January 1999 39

Throughout the history of GreatPlains grasslands NorthAmerican bison (Bos bison

also known as Bison bison Jones etal 1992) and other large herbivoreswere abundant and conspicuous com-ponents of the biota (Wedel 1961Stebbins 1981) Many of the earliestherbivores particularly those thatwere primarily browsers are nowextinct but their consumption ofwoody vegetation is thought to haveplayed a critical role in the post-Pleistocene rise of the grasslandbiome and the subsequent increasein bison populations (Axelrod 1985Hartnett et al 1997) Indeed thelarge herds of bison encountered byearly Europeans on the Great Plainswere likely the result of the rapidearly-Holocene proliferation of thisungulate into a relatively young andexpanding ldquotreelessrdquo grasslandbiome (Stebbins 1981 Axelrod1985) In the most productive re-gions of the Great Plains the easterntallgrass prairies abundant bisonherds were noted by early explorers(Shaw and Lee 1997) Although herdswere larger in the western shortgrass

steppe some have speculated that agreater density of bison could besupported in eastern tallgrass prai-ries than elsewhere in the plains(McHugh 1972) It is unfortunatethen that despite the historic abun-dance of bison in tallgrass prairiestheir ecological effects in these mesicgrasslands are poorly understood

Knowledge of the bisonrsquos role intallgrass prairies is lacking becausethe extent of this grassland and theabundance of its dominant ungulatehave declined dramatically and intandem over the last 150 years Al-though there is debate over the num-bers of bison inhabiting the GreatPlains before the 1800s (estimatesrange from 30 million to 60 millionMcHugh 1972 Flores 1991) it iswell documented that from 1830 to1880 the slaughter of bison in theGreat Plains reduced their numbersto an estimated low of a few thou-sand individuals Widespread culti-vation of the plains which accompa-nied the near extirpation of the bisonreduced the once-vast tallgrass prairie(approximately 68 million hectares)to less than 5 of its presettlement

range (Samson and Knopf 1994)The near-simultaneous reduction inherbivore abundance and grasslandextent left little opportunity to assessbisonndashtallgrass prairie interactions

Today thanks to conservationefforts (Berger and Cunningham1994) bison numbers in the GreatPlains have rebounded (to approxi-mately 150000) and significantpublic and private herds are main-tained in several mixed- and short-grass prairie preserves and ranchesIt is from these semi-arid grasslandsmany of which escaped cultivationthat the most extensive knowledgeof bisonndashgrassland interactions hasbeen generated (Peden et al 1974Coppock et al 1983) By contrastthe current understanding of tallgrassprairie structure and function hasbeen developed almost exclusivelyfrom studies of ungrazed tracts orfrom sites grazed by domestic cattle(Risser et al 1981 Collins 1987 Howe1994 Leach and Givnish 1996) Onlyrecently have bison been reintro-duced to tallgrass prairie sites thatare large enough to assess both theirinfluence on other biota and ecosys-tem processes as well as their inter-actions with other important fea-tures of these grasslands particularlyfire (Collins et al 1998 Coppedgeand Shaw 1998 Knapp et al 1998b)

The Konza Prairie Research Natu-ral Area in the Flint Hills of north-eastern Kansas is the largest tract ofunplowed tallgrass prairie (3500 ha)in North America dedicated to re-search (Knapp et al 1998b) KonzaPrairie was one of the original sitesselected in 1981 for inclusion in the

The Keystone Role of Bison inNorth American Tallgrass PrairieBison increase habitat heterogeneity and alter a broad array of

plant community and ecosystem processesAlan K Knapp John M Blair John M Briggs Scott L Collins David C Hartnett

Loretta C Johnson and E Gene Towne

Alan K Knapp John M Blair John MBriggs Scott L Collins David C Hartnettand Loretta C Johnson are professorsand E Gene Towne is a research associatein the Division of Biology Kansas StateUniversity Manhattan Kansas 66506Collins is also an adjunct professor withthe Department of Zoology University ofMaryland College Park MD 20742 anda program director in the Division of En-vironmental Biology National ScienceFoundation Arlington VA 22230 copy 1999American Institute of Biological Sciences

Ungulate grazingactivities and fire arekey to conserving and

restoring the bioticintegrity of the remainingtracts of tallgrass prairie

40 BioScience Vol 49 No 1

National Science Foundationrsquos LongTerm Ecological Research (LTER)program and ongoing experimentaltreatments on some parts of the sitedate to 1972 Past syntheses of re-search at Konza Prairie have focusedon the effects of fire and climatic

variability (Knapp and Seastedt1986 Knapp et al 1998b) but amajority of these data were collectedin the absence of bison grazing

In 1987 30 bison were reintro-duced to Konza Prairie and the herdwas allowed to increase until 1992

Since that time the herd has beenmaintained at approximately 200individuals who have had unre-stricted access to a 1012 ha portionof the landscape Within this areaare 10 watersheds that are subjectedto different frequencies of late-springprescribed fire The target animaldensity of 200 animals was selectedso that approximately 25 ofaboveground primary production isconsumed annually This consump-tion rate is approximately half thatof tallgrass prairie managed for do-mestic cattle The bison herd at KonzaPrairie is not provided supplementalfeed in winter nor is it actively man-aged Thus this herd provides anopportunity to document the impactof bison on a native tallgrass prairieecosystem

Although research on bisonndashtallgrass prairie interactions begansoon after bison were reintroducedto Konza Prairie comprehensivestudies spanning scales from the leafto the landscape level began in theearly 1990s In this article we pro-vide the first synthesis of these re-search efforts with the goal of high-lighting the keystone role (sensuPower et al 1996) that bison playedin the tallgrass prairies of the pastWithin this overview we addresstwo general questions What are thedirect and indirect effects of bisonon patterns and processes in tallgrassprairie What factors influence thespatial and temporal patterns of graz-ing activities by bison Perhaps asimportant as addressing these ques-tions the data that we present indi-cate that large-ungulate herbivorycan and should play a key role inthe management and conservationof the remaining tracts of this once-widespread grassland

Bison grazing activitiesand plant responsesLike all large herbivores bison donot graze indiscriminately across thelandscape or even within a local area(Senft et al 1987 Wallace et al1995) Rather they graze in twopatterns in tallgrass prairie creatingboth distinct grazing patches (typi-cally 20ndash50 m2 in Flint Hills tallgrassprairie Catchpole 1996) and moreextensive grazing lawns (larger than400 m2 McNaughton 1984) In both

Figure 1 Grazing by bison increases spatial variability in tallgrass prairie Immediatelyafter fire (April) in grazed watersheds on the Konza Prairie a native tallgrass prairie innortheastern Kansas the uneven consumption of fuel by fire in grazed areas (upperphoto) is caused by the patchiness of bison grazing the previous year Dark areasindicate sites where bison did not graze or where grazing was light and fuel loads weretherefore sufficient to carry a fire through the patch Lighter areas are patches that weregrazed repeatedly by bison such that fuel loads were reduced and the impact of fire wasminimal Later in the season bison graze preferentially in recently burned watersheds(lower photo) In this late spring (May) view of a burned watershed unburnedwatersheds can be seen in the background Bison at Konza Prairie have access to 10watersheds that differ in fire regime

January 1999 41

cases bison revisit grazed sitesthroughout the season such that re-peated defoliation of grazed plants isthe norm and relatively sharp bound-aries between grazed and ungrazedvegetation become evident (Figure1) Bison are primarily graminoidfeeders and consume higher propor-tions of the dominant grasses thanwould be predicted based on grassavailability in the landscape (Pedenet al 1974 Van Vuren and Bray1983 Steuter et al 1995) Bison tendto avoid forbs and woody specieswhich usually constitute less than10 of their diet Thus within abison grazing area forbs are oftenconspicuously left ungrazed and aresurrounded by grazed grasses (Fahne-stock and Knapp 1993 Damhoureyehand Hartnett 1997)

Preferential grazing of the domi-nant grasses by bison sets the stagefor significant alterations in com-petitive interactions among the C4grasses and the C3 forbs Such shiftsare important for plant communitystructure because in ungrazed andfrequently burned prairie a smallgroup of grass species (Andropogongerardii big bluestem Sorghastrumnutans Indian grass Panicumvirgatum switchgrass and Andro-pogon scoparium little bluestem)account for most biomass densityleaf area and resource consumption(Knapp 1985 Briggs and Knapp1995) However it is the species-rich forb component (more than 350species are recorded on Konza Prai-rie Freeman 1998) that is critical forthe maintenance of high levels ofbiotic diversity in tallgrass prairie(Gibson and Hulbert 1987 Glennand Collins 1990 Turner et al 1995)Thus by grazing on grasses and al-lowing forbs to flourish bison havethe potential to significantly influ-ence biodiversity in these grasslands(Collins et al 1998)

The short-term effects of bisonherbivory on the most abundant prai-rie grass A gerardii are differentfrom the long-term effects At theleaf level short-term responses toleaf removal are typical of manygraminoids in grazing systems(McNaughton 1983) Wallace (1990)reported a postgrazing enhancementof photosynthesis in A gerardii inOklahoma Similarly on Konza Prai-rie midseason photosynthetic rates

averaged 53 higher ingrazed tillers (individualgrass stems) than in ungrazedplants with a maximumstimulation of 150 (Figure2a) Mechanisms for this en-hancement of photosynthe-sis include increased lightavailability and reduced wa-ter stress for all species ingrazed patches (Fahnestockand Knapp 1993) and greatertissue nitrogen concentrationin A gerardii leaves as nitro-gen is reallocated from roots

These potential compensatory in-creases in photosynthesis after graz-ing may be augmented by the trans-location of carbon reserves frombelowground to aboveground tissuesVinton and Hartnett (1992) foundthat in the first year of grazinggrowth and biomass of grazed Agerardii tillers had completely com-pensated for the loss to grazing byseasonrsquos end But after several yearsof grazing the ability of tillers tocompensate for lost tissue was re-duced (Figure 2b) Vinton andHartnett (1992) attributed these dif-ferences in short- and long-term re-sponses to reductions in below-ground carbon allocation and storedcarbohydrate reserves after severalyears of grazing Turner et al (1993)also demonstrated the importance of

grazing history on productivity intal lgrass prairie by measuringaboveground primary production ina number of sites with different graz-ing histories They found that com-pensatory regrowth of biomass oc-curred in sites with little history ofgrazing but not in sites that had beengrazed heavily in previous years

Some researchers have argued thatthe inability of tallgrass prairiegrasses to compensate for the bio-mass lost in frequently grazed areasis evidence that prolonged intensivebison grazing did not occur in thesegrasslands (Shaw and Lee 1997) Ifthis hypothesis were true then grassabundance would decline continu-ally following repeated grazingHowever the tallgrass prairiersquos lossof production potential due to graz-ing is short lived At Konza Prairie

Figure 2 Response of tallgrassprairie vegetation to bison her-bivory at three temporal scales(a) Instantaneous leaf-level pho-tosynthetic responses of Andro-pogon gerardii to grazing (ErinQuestad and Alan K Knapp un-published data) (b) Relativegrowth rates of tillers of Agerardii in sites that differ in graz-ing history (data from Vintonand Hartnett 1992) Both un-grazed tillers and grazed tillerswith a history of grazing hadsimilar growth rates whereasgrazed tillers without a history ofgrazing had significantly greatergrowth rates (c) End-of-seasonaboveground biomass inungrazed sites that differ in graz-ing history (data from Knapp etal 1998a) Asterisks in panels (a)and (b) indicate significantly dif-ferent values (P lt 005) for com-parisons at specific dates in panel(c) they indicate comparisonswithin growth forms

42 BioScience Vol 49 No 1

permanent fenced exclosures haveexcluded bison from experimentalplots within grazed watersheds forseveral years When adjacent grazedsites were also protected from graz-ing with temporary exclosuresaboveground production in the firstyear in these newly protected siteswas reduced relative to that in theadjoining long-term ungrazed areas(Figure 2c) however these sites re-covered their production potentialby the second year Thus produc-tion potential can recover if bisongrazing is sufficiently dynamic ei-ther spatially or temporally suchthat sites are grazed intermittently

Removal of grass leaf area by bi-son and reductions in the capabilityof the dominant grasses to compen-sate for tissue lost after multipleyears of grazing suggest that the co-occurring subdominant forbs maybenefit from bison grazing Indeedcomparisons of forbs inside grazingpatches with those in adjacentungrazed prairie have shown thatgas exchange and aboveground bio-mass production density and cover

can be enhanced by the selective con-sumption of grasses by bison (Fahn-estock and Knapp 1993 Hartnett etal 1996 Damhoureyeh and Hartnett1997) Given the importance of pastgrazing pressures to the direct re-sponses of grasses to herbivory andthe indirect responses of forbs iden-tifying those factors that influencethe selection and reselection of graz-ing patches by bison in tallgrass prai-rie is critical for understanding thelong-term consequences of bisongrazing patterns

Factors influencing bisonselection of grazing sitesHistorical information regarding bi-son grazing patterns in the GreatPlains is replete with anecdotal ac-counts of herds attracted to recentlyburned grasslands (Figure 1McHugh 1972 Pyne 1982) butquantitative evidence of this prefer-ence in tallgrass prairie has beenlacking until recently (Coppedge andShaw 1998) As noted earlier bisonat Konza Prairie have free access to

10 watersheds that are subject todifferent fire frequencies Since 1991twice-weekly observations of the dis-tribution of bison within this areahave been made to assess patterns ofbison grazing (Nellis et al 1992)The results confirm that bison dograze preferentially within burnedwatersheds from April (burning takesplace in late Marchndashearly April)through June and July and in someyears through August (Figure 3Vinton et al 1993 Nellis and Briggs1997) In addition to grazing prefer-entially in burned sites bison in-crease their selective consumption ofsome grass species in burned sitesrelative to unburned sites (Pfeifferand Hartnett 1995) Late in the sum-mer lowland topographic positionswith deeper soils (and thereforegreater soil moisture and plant pro-ductivity Knapp et al 1993) inburned watersheds become preferredgrazing locations as the uplands dryThis preference for burned areas intallgrass prairie is consistent withpostfire responses in mixed grass prai-rie (Coppock and Detling 1986) aswell as with large ungulatesrsquo winterpreference for burned sites in the north-ern mixed grasslands of YellowstoneNational Park (Pearson et al 1995)

Within a watershed or at a spe-cific topographic position in tallgrassprairie several factors may influ-ence initial patch selection andreselection (see also Wallace et al1995) In addition patch selectionhas several long-term consequencesWhen bison were first reintroducedto Konza Prairie they encountered amosaic of burned and unburned wa-tersheds with significant differencesamong watersheds in the spatial het-erogeneity of plant community com-position For example frequentlyburned but ungrazed watersheds aredominated by C4 grasses and havelow species richness and diversitywhereas less frequently burned siteshave higher species richness and forbcover (Gibson and Hulbert 1987Collins 1992) Initial studies onKonza Prairie indicated clearly thatbison established grazing patches inareas strongly dominated by C4grasses and that these patches werereselected at a high rate (Vinton etal 1993) Six years later a survey offloristic composition indicated thatestablished bison grazing patches had

Figure 3 Spatial distribution of bison in 10 watersheds in the central portion of KonzaPrairie after the completion of spring burning During the months of April May andJune 1992 bison locations were determined twice per week and recorded on large-scalespatially rectified aerial photographs overlaid with a 30-meter grid Bison wereobserved most frequently in the recently burned watersheds B burned watersheds Uunburned watersheds

January 1999 43

a higher abundance of forbs and alower cover of grasses than adjacentungrazed patches (Catchpole 1996)Similar small-scale patterns of forband grass abundance were observedby comparing the floristic composi-tion inside and outside grazingexclosures in grazed watersheds atKonza Prairie (Hartnett et al 1996)These observations suggest that bi-son alter plant community composi-tion at the patch scale by selectingspecies-poor grass-dominated sitesand converting them to sites of lo-cally higher diversity (Figure 4)

A second factor that may influ-ence patch selection and reselectionby bison is plant quality The foliar

nitrogen content of plants is highlyvariable both spatially and tempo-rally Bison contribute to this patchi-ness through deposition of nitrogen-rich urine Steinauer (1994) appliedsynthetic bovine urine at randomlyselected locations along eighttransects at Konza Prairie (four eachin grazed and ungrazed areas) Inungrazed areas grass cover was sig-nificantly higher in plots that werefertilized with urine than in plotswithout urine But in the grazed areaby contrast grass cover was signifi-cantly lower on the urine-treatedplots than in plots without urinebecause bison preferentially grazedthe grasses on the urine-treated plots

