassemblages and indicator species: the need for aflexible asymmetrical approach. Ecological Mono-graphs 67:345–366.

ESQUE, T. C. AND C. R. SCHWALBE. 2002. Alien annualgrasses and their relationships to fire and bioticchange in Sonoran desertscrub. Pp. 165–194 in B.Tellman (ed.), Invasive exotic species in theSonoran region. University of Arizona Press andthe Arizona-Sonora Desert Museum, Tucson, AZ.

FLEMATTI, G. R., E. L. GHISALBERTI, K. W. DIXON,AND R. D. TRENGOVE. 2004. A compound fromsmoke that promotes seed germination. Science305:977.

GIBSON, A. C., M. R. SHARIFI, AND P. W. RUNDEL.2004. Resprout characteristics of creosote bush(Larrea tridentata) when subjected to repeatedvehicle damage. Journal of Arid Environments57:411–429.

GUO, Q., P. W. RUNDEL, AND D. W. GOODALL. 1998.Horizontal and vertical distribution of desert seedbanks: patterns, causes, and implications. Journalof Arid Environments 38:465–478.

JAMES, J. J., M. A. CAIRD, R. E. DRENOVSKY, AND

R. L. SHELEY. 2006. Influence of resource pulsesand perennial neighbors on the establishment of aninvasive annual grass in the Mojave Desert.Journal of Arid Environments 67:528–534.

LATO, L. J. 2006. Soil survey of Clark County area,Nevada. U.S. Department of Agriculture, NaturalResources Conservation Service. U.S. GovernmentPrinting Office, Washington, D.C.


LEI, S. A. 1999. Postfire woody vegetation recovery andsoil properties in blackbrush (Coleogyne ramosis-sima) shrubland ecotones. Journal of the Arizona-Nevada Academy of Science 32:105–115.

———. 2001. Postfire seed bank and soil conditions ina blackbrush (Coleogyne ramosissima Torr.) shrub-land. Bulletin of the Southern California Academyof Sciences 100:100–108.

LOVICH, J. E. AND D. BAINBRIDGE. 1999. Anthropo-genic degradation of the southern California desertecosystem and prospects for natural recoveryand restoration. Environmental Management24:309–326.

MCCUNE, B. AND M. J. MEFFORD. 1999. PC-ORD:multivariate analysis of ecological data. Version 4.User’s guide. MjM Software Design, GlenedenBeach, OR.

MCLAUGHLIN, S. P. AND J. E. BOWERS. 1982. Effectsof wildfire on a Sonoran Desert plant community.Ecology 63:246–248.

MINNICH, R. A. 1995. Wildland fire and early postfiresuccession in Joshua tree woodland and blackbrushscrub of the Mojave Desert of California. SanBernardino County Museum Association Quarter-ly 42:99–106.

NOY-MEIR, I. 1973. Desert ecosystems: environmentand producers. Annual Review of Ecology andSystematics 4:25–51.

O’LEARY, J. F. AND R. A. MINNICH. 1981. Postfirerecovery of the creosote bush scrub vegetation inthe western Colorado Desert. Madrono 28:61–66.

PATTEN, D. T. AND G. H. CAVE. 1984. Fire temper-atures and physical characteristics of a controlledburn in the upper Sonoran Desert. Journal ofRange Management 37:277–280.

PEET, R. K., T. R. WENTWORTH, AND P. S. WHITE.1998. A flexible, multipurpose method for record-ing vegetation composition and structure. Castanea63:262–274.

RAISON, R. J. 1979. Modification of the soil environ-ment by vegetation fires, with particular referenceto nitrogen transformations: a review. Plant andSoil 51:73–108.

ROGERS, G. F. AND J. STEELE. 1980. Sonoran Desertfire ecology. Pp. 15–19 in M. A. Stokes and J. H.

Dieterich (tech. coords.), Proceedings of the firehistory workshop. General Technical Report RM-81. U.S. Department of Agriculture, Rocky Moun-tain Forest and Range Experiment Station, FortCollins, CO.

SAS INSTITUTE. 1999. SAS/STAT user’s guide. SASInstitute, Inc., Cary, NC.

SAS INSTITUTE. 2004, JMP user’s guide. SAS Institute,Inc., Cary, NC.

UNITED STATES DEPARTMENT OF AGRICULTURE

(USDA), NATURAL RESOURCES CONSERVATION

SERVICE. 2007. The PLANTS database. NationalPlant Data Center, Baton Rouge, LA.Website http://plants.usda.gov. [accessed 23 November 2009].

VAN MANTGEM, P., M. SCHWARTZ, AND M. KEIFER.2001. Monitoring fire effects for managed burnsand wildfires: coming to terms with pseudoreplica-tion. Natural Areas Journal 21:266–273.

VASEK, F. C. 1983. Plant succession in the MojaveDesert. Crossosoma 9:1–23.

WALKER, L. R. AND E. A. POWELL. 1999. Effects ofseeding on road revegetation in the Mojave Desert,southern Nevada. Ecological Restoration17:150–155.

WARR, S. J., K. THOMPSON, AND M. KENT. 1993. Seedbanks as a neglected area of biogeographic research: areview of literature and sampling techniques. Progressin Physical Geography 17:329–347.

WILSON, R. C., M. G. NAROG, A. L. KOONCE, AND

AND B. M. CORCORAN. 1995. Postfire regenerationin Arizona’s giant saguaro shrub community.Pp. 424–431 in L. H. DeBano, P. H. Ffolliott, A.Ortega-Rubio, G. J. Gottfried, R. H. Hamre, andC. B. Edminster (tech. coords.), Biodiversity andmanagement of the madrean archipelago: thesky islands of southwestern United States andnorthwestern Mexico. General Technical ReportRM-GTR-264. U.S. Department of Agriculture,Forest Service, Rocky Mountain Forest and RangeExperiment Station, Fort Collins, CO.

ZIMMERMAN, G. M., H. GOETZ, AND P. W. MIELKE.1985. Use of an improved statistical method forgroup comparisons to study effects of prairie fire.Ecology 66:606–611.


early post-fire plant establishment on a mojave desert burn

Documents