Day and Detling (1990) have shownthat grasses growing on urine patchesin mixed-grass prairie have higherleaf nitrogen content and are there-fore more nutritious per bite thangrasses growing on patches withouturine Not only was grass cover loweron urine patches at Konza Prairiebut the total area of grazed patcheson the urine-treated transects wassignificantly larger than the area ofgrazed patches on transects that werenot treated with urine (Steinauer1994) Thus the enhanced produc-tivity of grasses growing on urinepatches represents a potential stimu-lus for the initiation and reselectionof grazing patches by bison

Figure 4 Conceptual model of the spatial and temporal dynamics of bison grazing activities and the responses of tallgrass prairieto the reintroduction of bison Before the reintroduction of bison watershed attributes at Konza Prairie differed strongly dependingon the fire regime imposed with frequently burned sites notable for their high productivity but low plant species diversity Selectivebison grazing at the watershed level (preference for burned sites particularly in the spring) and the patch level (selection of patchesdominated by C4 grasses) has led to increased similarity in watershed attributes despite differences in fire history Moreoversignificant increases in plant species diversity and spatial heterogeneity have been noted in all watersheds Superscripts denote studieson which this synthesis is based 1 Nellis and Briggs (1997) 2 Vinton et al (1993) 3 Briggs and Knapp (1995) 4 Collins (1992)Collins et al (1995) 5 Vinton and Hartnett (1992) Vinton et al (1993) 6 Hartnett et al (1997) 7 Fahnestock and Knapp (19931994) 8 Catchpole (1996) 9 Collins and Steinauer (1998) 10 Hartnett et al (1996)

44 BioScience Vol 49 No 1

Given that bison prefer grazingpatches that are initially dominatedby C4 grasses but that their selectiveforaging and reselection habits con-vert these patches to sites with agreater abundance of nonforage spe-cies it is likely that patch locationsare spatially dynamic across the land-scape Patches can ldquomoverdquo by twomechanisms patch abandonmentfollowed by selection of a new patchor patch migration wherein portionsof a patch are abandoned and thepatch expands into adjacent areasOver a 3-year period Catchpole(1996) found the rate of patch aban-donment to be approximately 6ndash7per year in both burned and un-burned watersheds thus at leastportions of previously establishedgrazing patches were reselected at ahigh rate (Figure 4) However whengrazing patches were mapped andcompared across years the extent ofspatial reselectionmdashalthough highlyvariable due to differences in totalburned area available to bison in anyone yearmdashaveraged approximately50 per year Thus grazing patchesin both burned and unburned water-sheds appear to migrate significantlyfrom year to year This local migra-tion permits periodic release of por-tions of the grassland from grazingpressures (Figure 4) and provides amechanism for recovery of below-ground carbohydrate storage reservesand production potential

At the watershed and landscapescales the long-term consequencesof bison activities include a reductionin cover dominance and productivityof grasses the competitive release ofmany subdominant species resultingin an increase in the abundance offorbs an overall increase in plantspecies richness and diversity andincreased spatial heterogeneity (Fig-ure 5 Hartnett et al 1996) Althoughalterations in plant community com-position can be attributed in largepart to the direct effects of grazingby bison increased plant species rich-ness is also likely to be a product ofincreased microsite diversity gener-ated by nongrazing activities suchas dung and urine deposition tram-pling and wallowing These andother bison activities contribute sig-nificantly to the increase in spatialheterogeneity that is characteristicof grazed tallgrass prairie (Figure 5)

Other impacts of bisonin tallgrass prairie

Effects of ungulates especially graz-ing on plant community composi-tion and structure have been studiedin many grasslands worldwide(McNaughton 1984 Milchunas etal 1988 Frank and McNaughton1992 Milchunas and Lauenroth1993) Ungulate activities howeveraffect many other aspects of grass-land structure and function includingthe physical structure of the environ-ment and the rates of a number ofecosystem-level processes (McNaugh-ton 1993 Frank and Evans 1997McNaughton et al 1997) There areseveral other important mechanismsby which bison alter ecosystem-levelprocesses and physical habitat struc-ture in tallgrass prairie

Nutrient redistribution and cyclingBison can substantially alter nutri-ent cycling processes and patterns ofnutrient availability in tallgrass prai-rie Their effects on nitrogen cyclingare critical because nitrogen avail-ability often limits plant productiv-ity in these grasslands (Seastedt et al1991 Blair 1997 Turner et al 1997)and influences plant species compo-sition (Gibson et al 1993 Wedinand Tilman 1993) Simulation mod-els of tallgrass prairie responses tograzing (Risser and Parton 1982)and studies of grazers in other grass-lands (Frank and Evans 1997McNaughton et al 1997) have dem-onstrated a disproportionate influ-ence of ungulates including bisonon the regulation of nitrogen cyclingprocesses Preliminary data fromKonza Prairie suggest that bison aresimilarly important in controllingnitrogen cycling in tallgrass prairie

Bison influence nitrogen cyclingconservation and availability intallgrass prairie ecosystems by alter-ing several soil and plant processesUngulates in grasslands consumerelatively recalcitrant plant biomassand return labile forms of nitrogen(ie urine) to soils (Ruess andMcNaughton 1988) thus bypassingthe otherwise slow mineralization ofnitrogen in plant litter Nitrogen inbison urine is largely urea which canbe hydrolyzed to ammonium in amatter of days (Ruess and McNaugh-ton 1988) Indeed application of

synthetic bison urine increased con-centrations of ammonium and ni-trate in Konza Prairie soils over 130-fold and 30-fold respectively 8 daysafter application (J R Matchett andLoretta C Johnson unpublished data)

Bison grazing can decrease theexport of nitrogen from tallgrassprairie by altering the magnitudes oftwo major pathways of nitrogenlossmdashcombustion and ammoniavolatilization Fire is the major path-way of nitrogen loss from ungrazedtallgrass prairie (Dodds et al 1996Blair 1997) nitrogen loss from burn-ing averages 1ndash4 gmiddotmndash2middotyrndash1 (Blair etal 1998) Grazing lowers combus-tion losses of nitrogen in tallgrassprairie by reducing the abovegroundplant detritus and increasing thepatchiness of a prairie fire (Figure 2Hobbs et al 1991) Although vola-tilization of ammonia can be in-creased by grazing in other types ofgrassland (Detling 1988) Hobbs etal (1991) suggested that any increasein ammonia volatilization in tallgrassprairie will be more than compen-sated for by a reduction in combus-tion losses of nitrogen

Finally bison grazing affects theamount and quality of plant litter re-turned to soils Grazing increases plantuptake of nutrients (Ruess 1984) andshoot nitrogen content in many grass-lands (Holland and Detling 1990Milchunas et al 1995) includingtallgrass prairie (Turner et al 1993)However the effects of grazers onroot growth and chemistry varyamong grasslands (Milchunas andLauenroth 1993) On Konza Prairieroot productivity and root biomasswere 30 and 20 lower respec-tively in bison grazing lawns than inungrazed exclosures In addition thenitrogen concentration of new rootgrowth in bison grazing lawns atKonza Prairie increased significantlyfrom 06 to 09 and the CN ratioof roots decreased A lower CN ratioreduces microbial immobilization andenhances nitrogen availability withingrazed areas Indeed recent studieson Konza Prairie indicated that netnitrogen mineralization in bison graz-ing lawns was 153 greater and netnitrification 126 greater than inungrazed prairie (Figure 6) Further-more net nitrogen mineralization rateswere proportional to the intensity ofbison use of a given area Thus the

January 1999 45

net effect of bison grazing appears tobe increased rates of nitrogen cyclingcoupled with a significant increase inspatial heterogeneity in nitrogen avail-ability together these effects can alterpatterns of plant productivity and spe-cies composition in tallgrass prairie(Figure 5 Steinauer and Collins 1995)

Wallowing One aspect of bison be-havior that differs from that of cattleand is primarily a physical activityis wallowing Wallows in Flint Hillstallgrass prairie which are estab-lished primarily in level upland orlowland sites dramatically alter thepatch structure of this prairie Bisonwallows develop as the animals pawthe ground and roll in the exposedsoil Continued use of wallows bybulls cows and calves creates a soildepression of 3ndash5 m in diameter (and10ndash30 cm in depth) that is devoid ofvegetation These denuded patcheseither gradually revegetate or remainas bare soil depending on the fre-quency of revisitation by bison Withthe vast numbers of bison that onceoccupied the Great Plains these soildepressions were probably abundantand widespread features of the land-scape (England and DeVos 1969)For example a number of relic wal-lows had to be filled to level the play-ing field for the first University ofOklahoma home football game in 1895(University of Oklahoma Athletic De-partment 1986) Relic wallows stillexist in many areas where bison havenot occurred in the past 125 years

Environmental conditions in relicand newly established wallowsstrongly influence prairie patch dy-namics (Polley and Collins 1984Polley and Wallace 1986) Becauseof soil compaction wallows oftenretain rainwater in the spring creat-ing localized habitats that are suit-able for ephemeral wetland speciessimilar to vernal pools in California(Holland and Jain 1981 Uno 1989)In the summer however the samewallows support only plants that cantolerate severe drought Vegetationcomposition and structure in wal-lows is different from that in thesurrounding prairie (Polley andCollins 1984) and these differencesare enhanced by fire (Collins andUno 1983) which may not spreadthrough wallows because of low fuelloads Consequently at larger spa-

tial scales grazed prairie that con-tains bison wallows has higher plantspecies diversity than grazed prairiewithout wallows (Collins and Bar-ber 1985) Thus bison can physi-cally alter grasslands in ways thatincrease environmental heterogeneityand enhance both local and regionalbiodiversity (Hartnett et al 1997)

Bison carcasses Bison not only af-fect vegetation patterns and soil pro-cesses through their grazing activi-ties but also have profound andlasting localized effects after theydie Although legal requirements andmanagement practices dictate theremoval of carcasses of domesticherbivores from public and privategrasslands native herbivores rou-tinely die of natural causes and theirbodies remain in situ As part of theminimal management strategy atKonza Prairie bison that die on siteare not removed As a result thesecarcasses create unique local distur-bances (Figure 7) that are the focusof studies to assess their effects onsoil nutrients and vegetation re-sponses in tallgrass prairie

When an individual bison diescopious quantities of fluids (withhigh nitrogen concentration) are re-leased during decomposition Adultbison can weigh more than 800 kgand these carcasses typically kill un-derlying and adjacent plants creat-ing a denuded zone of 4ndash6 m2 (Figure7) Although the fluids that are ini-tially released are toxic to vegeta-tion these sites eventually becomezones of high fertility For examplesoil cores extracted from the centerof carcass sites on Konza Prairie 3years after death had inorganic ni-trogen concentrations that were twoto three times higher than the sur-rounding prairie This nutrient enrich-ment may extend up to 25 m awayfrom the original carcass site andresults in patches dominated initiallyby early successional species Theaboveground primary production inthese patches is two to three timeshigher than in undisturbed prairie

Although disturbances created bybison carcasses are sporadic and lo-calized on Konza Prairie they pro-vide nutrient pulses that exceed allother natural processes even urineand fecal deposits Overall we canonly speculate about historical rates

of bison mortality Given the enor-mous size of the bison populationbefore their widespread slaughter inthe 1800s annual mortality wasprobably high High death rateswould have been especially commonduring droughts when there wouldbe the potential for large numbers ofcarcasses to occur across the land-scape Even though predators andscavengers may have consumed andrelocated many of these carcassesdecomposition of the remaining andpartial carcasses would still have re-sulted in patches of locally high nu-trient concentration Thus althoughit was variable bison mortality wouldhave led to a continual cycle of dis-turbance and recovery of thesepatches in presettlement grasslands

Are bison keystone speciesThe net effects of selective bison graz-ing activities at the landscape patchand individual plant level includeshifts in plant species compositionalterations of the physical and chemi-cal environment and increased spa-tial and temporal heterogeneity invegetation structure soil resourceavailability and a variety of ecosys-tem processes (Figures 4 5 and 6)Before bison reintroduction at KonzaPrairie the long-term burning ex-periments produced clear patterns ofresponse in the vegetation As firefrequency increased the dominanceof C4 grasses increased and the coverof C3 grasses forbs and woody spe-cies decreased (Figure 4 Gibson andHulbert 1987) Overall plant spe-cies diversity declined as fire fre-quency increased in ungrazed tall-grass prairie (Collins et al 1995)

These patterns in community struc-ture which had developed over 20years of burning treatments at KonzaPrairie are being rapidly and dra-matically altered by the grazing ac-tivity of the reintroduced bison Inparticular grazing by bison has low-ered the abundance of the dominant C4

grasses increased the abundance ofthe subdominant C3 grasses and forbsand markedly increased plant speciesdiversity (by 23) richness (by 38)and community heterogeneity (by13) relative to ungrazed sites evenunder annual burning conditions(Hartnett et al 1996 Collins andSteinauer 1998 Collins et al 1998)

46 BioScience Vol 49 No 1

Because of the multiple and dra-matic effects of bison on this land-scape we believe that bison are key-stone species in the tallgrass prairieOther authors have noted the poten-tial of large grazing mammals to actas ldquokeystone herbivoresrdquo capable ofmaintaining open grassland vegeta-tion that would otherwise undergosuccession to shrubland or wood-land (Owen-Smith 1987) Indeed thedisappearance of a grazing mega-fauna at the end of the Pleistocenemay have played a major role in thewidespread transition from steppe totundra at that time (Zimov et al 1995)However the concept of keystonespecies has been controversial since itsinception (Power et al 1996) One ofthe problems with this concept hasbeen the variable interpretation of cri-teria by which species are determinedto be keystone Power et al (1996)consider a keystone species to be ldquoonewhose impact on its community orecosystem is large and disproportion-ately large relative to its abundancerdquoTo make this definition operationalthese authors proposed a measure ofcommunity importance (CI) to be usedas an index of the strength of theimpact of a given species

CI = [(tN ndash tD)tN][1pi]

where tN is a quantitative measure ofa trait (eg diversity) in an intactcommunity tD is the measure of thetrait when species i has been deletedand pi is the proportional abundance(biomass) of species i before it wasdeleted CI values ldquomuch greater than1rdquo indicate that a species is key-stone We estimate bison biomass atKonza Prairie to be approximately11ndash12 gmiddotm2 (Collins and Steinauer1998) which is approximately 1of the total vegetative biomass OnKonza Prairie diversity is signifi-cantly higher (10ndash33) on grazedsites than ungrazed sites (Hartnett etal 1996) and these values yield arange of CIs from 6 to 25 For thisreason and others we consider bisonto be keystone species in tallgrassprairie ecosystems

Are bison and cattle functionalequivalents in tallgrass prairieThe historical presence of immenseherds of large ungulates in Great

Plains grasslands is undisputed(McHugh 1972) and we have em-phasized the keystone role that bisonplayed in determining the structureand function of tallgrass prairies atmultiple spatial and temporal scalesWith the replacement of native bisonby domesticated cattle in the remain-ing grasslands an obvious issue isthe degree of similarity between thesetwo ungulates with respect to theireffects on tallgrass prairie In otherwords can bison and cattle be con-sidered ecological equivalents

There have been several previousattempts to answer this questionbut the results have been equivocalat best (Plumb and Dodd 1993Hartnett et al 1997) The primarybarrier to resolving this issue restswith a lack of comparative studies inwhich management is held constantand the type of grazer is varied Suchstudies have recently been initiatedat Konza Prairie Results after 3 yearsindicate that the abundance and rich-ness of annual forbs and the spatialheterogeneity of biomass and coverare higher in sites with bison than insites with cattle No dramatic differ-ences have been detected howeverbetween cattle- and bison-grazed sitesin cover of the dominant C4 grass Agerardii or the dominant forb Am-brosia psilostachya total plant spe-cies richness is also not dramaticallydifferent (E Gene Towne and DavidC Hartnett unpublished data)

Results at Konza Prairie are con-sistent with previous assessments(eg Schwartz and Ellis [1981] VanVuren and Bray [1983] and Plumband Dodd [1993] in mixed and short-grass prairie) which noted that bothbison and cattle are generalist herbi-vores that graze preferentially ongraminoids Nevertheless some dif-ferences in the foraging patterns ofbison and cattle have been docu-mented that may have long-term im-plications for grasslands For examplebison have a higher proportion ofgraminoids in their diet than docattle consequently forb and browsespecies are more common in cattlediets (Van Vuren and Bray 1983Hartnett et al 1997) Also bisonspend less time grazing than cattleand more time in nonfeeding activi-ties (Plumb and Dodd 1993) andbison strongly prefer open grasslandareas for grazing whereas cattle use

wooded and grassland habitats op-portunistically (Hartnett et al 1997)

Studies that have focused exclu-sively on cattle generally concur thattheir grazing activities increase spa-tial heterogeneity and enhance plantspecies diversity so long as stockingdensity is not too high (Collins 1987Hartnett et al 1996) Because bisongrazing in tallgrass prairie has a simi-lar effect one could conclude thateither herbivore can alter resourceavailability and heterogeneity andreduce the cover of the dominantgrasses sufficiently to enhance thesuccess of the subdominant speciesPerhaps of greater importance thandifferences in foraging patterns be-tween bison and cattle however arethe number of nongrazing activitiessuch as wallowing and horning (ierubbing on trees) that are associatedexclusively with bison (Coppedgeand Shaw 1997 Hartnett et al 1997)These activities when combined withthe spatial redistribution of nutri-ents and selective consumption ofthe dominant grasses may furtherincrease plant species richness andresource heterogeneity particularlyat the landscape scale

Nevertheless it is likely that be-cause bison and cattle are function-ally similar as large grass-feedingherbivores management strategies(stocking intensity and duration) willhave a greater influence on the de-gree of ecological equivalencyachieved than inherent differences inthese ungulates (Hartnett et al1997) Clearly the degree of over-lap in diet and foraging patterns isgreater between bison and cattle thanbetween cattle and other historicallyimportant native herbivores (Hart-nett et al 1997) such as antelope(Antilocapra americana) deer (Odo-coileus virginianus) and elk (Cervuscanadensis) Indeed the loss of ante-lope and elk from the tallgrass prai-rie coupled with dramatic increasesin deer populations presents addi-tional challenges for managing theseecosystems

Conservation implicationsConserving small and moderate-sizedtracts of once-vast biomes such asthe tallgrass prairie presents a uniqueset of problems that are distinct fromthose associated with spatially re-

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

40 BioScience Vol 49 No 1

National Science Foundationrsquos LongTerm Ecological Research (LTER)program and ongoing experimentaltreatments on some parts of the sitedate to 1972 Past syntheses of re-search at Konza Prairie have focusedon the effects of fire and climatic

variability (Knapp and Seastedt1986 Knapp et al 1998b) but amajority of these data were collectedin the absence of bison grazing

In 1987 30 bison were reintro-duced to Konza Prairie and the herdwas allowed to increase until 1992

Since that time the herd has beenmaintained at approximately 200individuals who have had unre-stricted access to a 1012 ha portionof the landscape Within this areaare 10 watersheds that are subjectedto different frequencies of late-springprescribed fire The target animaldensity of 200 animals was selectedso that approximately 25 ofaboveground primary production isconsumed annually This consump-tion rate is approximately half thatof tallgrass prairie managed for do-mestic cattle The bison herd at KonzaPrairie is not provided supplementalfeed in winter nor is it actively man-aged Thus this herd provides anopportunity to document the impactof bison on a native tallgrass prairieecosystem

Although research on bisonndashtallgrass prairie interactions begansoon after bison were reintroducedto Konza Prairie comprehensivestudies spanning scales from the leafto the landscape level began in theearly 1990s In this article we pro-vide the first synthesis of these re-search efforts with the goal of high-lighting the keystone role (sensuPower et al 1996) that bison playedin the tallgrass prairies of the pastWithin this overview we addresstwo general questions What are thedirect and indirect effects of bisonon patterns and processes in tallgrassprairie What factors influence thespatial and temporal patterns of graz-ing activities by bison Perhaps asimportant as addressing these ques-tions the data that we present indi-cate that large-ungulate herbivorycan and should play a key role inthe management and conservationof the remaining tracts of this once-widespread grassland

Bison grazing activitiesand plant responsesLike all large herbivores bison donot graze indiscriminately across thelandscape or even within a local area(Senft et al 1987 Wallace et al1995) Rather they graze in twopatterns in tallgrass prairie creatingboth distinct grazing patches (typi-cally 20ndash50 m2 in Flint Hills tallgrassprairie Catchpole 1996) and moreextensive grazing lawns (larger than400 m2 McNaughton 1984) In both

Figure 1 Grazing by bison increases spatial variability in tallgrass prairie Immediatelyafter fire (April) in grazed watersheds on the Konza Prairie a native tallgrass prairie innortheastern Kansas the uneven consumption of fuel by fire in grazed areas (upperphoto) is caused by the patchiness of bison grazing the previous year Dark areasindicate sites where bison did not graze or where grazing was light and fuel loads weretherefore sufficient to carry a fire through the patch Lighter areas are patches that weregrazed repeatedly by bison such that fuel loads were reduced and the impact of fire wasminimal Later in the season bison graze preferentially in recently burned watersheds(lower photo) In this late spring (May) view of a burned watershed unburnedwatersheds can be seen in the background Bison at Konza Prairie have access to 10watersheds that differ in fire regime

January 1999 41

cases bison revisit grazed sitesthroughout the season such that re-peated defoliation of grazed plants isthe norm and relatively sharp bound-aries between grazed and ungrazedvegetation become evident (Figure1) Bison are primarily graminoidfeeders and consume higher propor-tions of the dominant grasses thanwould be predicted based on grassavailability in the landscape (Pedenet al 1974 Van Vuren and Bray1983 Steuter et al 1995) Bison tendto avoid forbs and woody specieswhich usually constitute less than10 of their diet Thus within abison grazing area forbs are oftenconspicuously left ungrazed and aresurrounded by grazed grasses (Fahne-stock and Knapp 1993 Damhoureyehand Hartnett 1997)

Preferential grazing of the domi-nant grasses by bison sets the stagefor significant alterations in com-petitive interactions among the C4grasses and the C3 forbs Such shiftsare important for plant communitystructure because in ungrazed andfrequently burned prairie a smallgroup of grass species (Andropogongerardii big bluestem Sorghastrumnutans Indian grass Panicumvirgatum switchgrass and Andro-pogon scoparium little bluestem)account for most biomass densityleaf area and resource consumption(Knapp 1985 Briggs and Knapp1995) However it is the species-rich forb component (more than 350species are recorded on Konza Prai-rie Freeman 1998) that is critical forthe maintenance of high levels ofbiotic diversity in tallgrass prairie(Gibson and Hulbert 1987 Glennand Collins 1990 Turner et al 1995)Thus by grazing on grasses and al-lowing forbs to flourish bison havethe potential to significantly influ-ence biodiversity in these grasslands(Collins et al 1998)

The short-term effects of bisonherbivory on the most abundant prai-rie grass A gerardii are differentfrom the long-term effects At theleaf level short-term responses toleaf removal are typical of manygraminoids in grazing systems(McNaughton 1983) Wallace (1990)reported a postgrazing enhancementof photosynthesis in A gerardii inOklahoma Similarly on Konza Prai-rie midseason photosynthetic rates

averaged 53 higher ingrazed tillers (individualgrass stems) than in ungrazedplants with a maximumstimulation of 150 (Figure2a) Mechanisms for this en-hancement of photosynthe-sis include increased lightavailability and reduced wa-ter stress for all species ingrazed patches (Fahnestockand Knapp 1993) and greatertissue nitrogen concentrationin A gerardii leaves as nitro-gen is reallocated from roots

These potential compensatory in-creases in photosynthesis after graz-ing may be augmented by the trans-location of carbon reserves frombelowground to aboveground tissuesVinton and Hartnett (1992) foundthat in the first year of grazinggrowth and biomass of grazed Agerardii tillers had completely com-pensated for the loss to grazing byseasonrsquos end But after several yearsof grazing the ability of tillers tocompensate for lost tissue was re-duced (Figure 2b) Vinton andHartnett (1992) attributed these dif-ferences in short- and long-term re-sponses to reductions in below-ground carbon allocation and storedcarbohydrate reserves after severalyears of grazing Turner et al (1993)also demonstrated the importance of

grazing history on productivity intal lgrass prairie by measuringaboveground primary production ina number of sites with different graz-ing histories They found that com-pensatory regrowth of biomass oc-curred in sites with little history ofgrazing but not in sites that had beengrazed heavily in previous years

Some researchers have argued thatthe inability of tallgrass prairiegrasses to compensate for the bio-mass lost in frequently grazed areasis evidence that prolonged intensivebison grazing did not occur in thesegrasslands (Shaw and Lee 1997) Ifthis hypothesis were true then grassabundance would decline continu-ally following repeated grazingHowever the tallgrass prairiersquos lossof production potential due to graz-ing is short lived At Konza Prairie

Figure 2 Response of tallgrassprairie vegetation to bison her-bivory at three temporal scales(a) Instantaneous leaf-level pho-tosynthetic responses of Andro-pogon gerardii to grazing (ErinQuestad and Alan K Knapp un-published data) (b) Relativegrowth rates of tillers of Agerardii in sites that differ in graz-ing history (data from Vintonand Hartnett 1992) Both un-grazed tillers and grazed tillerswith a history of grazing hadsimilar growth rates whereasgrazed tillers without a history ofgrazing had significantly greatergrowth rates (c) End-of-seasonaboveground biomass inungrazed sites that differ in graz-ing history (data from Knapp etal 1998a) Asterisks in panels (a)and (b) indicate significantly dif-ferent values (P lt 005) for com-parisons at specific dates in panel(c) they indicate comparisonswithin growth forms

42 BioScience Vol 49 No 1

permanent fenced exclosures haveexcluded bison from experimentalplots within grazed watersheds forseveral years When adjacent grazedsites were also protected from graz-ing with temporary exclosuresaboveground production in the firstyear in these newly protected siteswas reduced relative to that in theadjoining long-term ungrazed areas(Figure 2c) however these sites re-covered their production potentialby the second year Thus produc-tion potential can recover if bisongrazing is sufficiently dynamic ei-ther spatially or temporally suchthat sites are grazed intermittently

Removal of grass leaf area by bi-son and reductions in the capabilityof the dominant grasses to compen-sate for tissue lost after multipleyears of grazing suggest that the co-occurring subdominant forbs maybenefit from bison grazing Indeedcomparisons of forbs inside grazingpatches with those in adjacentungrazed prairie have shown thatgas exchange and aboveground bio-mass production density and cover

can be enhanced by the selective con-sumption of grasses by bison (Fahn-estock and Knapp 1993 Hartnett etal 1996 Damhoureyeh and Hartnett1997) Given the importance of pastgrazing pressures to the direct re-sponses of grasses to herbivory andthe indirect responses of forbs iden-tifying those factors that influencethe selection and reselection of graz-ing patches by bison in tallgrass prai-rie is critical for understanding thelong-term consequences of bisongrazing patterns

Factors influencing bisonselection of grazing sitesHistorical information regarding bi-son grazing patterns in the GreatPlains is replete with anecdotal ac-counts of herds attracted to recentlyburned grasslands (Figure 1McHugh 1972 Pyne 1982) butquantitative evidence of this prefer-ence in tallgrass prairie has beenlacking until recently (Coppedge andShaw 1998) As noted earlier bisonat Konza Prairie have free access to

10 watersheds that are subject todifferent fire frequencies Since 1991twice-weekly observations of the dis-tribution of bison within this areahave been made to assess patterns ofbison grazing (Nellis et al 1992)The results confirm that bison dograze preferentially within burnedwatersheds from April (burning takesplace in late Marchndashearly April)through June and July and in someyears through August (Figure 3Vinton et al 1993 Nellis and Briggs1997) In addition to grazing prefer-entially in burned sites bison in-crease their selective consumption ofsome grass species in burned sitesrelative to unburned sites (Pfeifferand Hartnett 1995) Late in the sum-mer lowland topographic positionswith deeper soils (and thereforegreater soil moisture and plant pro-ductivity Knapp et al 1993) inburned watersheds become preferredgrazing locations as the uplands dryThis preference for burned areas intallgrass prairie is consistent withpostfire responses in mixed grass prai-rie (Coppock and Detling 1986) aswell as with large ungulatesrsquo winterpreference for burned sites in the north-ern mixed grasslands of YellowstoneNational Park (Pearson et al 1995)

Within a watershed or at a spe-cific topographic position in tallgrassprairie several factors may influ-ence initial patch selection andreselection (see also Wallace et al1995) In addition patch selectionhas several long-term consequencesWhen bison were first reintroducedto Konza Prairie they encountered amosaic of burned and unburned wa-tersheds with significant differencesamong watersheds in the spatial het-erogeneity of plant community com-position For example frequentlyburned but ungrazed watersheds aredominated by C4 grasses and havelow species richness and diversitywhereas less frequently burned siteshave higher species richness and forbcover (Gibson and Hulbert 1987Collins 1992) Initial studies onKonza Prairie indicated clearly thatbison established grazing patches inareas strongly dominated by C4grasses and that these patches werereselected at a high rate (Vinton etal 1993) Six years later a survey offloristic composition indicated thatestablished bison grazing patches had

Figure 3 Spatial distribution of bison in 10 watersheds in the central portion of KonzaPrairie after the completion of spring burning During the months of April May andJune 1992 bison locations were determined twice per week and recorded on large-scalespatially rectified aerial photographs overlaid with a 30-meter grid Bison wereobserved most frequently in the recently burned watersheds B burned watersheds Uunburned watersheds

January 1999 43

a higher abundance of forbs and alower cover of grasses than adjacentungrazed patches (Catchpole 1996)Similar small-scale patterns of forband grass abundance were observedby comparing the floristic composi-tion inside and outside grazingexclosures in grazed watersheds atKonza Prairie (Hartnett et al 1996)These observations suggest that bi-son alter plant community composi-tion at the patch scale by selectingspecies-poor grass-dominated sitesand converting them to sites of lo-cally higher diversity (Figure 4)

A second factor that may influ-ence patch selection and reselectionby bison is plant quality The foliar

nitrogen content of plants is highlyvariable both spatially and tempo-rally Bison contribute to this patchi-ness through deposition of nitrogen-rich urine Steinauer (1994) appliedsynthetic bovine urine at randomlyselected locations along eighttransects at Konza Prairie (four eachin grazed and ungrazed areas) Inungrazed areas grass cover was sig-nificantly higher in plots that werefertilized with urine than in plotswithout urine But in the grazed areaby contrast grass cover was signifi-cantly lower on the urine-treatedplots than in plots without urinebecause bison preferentially grazedthe grasses on the urine-treated plots

Day and Detling (1990) have shownthat grasses growing on urine patchesin mixed-grass prairie have higherleaf nitrogen content and are there-fore more nutritious per bite thangrasses growing on patches withouturine Not only was grass cover loweron urine patches at Konza Prairiebut the total area of grazed patcheson the urine-treated transects wassignificantly larger than the area ofgrazed patches on transects that werenot treated with urine (Steinauer1994) Thus the enhanced produc-tivity of grasses growing on urinepatches represents a potential stimu-lus for the initiation and reselectionof grazing patches by bison

Figure 4 Conceptual model of the spatial and temporal dynamics of bison grazing activities and the responses of tallgrass prairieto the reintroduction of bison Before the reintroduction of bison watershed attributes at Konza Prairie differed strongly dependingon the fire regime imposed with frequently burned sites notable for their high productivity but low plant species diversity Selectivebison grazing at the watershed level (preference for burned sites particularly in the spring) and the patch level (selection of patchesdominated by C4 grasses) has led to increased similarity in watershed attributes despite differences in fire history Moreoversignificant increases in plant species diversity and spatial heterogeneity have been noted in all watersheds Superscripts denote studieson which this synthesis is based 1 Nellis and Briggs (1997) 2 Vinton et al (1993) 3 Briggs and Knapp (1995) 4 Collins (1992)Collins et al (1995) 5 Vinton and Hartnett (1992) Vinton et al (1993) 6 Hartnett et al (1997) 7 Fahnestock and Knapp (19931994) 8 Catchpole (1996) 9 Collins and Steinauer (1998) 10 Hartnett et al (1996)

44 BioScience Vol 49 No 1

Given that bison prefer grazingpatches that are initially dominatedby C4 grasses but that their selectiveforaging and reselection habits con-vert these patches to sites with agreater abundance of nonforage spe-cies it is likely that patch locationsare spatially dynamic across the land-scape Patches can ldquomoverdquo by twomechanisms patch abandonmentfollowed by selection of a new patchor patch migration wherein portionsof a patch are abandoned and thepatch expands into adjacent areasOver a 3-year period Catchpole(1996) found the rate of patch aban-donment to be approximately 6ndash7per year in both burned and un-burned watersheds thus at leastportions of previously establishedgrazing patches were reselected at ahigh rate (Figure 4) However whengrazing patches were mapped andcompared across years the extent ofspatial reselectionmdashalthough highlyvariable due to differences in totalburned area available to bison in anyone yearmdashaveraged approximately50 per year Thus grazing patchesin both burned and unburned water-sheds appear to migrate significantlyfrom year to year This local migra-tion permits periodic release of por-tions of the grassland from grazingpressures (Figure 4) and provides amechanism for recovery of below-ground carbohydrate storage reservesand production potential

At the watershed and landscapescales the long-term consequencesof bison activities include a reductionin cover dominance and productivityof grasses the competitive release ofmany subdominant species resultingin an increase in the abundance offorbs an overall increase in plantspecies richness and diversity andincreased spatial heterogeneity (Fig-ure 5 Hartnett et al 1996) Althoughalterations in plant community com-position can be attributed in largepart to the direct effects of grazingby bison increased plant species rich-ness is also likely to be a product ofincreased microsite diversity gener-ated by nongrazing activities suchas dung and urine deposition tram-pling and wallowing These andother bison activities contribute sig-nificantly to the increase in spatialheterogeneity that is characteristicof grazed tallgrass prairie (Figure 5)

Other impacts of bisonin tallgrass prairie

Effects of ungulates especially graz-ing on plant community composi-tion and structure have been studiedin many grasslands worldwide(McNaughton 1984 Milchunas etal 1988 Frank and McNaughton1992 Milchunas and Lauenroth1993) Ungulate activities howeveraffect many other aspects of grass-land structure and function includingthe physical structure of the environ-ment and the rates of a number ofecosystem-level processes (McNaugh-ton 1993 Frank and Evans 1997McNaughton et al 1997) There areseveral other important mechanismsby which bison alter ecosystem-levelprocesses and physical habitat struc-ture in tallgrass prairie

Nutrient redistribution and cyclingBison can substantially alter nutri-ent cycling processes and patterns ofnutrient availability in tallgrass prai-rie Their effects on nitrogen cyclingare critical because nitrogen avail-ability often limits plant productiv-ity in these grasslands (Seastedt et al1991 Blair 1997 Turner et al 1997)and influences plant species compo-sition (Gibson et al 1993 Wedinand Tilman 1993) Simulation mod-els of tallgrass prairie responses tograzing (Risser and Parton 1982)and studies of grazers in other grass-lands (Frank and Evans 1997McNaughton et al 1997) have dem-onstrated a disproportionate influ-ence of ungulates including bisonon the regulation of nitrogen cyclingprocesses Preliminary data fromKonza Prairie suggest that bison aresimilarly important in controllingnitrogen cycling in tallgrass prairie

Bison influence nitrogen cyclingconservation and availability intallgrass prairie ecosystems by alter-ing several soil and plant processesUngulates in grasslands consumerelatively recalcitrant plant biomassand return labile forms of nitrogen(ie urine) to soils (Ruess andMcNaughton 1988) thus bypassingthe otherwise slow mineralization ofnitrogen in plant litter Nitrogen inbison urine is largely urea which canbe hydrolyzed to ammonium in amatter of days (Ruess and McNaugh-ton 1988) Indeed application of

synthetic bison urine increased con-centrations of ammonium and ni-trate in Konza Prairie soils over 130-fold and 30-fold respectively 8 daysafter application (J R Matchett andLoretta C Johnson unpublished data)

Bison grazing can decrease theexport of nitrogen from tallgrassprairie by altering the magnitudes oftwo major pathways of nitrogenlossmdashcombustion and ammoniavolatilization Fire is the major path-way of nitrogen loss from ungrazedtallgrass prairie (Dodds et al 1996Blair 1997) nitrogen loss from burn-ing averages 1ndash4 gmiddotmndash2middotyrndash1 (Blair etal 1998) Grazing lowers combus-tion losses of nitrogen in tallgrassprairie by reducing the abovegroundplant detritus and increasing thepatchiness of a prairie fire (Figure 2Hobbs et al 1991) Although vola-tilization of ammonia can be in-creased by grazing in other types ofgrassland (Detling 1988) Hobbs etal (1991) suggested that any increasein ammonia volatilization in tallgrassprairie will be more than compen-sated for by a reduction in combus-tion losses of nitrogen

Finally bison grazing affects theamount and quality of plant litter re-turned to soils Grazing increases plantuptake of nutrients (Ruess 1984) andshoot nitrogen content in many grass-lands (Holland and Detling 1990Milchunas et al 1995) includingtallgrass prairie (Turner et al 1993)However the effects of grazers onroot growth and chemistry varyamong grasslands (Milchunas andLauenroth 1993) On Konza Prairieroot productivity and root biomasswere 30 and 20 lower respec-tively in bison grazing lawns than inungrazed exclosures In addition thenitrogen concentration of new rootgrowth in bison grazing lawns atKonza Prairie increased significantlyfrom 06 to 09 and the CN ratioof roots decreased A lower CN ratioreduces microbial immobilization andenhances nitrogen availability withingrazed areas Indeed recent studieson Konza Prairie indicated that netnitrogen mineralization in bison graz-ing lawns was 153 greater and netnitrification 126 greater than inungrazed prairie (Figure 6) Further-more net nitrogen mineralization rateswere proportional to the intensity ofbison use of a given area Thus the

January 1999 45

net effect of bison grazing appears tobe increased rates of nitrogen cyclingcoupled with a significant increase inspatial heterogeneity in nitrogen avail-ability together these effects can alterpatterns of plant productivity and spe-cies composition in tallgrass prairie(Figure 5 Steinauer and Collins 1995)

Wallowing One aspect of bison be-havior that differs from that of cattleand is primarily a physical activityis wallowing Wallows in Flint Hillstallgrass prairie which are estab-lished primarily in level upland orlowland sites dramatically alter thepatch structure of this prairie Bisonwallows develop as the animals pawthe ground and roll in the exposedsoil Continued use of wallows bybulls cows and calves creates a soildepression of 3ndash5 m in diameter (and10ndash30 cm in depth) that is devoid ofvegetation These denuded patcheseither gradually revegetate or remainas bare soil depending on the fre-quency of revisitation by bison Withthe vast numbers of bison that onceoccupied the Great Plains these soildepressions were probably abundantand widespread features of the land-scape (England and DeVos 1969)For example a number of relic wal-lows had to be filled to level the play-ing field for the first University ofOklahoma home football game in 1895(University of Oklahoma Athletic De-partment 1986) Relic wallows stillexist in many areas where bison havenot occurred in the past 125 years

Environmental conditions in relicand newly established wallowsstrongly influence prairie patch dy-namics (Polley and Collins 1984Polley and Wallace 1986) Becauseof soil compaction wallows oftenretain rainwater in the spring creat-ing localized habitats that are suit-able for ephemeral wetland speciessimilar to vernal pools in California(Holland and Jain 1981 Uno 1989)In the summer however the samewallows support only plants that cantolerate severe drought Vegetationcomposition and structure in wal-lows is different from that in thesurrounding prairie (Polley andCollins 1984) and these differencesare enhanced by fire (Collins andUno 1983) which may not spreadthrough wallows because of low fuelloads Consequently at larger spa-

tial scales grazed prairie that con-tains bison wallows has higher plantspecies diversity than grazed prairiewithout wallows (Collins and Bar-ber 1985) Thus bison can physi-cally alter grasslands in ways thatincrease environmental heterogeneityand enhance both local and regionalbiodiversity (Hartnett et al 1997)

Bison carcasses Bison not only af-fect vegetation patterns and soil pro-cesses through their grazing activi-ties but also have profound andlasting localized effects after theydie Although legal requirements andmanagement practices dictate theremoval of carcasses of domesticherbivores from public and privategrasslands native herbivores rou-tinely die of natural causes and theirbodies remain in situ As part of theminimal management strategy atKonza Prairie bison that die on siteare not removed As a result thesecarcasses create unique local distur-bances (Figure 7) that are the focusof studies to assess their effects onsoil nutrients and vegetation re-sponses in tallgrass prairie

When an individual bison diescopious quantities of fluids (withhigh nitrogen concentration) are re-leased during decomposition Adultbison can weigh more than 800 kgand these carcasses typically kill un-derlying and adjacent plants creat-ing a denuded zone of 4ndash6 m2 (Figure7) Although the fluids that are ini-tially released are toxic to vegeta-tion these sites eventually becomezones of high fertility For examplesoil cores extracted from the centerof carcass sites on Konza Prairie 3years after death had inorganic ni-trogen concentrations that were twoto three times higher than the sur-rounding prairie This nutrient enrich-ment may extend up to 25 m awayfrom the original carcass site andresults in patches dominated initiallyby early successional species Theaboveground primary production inthese patches is two to three timeshigher than in undisturbed prairie

Although disturbances created bybison carcasses are sporadic and lo-calized on Konza Prairie they pro-vide nutrient pulses that exceed allother natural processes even urineand fecal deposits Overall we canonly speculate about historical rates

of bison mortality Given the enor-mous size of the bison populationbefore their widespread slaughter inthe 1800s annual mortality wasprobably high High death rateswould have been especially commonduring droughts when there wouldbe the potential for large numbers ofcarcasses to occur across the land-scape Even though predators andscavengers may have consumed andrelocated many of these carcassesdecomposition of the remaining andpartial carcasses would still have re-sulted in patches of locally high nu-trient concentration Thus althoughit was variable bison mortality wouldhave led to a continual cycle of dis-turbance and recovery of thesepatches in presettlement grasslands

Are bison keystone speciesThe net effects of selective bison graz-ing activities at the landscape patchand individual plant level includeshifts in plant species compositionalterations of the physical and chemi-cal environment and increased spa-tial and temporal heterogeneity invegetation structure soil resourceavailability and a variety of ecosys-tem processes (Figures 4 5 and 6)Before bison reintroduction at KonzaPrairie the long-term burning ex-periments produced clear patterns ofresponse in the vegetation As firefrequency increased the dominanceof C4 grasses increased and the coverof C3 grasses forbs and woody spe-cies decreased (Figure 4 Gibson andHulbert 1987) Overall plant spe-cies diversity declined as fire fre-quency increased in ungrazed tall-grass prairie (Collins et al 1995)

These patterns in community struc-ture which had developed over 20years of burning treatments at KonzaPrairie are being rapidly and dra-matically altered by the grazing ac-tivity of the reintroduced bison Inparticular grazing by bison has low-ered the abundance of the dominant C4

grasses increased the abundance ofthe subdominant C3 grasses and forbsand markedly increased plant speciesdiversity (by 23) richness (by 38)and community heterogeneity (by13) relative to ungrazed sites evenunder annual burning conditions(Hartnett et al 1996 Collins andSteinauer 1998 Collins et al 1998)

46 BioScience Vol 49 No 1

Because of the multiple and dra-matic effects of bison on this land-scape we believe that bison are key-stone species in the tallgrass prairieOther authors have noted the poten-tial of large grazing mammals to actas ldquokeystone herbivoresrdquo capable ofmaintaining open grassland vegeta-tion that would otherwise undergosuccession to shrubland or wood-land (Owen-Smith 1987) Indeed thedisappearance of a grazing mega-fauna at the end of the Pleistocenemay have played a major role in thewidespread transition from steppe totundra at that time (Zimov et al 1995)However the concept of keystonespecies has been controversial since itsinception (Power et al 1996) One ofthe problems with this concept hasbeen the variable interpretation of cri-teria by which species are determinedto be keystone Power et al (1996)consider a keystone species to be ldquoonewhose impact on its community orecosystem is large and disproportion-ately large relative to its abundancerdquoTo make this definition operationalthese authors proposed a measure ofcommunity importance (CI) to be usedas an index of the strength of theimpact of a given species

CI = [(tN ndash tD)tN][1pi]

where tN is a quantitative measure ofa trait (eg diversity) in an intactcommunity tD is the measure of thetrait when species i has been deletedand pi is the proportional abundance(biomass) of species i before it wasdeleted CI values ldquomuch greater than1rdquo indicate that a species is key-stone We estimate bison biomass atKonza Prairie to be approximately11ndash12 gmiddotm2 (Collins and Steinauer1998) which is approximately 1of the total vegetative biomass OnKonza Prairie diversity is signifi-cantly higher (10ndash33) on grazedsites than ungrazed sites (Hartnett etal 1996) and these values yield arange of CIs from 6 to 25 For thisreason and others we consider bisonto be keystone species in tallgrassprairie ecosystems

Are bison and cattle functionalequivalents in tallgrass prairieThe historical presence of immenseherds of large ungulates in Great

Plains grasslands is undisputed(McHugh 1972) and we have em-phasized the keystone role that bisonplayed in determining the structureand function of tallgrass prairies atmultiple spatial and temporal scalesWith the replacement of native bisonby domesticated cattle in the remain-ing grasslands an obvious issue isthe degree of similarity between thesetwo ungulates with respect to theireffects on tallgrass prairie In otherwords can bison and cattle be con-sidered ecological equivalents

There have been several previousattempts to answer this questionbut the results have been equivocalat best (Plumb and Dodd 1993Hartnett et al 1997) The primarybarrier to resolving this issue restswith a lack of comparative studies inwhich management is held constantand the type of grazer is varied Suchstudies have recently been initiatedat Konza Prairie Results after 3 yearsindicate that the abundance and rich-ness of annual forbs and the spatialheterogeneity of biomass and coverare higher in sites with bison than insites with cattle No dramatic differ-ences have been detected howeverbetween cattle- and bison-grazed sitesin cover of the dominant C4 grass Agerardii or the dominant forb Am-brosia psilostachya total plant spe-cies richness is also not dramaticallydifferent (E Gene Towne and DavidC Hartnett unpublished data)

Results at Konza Prairie are con-sistent with previous assessments(eg Schwartz and Ellis [1981] VanVuren and Bray [1983] and Plumband Dodd [1993] in mixed and short-grass prairie) which noted that bothbison and cattle are generalist herbi-vores that graze preferentially ongraminoids Nevertheless some dif-ferences in the foraging patterns ofbison and cattle have been docu-mented that may have long-term im-plications for grasslands For examplebison have a higher proportion ofgraminoids in their diet than docattle consequently forb and browsespecies are more common in cattlediets (Van Vuren and Bray 1983Hartnett et al 1997) Also bisonspend less time grazing than cattleand more time in nonfeeding activi-ties (Plumb and Dodd 1993) andbison strongly prefer open grasslandareas for grazing whereas cattle use

wooded and grassland habitats op-portunistically (Hartnett et al 1997)

Studies that have focused exclu-sively on cattle generally concur thattheir grazing activities increase spa-tial heterogeneity and enhance plantspecies diversity so long as stockingdensity is not too high (Collins 1987Hartnett et al 1996) Because bisongrazing in tallgrass prairie has a simi-lar effect one could conclude thateither herbivore can alter resourceavailability and heterogeneity andreduce the cover of the dominantgrasses sufficiently to enhance thesuccess of the subdominant speciesPerhaps of greater importance thandifferences in foraging patterns be-tween bison and cattle however arethe number of nongrazing activitiessuch as wallowing and horning (ierubbing on trees) that are associatedexclusively with bison (Coppedgeand Shaw 1997 Hartnett et al 1997)These activities when combined withthe spatial redistribution of nutri-ents and selective consumption ofthe dominant grasses may furtherincrease plant species richness andresource heterogeneity particularlyat the landscape scale

Nevertheless it is likely that be-cause bison and cattle are function-ally similar as large grass-feedingherbivores management strategies(stocking intensity and duration) willhave a greater influence on the de-gree of ecological equivalencyachieved than inherent differences inthese ungulates (Hartnett et al1997) Clearly the degree of over-lap in diet and foraging patterns isgreater between bison and cattle thanbetween cattle and other historicallyimportant native herbivores (Hart-nett et al 1997) such as antelope(Antilocapra americana) deer (Odo-coileus virginianus) and elk (Cervuscanadensis) Indeed the loss of ante-lope and elk from the tallgrass prai-rie coupled with dramatic increasesin deer populations presents addi-tional challenges for managing theseecosystems

Conservation implicationsConserving small and moderate-sizedtracts of once-vast biomes such asthe tallgrass prairie presents a uniqueset of problems that are distinct fromthose associated with spatially re-

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

January 1999 41

cases bison revisit grazed sitesthroughout the season such that re-peated defoliation of grazed plants isthe norm and relatively sharp bound-aries between grazed and ungrazedvegetation become evident (Figure1) Bison are primarily graminoidfeeders and consume higher propor-tions of the dominant grasses thanwould be predicted based on grassavailability in the landscape (Pedenet al 1974 Van Vuren and Bray1983 Steuter et al 1995) Bison tendto avoid forbs and woody specieswhich usually constitute less than10 of their diet Thus within abison grazing area forbs are oftenconspicuously left ungrazed and aresurrounded by grazed grasses (Fahne-stock and Knapp 1993 Damhoureyehand Hartnett 1997)

Preferential grazing of the domi-nant grasses by bison sets the stagefor significant alterations in com-petitive interactions among the C4grasses and the C3 forbs Such shiftsare important for plant communitystructure because in ungrazed andfrequently burned prairie a smallgroup of grass species (Andropogongerardii big bluestem Sorghastrumnutans Indian grass Panicumvirgatum switchgrass and Andro-pogon scoparium little bluestem)account for most biomass densityleaf area and resource consumption(Knapp 1985 Briggs and Knapp1995) However it is the species-rich forb component (more than 350species are recorded on Konza Prai-rie Freeman 1998) that is critical forthe maintenance of high levels ofbiotic diversity in tallgrass prairie(Gibson and Hulbert 1987 Glennand Collins 1990 Turner et al 1995)Thus by grazing on grasses and al-lowing forbs to flourish bison havethe potential to significantly influ-ence biodiversity in these grasslands(Collins et al 1998)

The short-term effects of bisonherbivory on the most abundant prai-rie grass A gerardii are differentfrom the long-term effects At theleaf level short-term responses toleaf removal are typical of manygraminoids in grazing systems(McNaughton 1983) Wallace (1990)reported a postgrazing enhancementof photosynthesis in A gerardii inOklahoma Similarly on Konza Prai-rie midseason photosynthetic rates

averaged 53 higher ingrazed tillers (individualgrass stems) than in ungrazedplants with a maximumstimulation of 150 (Figure2a) Mechanisms for this en-hancement of photosynthe-sis include increased lightavailability and reduced wa-ter stress for all species ingrazed patches (Fahnestockand Knapp 1993) and greatertissue nitrogen concentrationin A gerardii leaves as nitro-gen is reallocated from roots

These potential compensatory in-creases in photosynthesis after graz-ing may be augmented by the trans-location of carbon reserves frombelowground to aboveground tissuesVinton and Hartnett (1992) foundthat in the first year of grazinggrowth and biomass of grazed Agerardii tillers had completely com-pensated for the loss to grazing byseasonrsquos end But after several yearsof grazing the ability of tillers tocompensate for lost tissue was re-duced (Figure 2b) Vinton andHartnett (1992) attributed these dif-ferences in short- and long-term re-sponses to reductions in below-ground carbon allocation and storedcarbohydrate reserves after severalyears of grazing Turner et al (1993)also demonstrated the importance of

grazing history on productivity intal lgrass prairie by measuringaboveground primary production ina number of sites with different graz-ing histories They found that com-pensatory regrowth of biomass oc-curred in sites with little history ofgrazing but not in sites that had beengrazed heavily in previous years

Some researchers have argued thatthe inability of tallgrass prairiegrasses to compensate for the bio-mass lost in frequently grazed areasis evidence that prolonged intensivebison grazing did not occur in thesegrasslands (Shaw and Lee 1997) Ifthis hypothesis were true then grassabundance would decline continu-ally following repeated grazingHowever the tallgrass prairiersquos lossof production potential due to graz-ing is short lived At Konza Prairie

Figure 2 Response of tallgrassprairie vegetation to bison her-bivory at three temporal scales(a) Instantaneous leaf-level pho-tosynthetic responses of Andro-pogon gerardii to grazing (ErinQuestad and Alan K Knapp un-published data) (b) Relativegrowth rates of tillers of Agerardii in sites that differ in graz-ing history (data from Vintonand Hartnett 1992) Both un-grazed tillers and grazed tillerswith a history of grazing hadsimilar growth rates whereasgrazed tillers without a history ofgrazing had significantly greatergrowth rates (c) End-of-seasonaboveground biomass inungrazed sites that differ in graz-ing history (data from Knapp etal 1998a) Asterisks in panels (a)and (b) indicate significantly dif-ferent values (P lt 005) for com-parisons at specific dates in panel(c) they indicate comparisonswithin growth forms

42 BioScience Vol 49 No 1

permanent fenced exclosures haveexcluded bison from experimentalplots within grazed watersheds forseveral years When adjacent grazedsites were also protected from graz-ing with temporary exclosuresaboveground production in the firstyear in these newly protected siteswas reduced relative to that in theadjoining long-term ungrazed areas(Figure 2c) however these sites re-covered their production potentialby the second year Thus produc-tion potential can recover if bisongrazing is sufficiently dynamic ei-ther spatially or temporally suchthat sites are grazed intermittently

Removal of grass leaf area by bi-son and reductions in the capabilityof the dominant grasses to compen-sate for tissue lost after multipleyears of grazing suggest that the co-occurring subdominant forbs maybenefit from bison grazing Indeedcomparisons of forbs inside grazingpatches with those in adjacentungrazed prairie have shown thatgas exchange and aboveground bio-mass production density and cover

can be enhanced by the selective con-sumption of grasses by bison (Fahn-estock and Knapp 1993 Hartnett etal 1996 Damhoureyeh and Hartnett1997) Given the importance of pastgrazing pressures to the direct re-sponses of grasses to herbivory andthe indirect responses of forbs iden-tifying those factors that influencethe selection and reselection of graz-ing patches by bison in tallgrass prai-rie is critical for understanding thelong-term consequences of bisongrazing patterns

Factors influencing bisonselection of grazing sitesHistorical information regarding bi-son grazing patterns in the GreatPlains is replete with anecdotal ac-counts of herds attracted to recentlyburned grasslands (Figure 1McHugh 1972 Pyne 1982) butquantitative evidence of this prefer-ence in tallgrass prairie has beenlacking until recently (Coppedge andShaw 1998) As noted earlier bisonat Konza Prairie have free access to

10 watersheds that are subject todifferent fire frequencies Since 1991twice-weekly observations of the dis-tribution of bison within this areahave been made to assess patterns ofbison grazing (Nellis et al 1992)The results confirm that bison dograze preferentially within burnedwatersheds from April (burning takesplace in late Marchndashearly April)through June and July and in someyears through August (Figure 3Vinton et al 1993 Nellis and Briggs1997) In addition to grazing prefer-entially in burned sites bison in-crease their selective consumption ofsome grass species in burned sitesrelative to unburned sites (Pfeifferand Hartnett 1995) Late in the sum-mer lowland topographic positionswith deeper soils (and thereforegreater soil moisture and plant pro-ductivity Knapp et al 1993) inburned watersheds become preferredgrazing locations as the uplands dryThis preference for burned areas intallgrass prairie is consistent withpostfire responses in mixed grass prai-rie (Coppock and Detling 1986) aswell as with large ungulatesrsquo winterpreference for burned sites in the north-ern mixed grasslands of YellowstoneNational Park (Pearson et al 1995)

Within a watershed or at a spe-cific topographic position in tallgrassprairie several factors may influ-ence initial patch selection andreselection (see also Wallace et al1995) In addition patch selectionhas several long-term consequencesWhen bison were first reintroducedto Konza Prairie they encountered amosaic of burned and unburned wa-tersheds with significant differencesamong watersheds in the spatial het-erogeneity of plant community com-position For example frequentlyburned but ungrazed watersheds aredominated by C4 grasses and havelow species richness and diversitywhereas less frequently burned siteshave higher species richness and forbcover (Gibson and Hulbert 1987Collins 1992) Initial studies onKonza Prairie indicated clearly thatbison established grazing patches inareas strongly dominated by C4grasses and that these patches werereselected at a high rate (Vinton etal 1993) Six years later a survey offloristic composition indicated thatestablished bison grazing patches had

Figure 3 Spatial distribution of bison in 10 watersheds in the central portion of KonzaPrairie after the completion of spring burning During the months of April May andJune 1992 bison locations were determined twice per week and recorded on large-scalespatially rectified aerial photographs overlaid with a 30-meter grid Bison wereobserved most frequently in the recently burned watersheds B burned watersheds Uunburned watersheds

January 1999 43

a higher abundance of forbs and alower cover of grasses than adjacentungrazed patches (Catchpole 1996)Similar small-scale patterns of forband grass abundance were observedby comparing the floristic composi-tion inside and outside grazingexclosures in grazed watersheds atKonza Prairie (Hartnett et al 1996)These observations suggest that bi-son alter plant community composi-tion at the patch scale by selectingspecies-poor grass-dominated sitesand converting them to sites of lo-cally higher diversity (Figure 4)

A second factor that may influ-ence patch selection and reselectionby bison is plant quality The foliar

nitrogen content of plants is highlyvariable both spatially and tempo-rally Bison contribute to this patchi-ness through deposition of nitrogen-rich urine Steinauer (1994) appliedsynthetic bovine urine at randomlyselected locations along eighttransects at Konza Prairie (four eachin grazed and ungrazed areas) Inungrazed areas grass cover was sig-nificantly higher in plots that werefertilized with urine than in plotswithout urine But in the grazed areaby contrast grass cover was signifi-cantly lower on the urine-treatedplots than in plots without urinebecause bison preferentially grazedthe grasses on the urine-treated plots

Day and Detling (1990) have shownthat grasses growing on urine patchesin mixed-grass prairie have higherleaf nitrogen content and are there-fore more nutritious per bite thangrasses growing on patches withouturine Not only was grass cover loweron urine patches at Konza Prairiebut the total area of grazed patcheson the urine-treated transects wassignificantly larger than the area ofgrazed patches on transects that werenot treated with urine (Steinauer1994) Thus the enhanced produc-tivity of grasses growing on urinepatches represents a potential stimu-lus for the initiation and reselectionof grazing patches by bison

Figure 4 Conceptual model of the spatial and temporal dynamics of bison grazing activities and the responses of tallgrass prairieto the reintroduction of bison Before the reintroduction of bison watershed attributes at Konza Prairie differed strongly dependingon the fire regime imposed with frequently burned sites notable for their high productivity but low plant species diversity Selectivebison grazing at the watershed level (preference for burned sites particularly in the spring) and the patch level (selection of patchesdominated by C4 grasses) has led to increased similarity in watershed attributes despite differences in fire history Moreoversignificant increases in plant species diversity and spatial heterogeneity have been noted in all watersheds Superscripts denote studieson which this synthesis is based 1 Nellis and Briggs (1997) 2 Vinton et al (1993) 3 Briggs and Knapp (1995) 4 Collins (1992)Collins et al (1995) 5 Vinton and Hartnett (1992) Vinton et al (1993) 6 Hartnett et al (1997) 7 Fahnestock and Knapp (19931994) 8 Catchpole (1996) 9 Collins and Steinauer (1998) 10 Hartnett et al (1996)

44 BioScience Vol 49 No 1

Given that bison prefer grazingpatches that are initially dominatedby C4 grasses but that their selectiveforaging and reselection habits con-vert these patches to sites with agreater abundance of nonforage spe-cies it is likely that patch locationsare spatially dynamic across the land-scape Patches can ldquomoverdquo by twomechanisms patch abandonmentfollowed by selection of a new patchor patch migration wherein portionsof a patch are abandoned and thepatch expands into adjacent areasOver a 3-year period Catchpole(1996) found the rate of patch aban-donment to be approximately 6ndash7per year in both burned and un-burned watersheds thus at leastportions of previously establishedgrazing patches were reselected at ahigh rate (Figure 4) However whengrazing patches were mapped andcompared across years the extent ofspatial reselectionmdashalthough highlyvariable due to differences in totalburned area available to bison in anyone yearmdashaveraged approximately50 per year Thus grazing patchesin both burned and unburned water-sheds appear to migrate significantlyfrom year to year This local migra-tion permits periodic release of por-tions of the grassland from grazingpressures (Figure 4) and provides amechanism for recovery of below-ground carbohydrate storage reservesand production potential

At the watershed and landscapescales the long-term consequencesof bison activities include a reductionin cover dominance and productivityof grasses the competitive release ofmany subdominant species resultingin an increase in the abundance offorbs an overall increase in plantspecies richness and diversity andincreased spatial heterogeneity (Fig-ure 5 Hartnett et al 1996) Althoughalterations in plant community com-position can be attributed in largepart to the direct effects of grazingby bison increased plant species rich-ness is also likely to be a product ofincreased microsite diversity gener-ated by nongrazing activities suchas dung and urine deposition tram-pling and wallowing These andother bison activities contribute sig-nificantly to the increase in spatialheterogeneity that is characteristicof grazed tallgrass prairie (Figure 5)

Other impacts of bisonin tallgrass prairie

Effects of ungulates especially graz-ing on plant community composi-tion and structure have been studiedin many grasslands worldwide(McNaughton 1984 Milchunas etal 1988 Frank and McNaughton1992 Milchunas and Lauenroth1993) Ungulate activities howeveraffect many other aspects of grass-land structure and function includingthe physical structure of the environ-ment and the rates of a number ofecosystem-level processes (McNaugh-ton 1993 Frank and Evans 1997McNaughton et al 1997) There areseveral other important mechanismsby which bison alter ecosystem-levelprocesses and physical habitat struc-ture in tallgrass prairie

Nutrient redistribution and cyclingBison can substantially alter nutri-ent cycling processes and patterns ofnutrient availability in tallgrass prai-rie Their effects on nitrogen cyclingare critical because nitrogen avail-ability often limits plant productiv-ity in these grasslands (Seastedt et al1991 Blair 1997 Turner et al 1997)and influences plant species compo-sition (Gibson et al 1993 Wedinand Tilman 1993) Simulation mod-els of tallgrass prairie responses tograzing (Risser and Parton 1982)and studies of grazers in other grass-lands (Frank and Evans 1997McNaughton et al 1997) have dem-onstrated a disproportionate influ-ence of ungulates including bisonon the regulation of nitrogen cyclingprocesses Preliminary data fromKonza Prairie suggest that bison aresimilarly important in controllingnitrogen cycling in tallgrass prairie

Bison influence nitrogen cyclingconservation and availability intallgrass prairie ecosystems by alter-ing several soil and plant processesUngulates in grasslands consumerelatively recalcitrant plant biomassand return labile forms of nitrogen(ie urine) to soils (Ruess andMcNaughton 1988) thus bypassingthe otherwise slow mineralization ofnitrogen in plant litter Nitrogen inbison urine is largely urea which canbe hydrolyzed to ammonium in amatter of days (Ruess and McNaugh-ton 1988) Indeed application of

synthetic bison urine increased con-centrations of ammonium and ni-trate in Konza Prairie soils over 130-fold and 30-fold respectively 8 daysafter application (J R Matchett andLoretta C Johnson unpublished data)

Bison grazing can decrease theexport of nitrogen from tallgrassprairie by altering the magnitudes oftwo major pathways of nitrogenlossmdashcombustion and ammoniavolatilization Fire is the major path-way of nitrogen loss from ungrazedtallgrass prairie (Dodds et al 1996Blair 1997) nitrogen loss from burn-ing averages 1ndash4 gmiddotmndash2middotyrndash1 (Blair etal 1998) Grazing lowers combus-tion losses of nitrogen in tallgrassprairie by reducing the abovegroundplant detritus and increasing thepatchiness of a prairie fire (Figure 2Hobbs et al 1991) Although vola-tilization of ammonia can be in-creased by grazing in other types ofgrassland (Detling 1988) Hobbs etal (1991) suggested that any increasein ammonia volatilization in tallgrassprairie will be more than compen-sated for by a reduction in combus-tion losses of nitrogen

Finally bison grazing affects theamount and quality of plant litter re-turned to soils Grazing increases plantuptake of nutrients (Ruess 1984) andshoot nitrogen content in many grass-lands (Holland and Detling 1990Milchunas et al 1995) includingtallgrass prairie (Turner et al 1993)However the effects of grazers onroot growth and chemistry varyamong grasslands (Milchunas andLauenroth 1993) On Konza Prairieroot productivity and root biomasswere 30 and 20 lower respec-tively in bison grazing lawns than inungrazed exclosures In addition thenitrogen concentration of new rootgrowth in bison grazing lawns atKonza Prairie increased significantlyfrom 06 to 09 and the CN ratioof roots decreased A lower CN ratioreduces microbial immobilization andenhances nitrogen availability withingrazed areas Indeed recent studieson Konza Prairie indicated that netnitrogen mineralization in bison graz-ing lawns was 153 greater and netnitrification 126 greater than inungrazed prairie (Figure 6) Further-more net nitrogen mineralization rateswere proportional to the intensity ofbison use of a given area Thus the

January 1999 45

net effect of bison grazing appears tobe increased rates of nitrogen cyclingcoupled with a significant increase inspatial heterogeneity in nitrogen avail-ability together these effects can alterpatterns of plant productivity and spe-cies composition in tallgrass prairie(Figure 5 Steinauer and Collins 1995)

Wallowing One aspect of bison be-havior that differs from that of cattleand is primarily a physical activityis wallowing Wallows in Flint Hillstallgrass prairie which are estab-lished primarily in level upland orlowland sites dramatically alter thepatch structure of this prairie Bisonwallows develop as the animals pawthe ground and roll in the exposedsoil Continued use of wallows bybulls cows and calves creates a soildepression of 3ndash5 m in diameter (and10ndash30 cm in depth) that is devoid ofvegetation These denuded patcheseither gradually revegetate or remainas bare soil depending on the fre-quency of revisitation by bison Withthe vast numbers of bison that onceoccupied the Great Plains these soildepressions were probably abundantand widespread features of the land-scape (England and DeVos 1969)For example a number of relic wal-lows had to be filled to level the play-ing field for the first University ofOklahoma home football game in 1895(University of Oklahoma Athletic De-partment 1986) Relic wallows stillexist in many areas where bison havenot occurred in the past 125 years

Environmental conditions in relicand newly established wallowsstrongly influence prairie patch dy-namics (Polley and Collins 1984Polley and Wallace 1986) Becauseof soil compaction wallows oftenretain rainwater in the spring creat-ing localized habitats that are suit-able for ephemeral wetland speciessimilar to vernal pools in California(Holland and Jain 1981 Uno 1989)In the summer however the samewallows support only plants that cantolerate severe drought Vegetationcomposition and structure in wal-lows is different from that in thesurrounding prairie (Polley andCollins 1984) and these differencesare enhanced by fire (Collins andUno 1983) which may not spreadthrough wallows because of low fuelloads Consequently at larger spa-

tial scales grazed prairie that con-tains bison wallows has higher plantspecies diversity than grazed prairiewithout wallows (Collins and Bar-ber 1985) Thus bison can physi-cally alter grasslands in ways thatincrease environmental heterogeneityand enhance both local and regionalbiodiversity (Hartnett et al 1997)

Bison carcasses Bison not only af-fect vegetation patterns and soil pro-cesses through their grazing activi-ties but also have profound andlasting localized effects after theydie Although legal requirements andmanagement practices dictate theremoval of carcasses of domesticherbivores from public and privategrasslands native herbivores rou-tinely die of natural causes and theirbodies remain in situ As part of theminimal management strategy atKonza Prairie bison that die on siteare not removed As a result thesecarcasses create unique local distur-bances (Figure 7) that are the focusof studies to assess their effects onsoil nutrients and vegetation re-sponses in tallgrass prairie

When an individual bison diescopious quantities of fluids (withhigh nitrogen concentration) are re-leased during decomposition Adultbison can weigh more than 800 kgand these carcasses typically kill un-derlying and adjacent plants creat-ing a denuded zone of 4ndash6 m2 (Figure7) Although the fluids that are ini-tially released are toxic to vegeta-tion these sites eventually becomezones of high fertility For examplesoil cores extracted from the centerof carcass sites on Konza Prairie 3years after death had inorganic ni-trogen concentrations that were twoto three times higher than the sur-rounding prairie This nutrient enrich-ment may extend up to 25 m awayfrom the original carcass site andresults in patches dominated initiallyby early successional species Theaboveground primary production inthese patches is two to three timeshigher than in undisturbed prairie

Although disturbances created bybison carcasses are sporadic and lo-calized on Konza Prairie they pro-vide nutrient pulses that exceed allother natural processes even urineand fecal deposits Overall we canonly speculate about historical rates

of bison mortality Given the enor-mous size of the bison populationbefore their widespread slaughter inthe 1800s annual mortality wasprobably high High death rateswould have been especially commonduring droughts when there wouldbe the potential for large numbers ofcarcasses to occur across the land-scape Even though predators andscavengers may have consumed andrelocated many of these carcassesdecomposition of the remaining andpartial carcasses would still have re-sulted in patches of locally high nu-trient concentration Thus althoughit was variable bison mortality wouldhave led to a continual cycle of dis-turbance and recovery of thesepatches in presettlement grasslands

Are bison keystone speciesThe net effects of selective bison graz-ing activities at the landscape patchand individual plant level includeshifts in plant species compositionalterations of the physical and chemi-cal environment and increased spa-tial and temporal heterogeneity invegetation structure soil resourceavailability and a variety of ecosys-tem processes (Figures 4 5 and 6)Before bison reintroduction at KonzaPrairie the long-term burning ex-periments produced clear patterns ofresponse in the vegetation As firefrequency increased the dominanceof C4 grasses increased and the coverof C3 grasses forbs and woody spe-cies decreased (Figure 4 Gibson andHulbert 1987) Overall plant spe-cies diversity declined as fire fre-quency increased in ungrazed tall-grass prairie (Collins et al 1995)

These patterns in community struc-ture which had developed over 20years of burning treatments at KonzaPrairie are being rapidly and dra-matically altered by the grazing ac-tivity of the reintroduced bison Inparticular grazing by bison has low-ered the abundance of the dominant C4

grasses increased the abundance ofthe subdominant C3 grasses and forbsand markedly increased plant speciesdiversity (by 23) richness (by 38)and community heterogeneity (by13) relative to ungrazed sites evenunder annual burning conditions(Hartnett et al 1996 Collins andSteinauer 1998 Collins et al 1998)

46 BioScience Vol 49 No 1

Because of the multiple and dra-matic effects of bison on this land-scape we believe that bison are key-stone species in the tallgrass prairieOther authors have noted the poten-tial of large grazing mammals to actas ldquokeystone herbivoresrdquo capable ofmaintaining open grassland vegeta-tion that would otherwise undergosuccession to shrubland or wood-land (Owen-Smith 1987) Indeed thedisappearance of a grazing mega-fauna at the end of the Pleistocenemay have played a major role in thewidespread transition from steppe totundra at that time (Zimov et al 1995)However the concept of keystonespecies has been controversial since itsinception (Power et al 1996) One ofthe problems with this concept hasbeen the variable interpretation of cri-teria by which species are determinedto be keystone Power et al (1996)consider a keystone species to be ldquoonewhose impact on its community orecosystem is large and disproportion-ately large relative to its abundancerdquoTo make this definition operationalthese authors proposed a measure ofcommunity importance (CI) to be usedas an index of the strength of theimpact of a given species

CI = [(tN ndash tD)tN][1pi]

where tN is a quantitative measure ofa trait (eg diversity) in an intactcommunity tD is the measure of thetrait when species i has been deletedand pi is the proportional abundance(biomass) of species i before it wasdeleted CI values ldquomuch greater than1rdquo indicate that a species is key-stone We estimate bison biomass atKonza Prairie to be approximately11ndash12 gmiddotm2 (Collins and Steinauer1998) which is approximately 1of the total vegetative biomass OnKonza Prairie diversity is signifi-cantly higher (10ndash33) on grazedsites than ungrazed sites (Hartnett etal 1996) and these values yield arange of CIs from 6 to 25 For thisreason and others we consider bisonto be keystone species in tallgrassprairie ecosystems

Are bison and cattle functionalequivalents in tallgrass prairieThe historical presence of immenseherds of large ungulates in Great

Plains grasslands is undisputed(McHugh 1972) and we have em-phasized the keystone role that bisonplayed in determining the structureand function of tallgrass prairies atmultiple spatial and temporal scalesWith the replacement of native bisonby domesticated cattle in the remain-ing grasslands an obvious issue isthe degree of similarity between thesetwo ungulates with respect to theireffects on tallgrass prairie In otherwords can bison and cattle be con-sidered ecological equivalents

There have been several previousattempts to answer this questionbut the results have been equivocalat best (Plumb and Dodd 1993Hartnett et al 1997) The primarybarrier to resolving this issue restswith a lack of comparative studies inwhich management is held constantand the type of grazer is varied Suchstudies have recently been initiatedat Konza Prairie Results after 3 yearsindicate that the abundance and rich-ness of annual forbs and the spatialheterogeneity of biomass and coverare higher in sites with bison than insites with cattle No dramatic differ-ences have been detected howeverbetween cattle- and bison-grazed sitesin cover of the dominant C4 grass Agerardii or the dominant forb Am-brosia psilostachya total plant spe-cies richness is also not dramaticallydifferent (E Gene Towne and DavidC Hartnett unpublished data)

Results at Konza Prairie are con-sistent with previous assessments(eg Schwartz and Ellis [1981] VanVuren and Bray [1983] and Plumband Dodd [1993] in mixed and short-grass prairie) which noted that bothbison and cattle are generalist herbi-vores that graze preferentially ongraminoids Nevertheless some dif-ferences in the foraging patterns ofbison and cattle have been docu-mented that may have long-term im-plications for grasslands For examplebison have a higher proportion ofgraminoids in their diet than docattle consequently forb and browsespecies are more common in cattlediets (Van Vuren and Bray 1983Hartnett et al 1997) Also bisonspend less time grazing than cattleand more time in nonfeeding activi-ties (Plumb and Dodd 1993) andbison strongly prefer open grasslandareas for grazing whereas cattle use

wooded and grassland habitats op-portunistically (Hartnett et al 1997)

Studies that have focused exclu-sively on cattle generally concur thattheir grazing activities increase spa-tial heterogeneity and enhance plantspecies diversity so long as stockingdensity is not too high (Collins 1987Hartnett et al 1996) Because bisongrazing in tallgrass prairie has a simi-lar effect one could conclude thateither herbivore can alter resourceavailability and heterogeneity andreduce the cover of the dominantgrasses sufficiently to enhance thesuccess of the subdominant speciesPerhaps of greater importance thandifferences in foraging patterns be-tween bison and cattle however arethe number of nongrazing activitiessuch as wallowing and horning (ierubbing on trees) that are associatedexclusively with bison (Coppedgeand Shaw 1997 Hartnett et al 1997)These activities when combined withthe spatial redistribution of nutri-ents and selective consumption ofthe dominant grasses may furtherincrease plant species richness andresource heterogeneity particularlyat the landscape scale

Nevertheless it is likely that be-cause bison and cattle are function-ally similar as large grass-feedingherbivores management strategies(stocking intensity and duration) willhave a greater influence on the de-gree of ecological equivalencyachieved than inherent differences inthese ungulates (Hartnett et al1997) Clearly the degree of over-lap in diet and foraging patterns isgreater between bison and cattle thanbetween cattle and other historicallyimportant native herbivores (Hart-nett et al 1997) such as antelope(Antilocapra americana) deer (Odo-coileus virginianus) and elk (Cervuscanadensis) Indeed the loss of ante-lope and elk from the tallgrass prai-rie coupled with dramatic increasesin deer populations presents addi-tional challenges for managing theseecosystems

Conservation implicationsConserving small and moderate-sizedtracts of once-vast biomes such asthe tallgrass prairie presents a uniqueset of problems that are distinct fromthose associated with spatially re-

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

42 BioScience Vol 49 No 1

permanent fenced exclosures haveexcluded bison from experimentalplots within grazed watersheds forseveral years When adjacent grazedsites were also protected from graz-ing with temporary exclosuresaboveground production in the firstyear in these newly protected siteswas reduced relative to that in theadjoining long-term ungrazed areas(Figure 2c) however these sites re-covered their production potentialby the second year Thus produc-tion potential can recover if bisongrazing is sufficiently dynamic ei-ther spatially or temporally suchthat sites are grazed intermittently

Removal of grass leaf area by bi-son and reductions in the capabilityof the dominant grasses to compen-sate for tissue lost after multipleyears of grazing suggest that the co-occurring subdominant forbs maybenefit from bison grazing Indeedcomparisons of forbs inside grazingpatches with those in adjacentungrazed prairie have shown thatgas exchange and aboveground bio-mass production density and cover

can be enhanced by the selective con-sumption of grasses by bison (Fahn-estock and Knapp 1993 Hartnett etal 1996 Damhoureyeh and Hartnett1997) Given the importance of pastgrazing pressures to the direct re-sponses of grasses to herbivory andthe indirect responses of forbs iden-tifying those factors that influencethe selection and reselection of graz-ing patches by bison in tallgrass prai-rie is critical for understanding thelong-term consequences of bisongrazing patterns

Factors influencing bisonselection of grazing sitesHistorical information regarding bi-son grazing patterns in the GreatPlains is replete with anecdotal ac-counts of herds attracted to recentlyburned grasslands (Figure 1McHugh 1972 Pyne 1982) butquantitative evidence of this prefer-ence in tallgrass prairie has beenlacking until recently (Coppedge andShaw 1998) As noted earlier bisonat Konza Prairie have free access to

10 watersheds that are subject todifferent fire frequencies Since 1991twice-weekly observations of the dis-tribution of bison within this areahave been made to assess patterns ofbison grazing (Nellis et al 1992)The results confirm that bison dograze preferentially within burnedwatersheds from April (burning takesplace in late Marchndashearly April)through June and July and in someyears through August (Figure 3Vinton et al 1993 Nellis and Briggs1997) In addition to grazing prefer-entially in burned sites bison in-crease their selective consumption ofsome grass species in burned sitesrelative to unburned sites (Pfeifferand Hartnett 1995) Late in the sum-mer lowland topographic positionswith deeper soils (and thereforegreater soil moisture and plant pro-ductivity Knapp et al 1993) inburned watersheds become preferredgrazing locations as the uplands dryThis preference for burned areas intallgrass prairie is consistent withpostfire responses in mixed grass prai-rie (Coppock and Detling 1986) aswell as with large ungulatesrsquo winterpreference for burned sites in the north-ern mixed grasslands of YellowstoneNational Park (Pearson et al 1995)

Within a watershed or at a spe-cific topographic position in tallgrassprairie several factors may influ-ence initial patch selection andreselection (see also Wallace et al1995) In addition patch selectionhas several long-term consequencesWhen bison were first reintroducedto Konza Prairie they encountered amosaic of burned and unburned wa-tersheds with significant differencesamong watersheds in the spatial het-erogeneity of plant community com-position For example frequentlyburned but ungrazed watersheds aredominated by C4 grasses and havelow species richness and diversitywhereas less frequently burned siteshave higher species richness and forbcover (Gibson and Hulbert 1987Collins 1992) Initial studies onKonza Prairie indicated clearly thatbison established grazing patches inareas strongly dominated by C4grasses and that these patches werereselected at a high rate (Vinton etal 1993) Six years later a survey offloristic composition indicated thatestablished bison grazing patches had

Figure 3 Spatial distribution of bison in 10 watersheds in the central portion of KonzaPrairie after the completion of spring burning During the months of April May andJune 1992 bison locations were determined twice per week and recorded on large-scalespatially rectified aerial photographs overlaid with a 30-meter grid Bison wereobserved most frequently in the recently burned watersheds B burned watersheds Uunburned watersheds

January 1999 43

a higher abundance of forbs and alower cover of grasses than adjacentungrazed patches (Catchpole 1996)Similar small-scale patterns of forband grass abundance were observedby comparing the floristic composi-tion inside and outside grazingexclosures in grazed watersheds atKonza Prairie (Hartnett et al 1996)These observations suggest that bi-son alter plant community composi-tion at the patch scale by selectingspecies-poor grass-dominated sitesand converting them to sites of lo-cally higher diversity (Figure 4)

A second factor that may influ-ence patch selection and reselectionby bison is plant quality The foliar

nitrogen content of plants is highlyvariable both spatially and tempo-rally Bison contribute to this patchi-ness through deposition of nitrogen-rich urine Steinauer (1994) appliedsynthetic bovine urine at randomlyselected locations along eighttransects at Konza Prairie (four eachin grazed and ungrazed areas) Inungrazed areas grass cover was sig-nificantly higher in plots that werefertilized with urine than in plotswithout urine But in the grazed areaby contrast grass cover was signifi-cantly lower on the urine-treatedplots than in plots without urinebecause bison preferentially grazedthe grasses on the urine-treated plots

Day and Detling (1990) have shownthat grasses growing on urine patchesin mixed-grass prairie have higherleaf nitrogen content and are there-fore more nutritious per bite thangrasses growing on patches withouturine Not only was grass cover loweron urine patches at Konza Prairiebut the total area of grazed patcheson the urine-treated transects wassignificantly larger than the area ofgrazed patches on transects that werenot treated with urine (Steinauer1994) Thus the enhanced produc-tivity of grasses growing on urinepatches represents a potential stimu-lus for the initiation and reselectionof grazing patches by bison

Figure 4 Conceptual model of the spatial and temporal dynamics of bison grazing activities and the responses of tallgrass prairieto the reintroduction of bison Before the reintroduction of bison watershed attributes at Konza Prairie differed strongly dependingon the fire regime imposed with frequently burned sites notable for their high productivity but low plant species diversity Selectivebison grazing at the watershed level (preference for burned sites particularly in the spring) and the patch level (selection of patchesdominated by C4 grasses) has led to increased similarity in watershed attributes despite differences in fire history Moreoversignificant increases in plant species diversity and spatial heterogeneity have been noted in all watersheds Superscripts denote studieson which this synthesis is based 1 Nellis and Briggs (1997) 2 Vinton et al (1993) 3 Briggs and Knapp (1995) 4 Collins (1992)Collins et al (1995) 5 Vinton and Hartnett (1992) Vinton et al (1993) 6 Hartnett et al (1997) 7 Fahnestock and Knapp (19931994) 8 Catchpole (1996) 9 Collins and Steinauer (1998) 10 Hartnett et al (1996)

44 BioScience Vol 49 No 1

Given that bison prefer grazingpatches that are initially dominatedby C4 grasses but that their selectiveforaging and reselection habits con-vert these patches to sites with agreater abundance of nonforage spe-cies it is likely that patch locationsare spatially dynamic across the land-scape Patches can ldquomoverdquo by twomechanisms patch abandonmentfollowed by selection of a new patchor patch migration wherein portionsof a patch are abandoned and thepatch expands into adjacent areasOver a 3-year period Catchpole(1996) found the rate of patch aban-donment to be approximately 6ndash7per year in both burned and un-burned watersheds thus at leastportions of previously establishedgrazing patches were reselected at ahigh rate (Figure 4) However whengrazing patches were mapped andcompared across years the extent ofspatial reselectionmdashalthough highlyvariable due to differences in totalburned area available to bison in anyone yearmdashaveraged approximately50 per year Thus grazing patchesin both burned and unburned water-sheds appear to migrate significantlyfrom year to year This local migra-tion permits periodic release of por-tions of the grassland from grazingpressures (Figure 4) and provides amechanism for recovery of below-ground carbohydrate storage reservesand production potential

At the watershed and landscapescales the long-term consequencesof bison activities include a reductionin cover dominance and productivityof grasses the competitive release ofmany subdominant species resultingin an increase in the abundance offorbs an overall increase in plantspecies richness and diversity andincreased spatial heterogeneity (Fig-ure 5 Hartnett et al 1996) Althoughalterations in plant community com-position can be attributed in largepart to the direct effects of grazingby bison increased plant species rich-ness is also likely to be a product ofincreased microsite diversity gener-ated by nongrazing activities suchas dung and urine deposition tram-pling and wallowing These andother bison activities contribute sig-nificantly to the increase in spatialheterogeneity that is characteristicof grazed tallgrass prairie (Figure 5)

Other impacts of bisonin tallgrass prairie

Effects of ungulates especially graz-ing on plant community composi-tion and structure have been studiedin many grasslands worldwide(McNaughton 1984 Milchunas etal 1988 Frank and McNaughton1992 Milchunas and Lauenroth1993) Ungulate activities howeveraffect many other aspects of grass-land structure and function includingthe physical structure of the environ-ment and the rates of a number ofecosystem-level processes (McNaugh-ton 1993 Frank and Evans 1997McNaughton et al 1997) There areseveral other important mechanismsby which bison alter ecosystem-levelprocesses and physical habitat struc-ture in tallgrass prairie

Nutrient redistribution and cyclingBison can substantially alter nutri-ent cycling processes and patterns ofnutrient availability in tallgrass prai-rie Their effects on nitrogen cyclingare critical because nitrogen avail-ability often limits plant productiv-ity in these grasslands (Seastedt et al1991 Blair 1997 Turner et al 1997)and influences plant species compo-sition (Gibson et al 1993 Wedinand Tilman 1993) Simulation mod-els of tallgrass prairie responses tograzing (Risser and Parton 1982)and studies of grazers in other grass-lands (Frank and Evans 1997McNaughton et al 1997) have dem-onstrated a disproportionate influ-ence of ungulates including bisonon the regulation of nitrogen cyclingprocesses Preliminary data fromKonza Prairie suggest that bison aresimilarly important in controllingnitrogen cycling in tallgrass prairie

Bison influence nitrogen cyclingconservation and availability intallgrass prairie ecosystems by alter-ing several soil and plant processesUngulates in grasslands consumerelatively recalcitrant plant biomassand return labile forms of nitrogen(ie urine) to soils (Ruess andMcNaughton 1988) thus bypassingthe otherwise slow mineralization ofnitrogen in plant litter Nitrogen inbison urine is largely urea which canbe hydrolyzed to ammonium in amatter of days (Ruess and McNaugh-ton 1988) Indeed application of

synthetic bison urine increased con-centrations of ammonium and ni-trate in Konza Prairie soils over 130-fold and 30-fold respectively 8 daysafter application (J R Matchett andLoretta C Johnson unpublished data)

Bison grazing can decrease theexport of nitrogen from tallgrassprairie by altering the magnitudes oftwo major pathways of nitrogenlossmdashcombustion and ammoniavolatilization Fire is the major path-way of nitrogen loss from ungrazedtallgrass prairie (Dodds et al 1996Blair 1997) nitrogen loss from burn-ing averages 1ndash4 gmiddotmndash2middotyrndash1 (Blair etal 1998) Grazing lowers combus-tion losses of nitrogen in tallgrassprairie by reducing the abovegroundplant detritus and increasing thepatchiness of a prairie fire (Figure 2Hobbs et al 1991) Although vola-tilization of ammonia can be in-creased by grazing in other types ofgrassland (Detling 1988) Hobbs etal (1991) suggested that any increasein ammonia volatilization in tallgrassprairie will be more than compen-sated for by a reduction in combus-tion losses of nitrogen

Finally bison grazing affects theamount and quality of plant litter re-turned to soils Grazing increases plantuptake of nutrients (Ruess 1984) andshoot nitrogen content in many grass-lands (Holland and Detling 1990Milchunas et al 1995) includingtallgrass prairie (Turner et al 1993)However the effects of grazers onroot growth and chemistry varyamong grasslands (Milchunas andLauenroth 1993) On Konza Prairieroot productivity and root biomasswere 30 and 20 lower respec-tively in bison grazing lawns than inungrazed exclosures In addition thenitrogen concentration of new rootgrowth in bison grazing lawns atKonza Prairie increased significantlyfrom 06 to 09 and the CN ratioof roots decreased A lower CN ratioreduces microbial immobilization andenhances nitrogen availability withingrazed areas Indeed recent studieson Konza Prairie indicated that netnitrogen mineralization in bison graz-ing lawns was 153 greater and netnitrification 126 greater than inungrazed prairie (Figure 6) Further-more net nitrogen mineralization rateswere proportional to the intensity ofbison use of a given area Thus the

January 1999 45

net effect of bison grazing appears tobe increased rates of nitrogen cyclingcoupled with a significant increase inspatial heterogeneity in nitrogen avail-ability together these effects can alterpatterns of plant productivity and spe-cies composition in tallgrass prairie(Figure 5 Steinauer and Collins 1995)

Wallowing One aspect of bison be-havior that differs from that of cattleand is primarily a physical activityis wallowing Wallows in Flint Hillstallgrass prairie which are estab-lished primarily in level upland orlowland sites dramatically alter thepatch structure of this prairie Bisonwallows develop as the animals pawthe ground and roll in the exposedsoil Continued use of wallows bybulls cows and calves creates a soildepression of 3ndash5 m in diameter (and10ndash30 cm in depth) that is devoid ofvegetation These denuded patcheseither gradually revegetate or remainas bare soil depending on the fre-quency of revisitation by bison Withthe vast numbers of bison that onceoccupied the Great Plains these soildepressions were probably abundantand widespread features of the land-scape (England and DeVos 1969)For example a number of relic wal-lows had to be filled to level the play-ing field for the first University ofOklahoma home football game in 1895(University of Oklahoma Athletic De-partment 1986) Relic wallows stillexist in many areas where bison havenot occurred in the past 125 years

Environmental conditions in relicand newly established wallowsstrongly influence prairie patch dy-namics (Polley and Collins 1984Polley and Wallace 1986) Becauseof soil compaction wallows oftenretain rainwater in the spring creat-ing localized habitats that are suit-able for ephemeral wetland speciessimilar to vernal pools in California(Holland and Jain 1981 Uno 1989)In the summer however the samewallows support only plants that cantolerate severe drought Vegetationcomposition and structure in wal-lows is different from that in thesurrounding prairie (Polley andCollins 1984) and these differencesare enhanced by fire (Collins andUno 1983) which may not spreadthrough wallows because of low fuelloads Consequently at larger spa-

tial scales grazed prairie that con-tains bison wallows has higher plantspecies diversity than grazed prairiewithout wallows (Collins and Bar-ber 1985) Thus bison can physi-cally alter grasslands in ways thatincrease environmental heterogeneityand enhance both local and regionalbiodiversity (Hartnett et al 1997)

Bison carcasses Bison not only af-fect vegetation patterns and soil pro-cesses through their grazing activi-ties but also have profound andlasting localized effects after theydie Although legal requirements andmanagement practices dictate theremoval of carcasses of domesticherbivores from public and privategrasslands native herbivores rou-tinely die of natural causes and theirbodies remain in situ As part of theminimal management strategy atKonza Prairie bison that die on siteare not removed As a result thesecarcasses create unique local distur-bances (Figure 7) that are the focusof studies to assess their effects onsoil nutrients and vegetation re-sponses in tallgrass prairie

When an individual bison diescopious quantities of fluids (withhigh nitrogen concentration) are re-leased during decomposition Adultbison can weigh more than 800 kgand these carcasses typically kill un-derlying and adjacent plants creat-ing a denuded zone of 4ndash6 m2 (Figure7) Although the fluids that are ini-tially released are toxic to vegeta-tion these sites eventually becomezones of high fertility For examplesoil cores extracted from the centerof carcass sites on Konza Prairie 3years after death had inorganic ni-trogen concentrations that were twoto three times higher than the sur-rounding prairie This nutrient enrich-ment may extend up to 25 m awayfrom the original carcass site andresults in patches dominated initiallyby early successional species Theaboveground primary production inthese patches is two to three timeshigher than in undisturbed prairie

Although disturbances created bybison carcasses are sporadic and lo-calized on Konza Prairie they pro-vide nutrient pulses that exceed allother natural processes even urineand fecal deposits Overall we canonly speculate about historical rates

of bison mortality Given the enor-mous size of the bison populationbefore their widespread slaughter inthe 1800s annual mortality wasprobably high High death rateswould have been especially commonduring droughts when there wouldbe the potential for large numbers ofcarcasses to occur across the land-scape Even though predators andscavengers may have consumed andrelocated many of these carcassesdecomposition of the remaining andpartial carcasses would still have re-sulted in patches of locally high nu-trient concentration Thus althoughit was variable bison mortality wouldhave led to a continual cycle of dis-turbance and recovery of thesepatches in presettlement grasslands

Are bison keystone speciesThe net effects of selective bison graz-ing activities at the landscape patchand individual plant level includeshifts in plant species compositionalterations of the physical and chemi-cal environment and increased spa-tial and temporal heterogeneity invegetation structure soil resourceavailability and a variety of ecosys-tem processes (Figures 4 5 and 6)Before bison reintroduction at KonzaPrairie the long-term burning ex-periments produced clear patterns ofresponse in the vegetation As firefrequency increased the dominanceof C4 grasses increased and the coverof C3 grasses forbs and woody spe-cies decreased (Figure 4 Gibson andHulbert 1987) Overall plant spe-cies diversity declined as fire fre-quency increased in ungrazed tall-grass prairie (Collins et al 1995)

These patterns in community struc-ture which had developed over 20years of burning treatments at KonzaPrairie are being rapidly and dra-matically altered by the grazing ac-tivity of the reintroduced bison Inparticular grazing by bison has low-ered the abundance of the dominant C4

grasses increased the abundance ofthe subdominant C3 grasses and forbsand markedly increased plant speciesdiversity (by 23) richness (by 38)and community heterogeneity (by13) relative to ungrazed sites evenunder annual burning conditions(Hartnett et al 1996 Collins andSteinauer 1998 Collins et al 1998)

46 BioScience Vol 49 No 1

Because of the multiple and dra-matic effects of bison on this land-scape we believe that bison are key-stone species in the tallgrass prairieOther authors have noted the poten-tial of large grazing mammals to actas ldquokeystone herbivoresrdquo capable ofmaintaining open grassland vegeta-tion that would otherwise undergosuccession to shrubland or wood-land (Owen-Smith 1987) Indeed thedisappearance of a grazing mega-fauna at the end of the Pleistocenemay have played a major role in thewidespread transition from steppe totundra at that time (Zimov et al 1995)However the concept of keystonespecies has been controversial since itsinception (Power et al 1996) One ofthe problems with this concept hasbeen the variable interpretation of cri-teria by which species are determinedto be keystone Power et al (1996)consider a keystone species to be ldquoonewhose impact on its community orecosystem is large and disproportion-ately large relative to its abundancerdquoTo make this definition operationalthese authors proposed a measure ofcommunity importance (CI) to be usedas an index of the strength of theimpact of a given species

CI = [(tN ndash tD)tN][1pi]

where tN is a quantitative measure ofa trait (eg diversity) in an intactcommunity tD is the measure of thetrait when species i has been deletedand pi is the proportional abundance(biomass) of species i before it wasdeleted CI values ldquomuch greater than1rdquo indicate that a species is key-stone We estimate bison biomass atKonza Prairie to be approximately11ndash12 gmiddotm2 (Collins and Steinauer1998) which is approximately 1of the total vegetative biomass OnKonza Prairie diversity is signifi-cantly higher (10ndash33) on grazedsites than ungrazed sites (Hartnett etal 1996) and these values yield arange of CIs from 6 to 25 For thisreason and others we consider bisonto be keystone species in tallgrassprairie ecosystems

Are bison and cattle functionalequivalents in tallgrass prairieThe historical presence of immenseherds of large ungulates in Great

Plains grasslands is undisputed(McHugh 1972) and we have em-phasized the keystone role that bisonplayed in determining the structureand function of tallgrass prairies atmultiple spatial and temporal scalesWith the replacement of native bisonby domesticated cattle in the remain-ing grasslands an obvious issue isthe degree of similarity between thesetwo ungulates with respect to theireffects on tallgrass prairie In otherwords can bison and cattle be con-sidered ecological equivalents

There have been several previousattempts to answer this questionbut the results have been equivocalat best (Plumb and Dodd 1993Hartnett et al 1997) The primarybarrier to resolving this issue restswith a lack of comparative studies inwhich management is held constantand the type of grazer is varied Suchstudies have recently been initiatedat Konza Prairie Results after 3 yearsindicate that the abundance and rich-ness of annual forbs and the spatialheterogeneity of biomass and coverare higher in sites with bison than insites with cattle No dramatic differ-ences have been detected howeverbetween cattle- and bison-grazed sitesin cover of the dominant C4 grass Agerardii or the dominant forb Am-brosia psilostachya total plant spe-cies richness is also not dramaticallydifferent (E Gene Towne and DavidC Hartnett unpublished data)

Results at Konza Prairie are con-sistent with previous assessments(eg Schwartz and Ellis [1981] VanVuren and Bray [1983] and Plumband Dodd [1993] in mixed and short-grass prairie) which noted that bothbison and cattle are generalist herbi-vores that graze preferentially ongraminoids Nevertheless some dif-ferences in the foraging patterns ofbison and cattle have been docu-mented that may have long-term im-plications for grasslands For examplebison have a higher proportion ofgraminoids in their diet than docattle consequently forb and browsespecies are more common in cattlediets (Van Vuren and Bray 1983Hartnett et al 1997) Also bisonspend less time grazing than cattleand more time in nonfeeding activi-ties (Plumb and Dodd 1993) andbison strongly prefer open grasslandareas for grazing whereas cattle use

wooded and grassland habitats op-portunistically (Hartnett et al 1997)

Studies that have focused exclu-sively on cattle generally concur thattheir grazing activities increase spa-tial heterogeneity and enhance plantspecies diversity so long as stockingdensity is not too high (Collins 1987Hartnett et al 1996) Because bisongrazing in tallgrass prairie has a simi-lar effect one could conclude thateither herbivore can alter resourceavailability and heterogeneity andreduce the cover of the dominantgrasses sufficiently to enhance thesuccess of the subdominant speciesPerhaps of greater importance thandifferences in foraging patterns be-tween bison and cattle however arethe number of nongrazing activitiessuch as wallowing and horning (ierubbing on trees) that are associatedexclusively with bison (Coppedgeand Shaw 1997 Hartnett et al 1997)These activities when combined withthe spatial redistribution of nutri-ents and selective consumption ofthe dominant grasses may furtherincrease plant species richness andresource heterogeneity particularlyat the landscape scale

Nevertheless it is likely that be-cause bison and cattle are function-ally similar as large grass-feedingherbivores management strategies(stocking intensity and duration) willhave a greater influence on the de-gree of ecological equivalencyachieved than inherent differences inthese ungulates (Hartnett et al1997) Clearly the degree of over-lap in diet and foraging patterns isgreater between bison and cattle thanbetween cattle and other historicallyimportant native herbivores (Hart-nett et al 1997) such as antelope(Antilocapra americana) deer (Odo-coileus virginianus) and elk (Cervuscanadensis) Indeed the loss of ante-lope and elk from the tallgrass prai-rie coupled with dramatic increasesin deer populations presents addi-tional challenges for managing theseecosystems

Conservation implicationsConserving small and moderate-sizedtracts of once-vast biomes such asthe tallgrass prairie presents a uniqueset of problems that are distinct fromthose associated with spatially re-

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

January 1999 43

a higher abundance of forbs and alower cover of grasses than adjacentungrazed patches (Catchpole 1996)Similar small-scale patterns of forband grass abundance were observedby comparing the floristic composi-tion inside and outside grazingexclosures in grazed watersheds atKonza Prairie (Hartnett et al 1996)These observations suggest that bi-son alter plant community composi-tion at the patch scale by selectingspecies-poor grass-dominated sitesand converting them to sites of lo-cally higher diversity (Figure 4)

A second factor that may influ-ence patch selection and reselectionby bison is plant quality The foliar

nitrogen content of plants is highlyvariable both spatially and tempo-rally Bison contribute to this patchi-ness through deposition of nitrogen-rich urine Steinauer (1994) appliedsynthetic bovine urine at randomlyselected locations along eighttransects at Konza Prairie (four eachin grazed and ungrazed areas) Inungrazed areas grass cover was sig-nificantly higher in plots that werefertilized with urine than in plotswithout urine But in the grazed areaby contrast grass cover was signifi-cantly lower on the urine-treatedplots than in plots without urinebecause bison preferentially grazedthe grasses on the urine-treated plots

Day and Detling (1990) have shownthat grasses growing on urine patchesin mixed-grass prairie have higherleaf nitrogen content and are there-fore more nutritious per bite thangrasses growing on patches withouturine Not only was grass cover loweron urine patches at Konza Prairiebut the total area of grazed patcheson the urine-treated transects wassignificantly larger than the area ofgrazed patches on transects that werenot treated with urine (Steinauer1994) Thus the enhanced produc-tivity of grasses growing on urinepatches represents a potential stimu-lus for the initiation and reselectionof grazing patches by bison

Figure 4 Conceptual model of the spatial and temporal dynamics of bison grazing activities and the responses of tallgrass prairieto the reintroduction of bison Before the reintroduction of bison watershed attributes at Konza Prairie differed strongly dependingon the fire regime imposed with frequently burned sites notable for their high productivity but low plant species diversity Selectivebison grazing at the watershed level (preference for burned sites particularly in the spring) and the patch level (selection of patchesdominated by C4 grasses) has led to increased similarity in watershed attributes despite differences in fire history Moreoversignificant increases in plant species diversity and spatial heterogeneity have been noted in all watersheds Superscripts denote studieson which this synthesis is based 1 Nellis and Briggs (1997) 2 Vinton et al (1993) 3 Briggs and Knapp (1995) 4 Collins (1992)Collins et al (1995) 5 Vinton and Hartnett (1992) Vinton et al (1993) 6 Hartnett et al (1997) 7 Fahnestock and Knapp (19931994) 8 Catchpole (1996) 9 Collins and Steinauer (1998) 10 Hartnett et al (1996)

44 BioScience Vol 49 No 1

Given that bison prefer grazingpatches that are initially dominatedby C4 grasses but that their selectiveforaging and reselection habits con-vert these patches to sites with agreater abundance of nonforage spe-cies it is likely that patch locationsare spatially dynamic across the land-scape Patches can ldquomoverdquo by twomechanisms patch abandonmentfollowed by selection of a new patchor patch migration wherein portionsof a patch are abandoned and thepatch expands into adjacent areasOver a 3-year period Catchpole(1996) found the rate of patch aban-donment to be approximately 6ndash7per year in both burned and un-burned watersheds thus at leastportions of previously establishedgrazing patches were reselected at ahigh rate (Figure 4) However whengrazing patches were mapped andcompared across years the extent ofspatial reselectionmdashalthough highlyvariable due to differences in totalburned area available to bison in anyone yearmdashaveraged approximately50 per year Thus grazing patchesin both burned and unburned water-sheds appear to migrate significantlyfrom year to year This local migra-tion permits periodic release of por-tions of the grassland from grazingpressures (Figure 4) and provides amechanism for recovery of below-ground carbohydrate storage reservesand production potential

At the watershed and landscapescales the long-term consequencesof bison activities include a reductionin cover dominance and productivityof grasses the competitive release ofmany subdominant species resultingin an increase in the abundance offorbs an overall increase in plantspecies richness and diversity andincreased spatial heterogeneity (Fig-ure 5 Hartnett et al 1996) Althoughalterations in plant community com-position can be attributed in largepart to the direct effects of grazingby bison increased plant species rich-ness is also likely to be a product ofincreased microsite diversity gener-ated by nongrazing activities suchas dung and urine deposition tram-pling and wallowing These andother bison activities contribute sig-nificantly to the increase in spatialheterogeneity that is characteristicof grazed tallgrass prairie (Figure 5)

Other impacts of bisonin tallgrass prairie

Effects of ungulates especially graz-ing on plant community composi-tion and structure have been studiedin many grasslands worldwide(McNaughton 1984 Milchunas etal 1988 Frank and McNaughton1992 Milchunas and Lauenroth1993) Ungulate activities howeveraffect many other aspects of grass-land structure and function includingthe physical structure of the environ-ment and the rates of a number ofecosystem-level processes (McNaugh-ton 1993 Frank and Evans 1997McNaughton et al 1997) There areseveral other important mechanismsby which bison alter ecosystem-levelprocesses and physical habitat struc-ture in tallgrass prairie

Nutrient redistribution and cyclingBison can substantially alter nutri-ent cycling processes and patterns ofnutrient availability in tallgrass prai-rie Their effects on nitrogen cyclingare critical because nitrogen avail-ability often limits plant productiv-ity in these grasslands (Seastedt et al1991 Blair 1997 Turner et al 1997)and influences plant species compo-sition (Gibson et al 1993 Wedinand Tilman 1993) Simulation mod-els of tallgrass prairie responses tograzing (Risser and Parton 1982)and studies of grazers in other grass-lands (Frank and Evans 1997McNaughton et al 1997) have dem-onstrated a disproportionate influ-ence of ungulates including bisonon the regulation of nitrogen cyclingprocesses Preliminary data fromKonza Prairie suggest that bison aresimilarly important in controllingnitrogen cycling in tallgrass prairie

Bison influence nitrogen cyclingconservation and availability intallgrass prairie ecosystems by alter-ing several soil and plant processesUngulates in grasslands consumerelatively recalcitrant plant biomassand return labile forms of nitrogen(ie urine) to soils (Ruess andMcNaughton 1988) thus bypassingthe otherwise slow mineralization ofnitrogen in plant litter Nitrogen inbison urine is largely urea which canbe hydrolyzed to ammonium in amatter of days (Ruess and McNaugh-ton 1988) Indeed application of

synthetic bison urine increased con-centrations of ammonium and ni-trate in Konza Prairie soils over 130-fold and 30-fold respectively 8 daysafter application (J R Matchett andLoretta C Johnson unpublished data)

Bison grazing can decrease theexport of nitrogen from tallgrassprairie by altering the magnitudes oftwo major pathways of nitrogenlossmdashcombustion and ammoniavolatilization Fire is the major path-way of nitrogen loss from ungrazedtallgrass prairie (Dodds et al 1996Blair 1997) nitrogen loss from burn-ing averages 1ndash4 gmiddotmndash2middotyrndash1 (Blair etal 1998) Grazing lowers combus-tion losses of nitrogen in tallgrassprairie by reducing the abovegroundplant detritus and increasing thepatchiness of a prairie fire (Figure 2Hobbs et al 1991) Although vola-tilization of ammonia can be in-creased by grazing in other types ofgrassland (Detling 1988) Hobbs etal (1991) suggested that any increasein ammonia volatilization in tallgrassprairie will be more than compen-sated for by a reduction in combus-tion losses of nitrogen

Finally bison grazing affects theamount and quality of plant litter re-turned to soils Grazing increases plantuptake of nutrients (Ruess 1984) andshoot nitrogen content in many grass-lands (Holland and Detling 1990Milchunas et al 1995) includingtallgrass prairie (Turner et al 1993)However the effects of grazers onroot growth and chemistry varyamong grasslands (Milchunas andLauenroth 1993) On Konza Prairieroot productivity and root biomasswere 30 and 20 lower respec-tively in bison grazing lawns than inungrazed exclosures In addition thenitrogen concentration of new rootgrowth in bison grazing lawns atKonza Prairie increased significantlyfrom 06 to 09 and the CN ratioof roots decreased A lower CN ratioreduces microbial immobilization andenhances nitrogen availability withingrazed areas Indeed recent studieson Konza Prairie indicated that netnitrogen mineralization in bison graz-ing lawns was 153 greater and netnitrification 126 greater than inungrazed prairie (Figure 6) Further-more net nitrogen mineralization rateswere proportional to the intensity ofbison use of a given area Thus the

January 1999 45

net effect of bison grazing appears tobe increased rates of nitrogen cyclingcoupled with a significant increase inspatial heterogeneity in nitrogen avail-ability together these effects can alterpatterns of plant productivity and spe-cies composition in tallgrass prairie(Figure 5 Steinauer and Collins 1995)

Wallowing One aspect of bison be-havior that differs from that of cattleand is primarily a physical activityis wallowing Wallows in Flint Hillstallgrass prairie which are estab-lished primarily in level upland orlowland sites dramatically alter thepatch structure of this prairie Bisonwallows develop as the animals pawthe ground and roll in the exposedsoil Continued use of wallows bybulls cows and calves creates a soildepression of 3ndash5 m in diameter (and10ndash30 cm in depth) that is devoid ofvegetation These denuded patcheseither gradually revegetate or remainas bare soil depending on the fre-quency of revisitation by bison Withthe vast numbers of bison that onceoccupied the Great Plains these soildepressions were probably abundantand widespread features of the land-scape (England and DeVos 1969)For example a number of relic wal-lows had to be filled to level the play-ing field for the first University ofOklahoma home football game in 1895(University of Oklahoma Athletic De-partment 1986) Relic wallows stillexist in many areas where bison havenot occurred in the past 125 years

Environmental conditions in relicand newly established wallowsstrongly influence prairie patch dy-namics (Polley and Collins 1984Polley and Wallace 1986) Becauseof soil compaction wallows oftenretain rainwater in the spring creat-ing localized habitats that are suit-able for ephemeral wetland speciessimilar to vernal pools in California(Holland and Jain 1981 Uno 1989)In the summer however the samewallows support only plants that cantolerate severe drought Vegetationcomposition and structure in wal-lows is different from that in thesurrounding prairie (Polley andCollins 1984) and these differencesare enhanced by fire (Collins andUno 1983) which may not spreadthrough wallows because of low fuelloads Consequently at larger spa-

tial scales grazed prairie that con-tains bison wallows has higher plantspecies diversity than grazed prairiewithout wallows (Collins and Bar-ber 1985) Thus bison can physi-cally alter grasslands in ways thatincrease environmental heterogeneityand enhance both local and regionalbiodiversity (Hartnett et al 1997)

Bison carcasses Bison not only af-fect vegetation patterns and soil pro-cesses through their grazing activi-ties but also have profound andlasting localized effects after theydie Although legal requirements andmanagement practices dictate theremoval of carcasses of domesticherbivores from public and privategrasslands native herbivores rou-tinely die of natural causes and theirbodies remain in situ As part of theminimal management strategy atKonza Prairie bison that die on siteare not removed As a result thesecarcasses create unique local distur-bances (Figure 7) that are the focusof studies to assess their effects onsoil nutrients and vegetation re-sponses in tallgrass prairie

When an individual bison diescopious quantities of fluids (withhigh nitrogen concentration) are re-leased during decomposition Adultbison can weigh more than 800 kgand these carcasses typically kill un-derlying and adjacent plants creat-ing a denuded zone of 4ndash6 m2 (Figure7) Although the fluids that are ini-tially released are toxic to vegeta-tion these sites eventually becomezones of high fertility For examplesoil cores extracted from the centerof carcass sites on Konza Prairie 3years after death had inorganic ni-trogen concentrations that were twoto three times higher than the sur-rounding prairie This nutrient enrich-ment may extend up to 25 m awayfrom the original carcass site andresults in patches dominated initiallyby early successional species Theaboveground primary production inthese patches is two to three timeshigher than in undisturbed prairie

Although disturbances created bybison carcasses are sporadic and lo-calized on Konza Prairie they pro-vide nutrient pulses that exceed allother natural processes even urineand fecal deposits Overall we canonly speculate about historical rates

of bison mortality Given the enor-mous size of the bison populationbefore their widespread slaughter inthe 1800s annual mortality wasprobably high High death rateswould have been especially commonduring droughts when there wouldbe the potential for large numbers ofcarcasses to occur across the land-scape Even though predators andscavengers may have consumed andrelocated many of these carcassesdecomposition of the remaining andpartial carcasses would still have re-sulted in patches of locally high nu-trient concentration Thus althoughit was variable bison mortality wouldhave led to a continual cycle of dis-turbance and recovery of thesepatches in presettlement grasslands

Are bison keystone speciesThe net effects of selective bison graz-ing activities at the landscape patchand individual plant level includeshifts in plant species compositionalterations of the physical and chemi-cal environment and increased spa-tial and temporal heterogeneity invegetation structure soil resourceavailability and a variety of ecosys-tem processes (Figures 4 5 and 6)Before bison reintroduction at KonzaPrairie the long-term burning ex-periments produced clear patterns ofresponse in the vegetation As firefrequency increased the dominanceof C4 grasses increased and the coverof C3 grasses forbs and woody spe-cies decreased (Figure 4 Gibson andHulbert 1987) Overall plant spe-cies diversity declined as fire fre-quency increased in ungrazed tall-grass prairie (Collins et al 1995)

These patterns in community struc-ture which had developed over 20years of burning treatments at KonzaPrairie are being rapidly and dra-matically altered by the grazing ac-tivity of the reintroduced bison Inparticular grazing by bison has low-ered the abundance of the dominant C4

grasses increased the abundance ofthe subdominant C3 grasses and forbsand markedly increased plant speciesdiversity (by 23) richness (by 38)and community heterogeneity (by13) relative to ungrazed sites evenunder annual burning conditions(Hartnett et al 1996 Collins andSteinauer 1998 Collins et al 1998)

46 BioScience Vol 49 No 1

Because of the multiple and dra-matic effects of bison on this land-scape we believe that bison are key-stone species in the tallgrass prairieOther authors have noted the poten-tial of large grazing mammals to actas ldquokeystone herbivoresrdquo capable ofmaintaining open grassland vegeta-tion that would otherwise undergosuccession to shrubland or wood-land (Owen-Smith 1987) Indeed thedisappearance of a grazing mega-fauna at the end of the Pleistocenemay have played a major role in thewidespread transition from steppe totundra at that time (Zimov et al 1995)However the concept of keystonespecies has been controversial since itsinception (Power et al 1996) One ofthe problems with this concept hasbeen the variable interpretation of cri-teria by which species are determinedto be keystone Power et al (1996)consider a keystone species to be ldquoonewhose impact on its community orecosystem is large and disproportion-ately large relative to its abundancerdquoTo make this definition operationalthese authors proposed a measure ofcommunity importance (CI) to be usedas an index of the strength of theimpact of a given species

CI = [(tN ndash tD)tN][1pi]

where tN is a quantitative measure ofa trait (eg diversity) in an intactcommunity tD is the measure of thetrait when species i has been deletedand pi is the proportional abundance(biomass) of species i before it wasdeleted CI values ldquomuch greater than1rdquo indicate that a species is key-stone We estimate bison biomass atKonza Prairie to be approximately11ndash12 gmiddotm2 (Collins and Steinauer1998) which is approximately 1of the total vegetative biomass OnKonza Prairie diversity is signifi-cantly higher (10ndash33) on grazedsites than ungrazed sites (Hartnett etal 1996) and these values yield arange of CIs from 6 to 25 For thisreason and others we consider bisonto be keystone species in tallgrassprairie ecosystems

Are bison and cattle functionalequivalents in tallgrass prairieThe historical presence of immenseherds of large ungulates in Great

Plains grasslands is undisputed(McHugh 1972) and we have em-phasized the keystone role that bisonplayed in determining the structureand function of tallgrass prairies atmultiple spatial and temporal scalesWith the replacement of native bisonby domesticated cattle in the remain-ing grasslands an obvious issue isthe degree of similarity between thesetwo ungulates with respect to theireffects on tallgrass prairie In otherwords can bison and cattle be con-sidered ecological equivalents

There have been several previousattempts to answer this questionbut the results have been equivocalat best (Plumb and Dodd 1993Hartnett et al 1997) The primarybarrier to resolving this issue restswith a lack of comparative studies inwhich management is held constantand the type of grazer is varied Suchstudies have recently been initiatedat Konza Prairie Results after 3 yearsindicate that the abundance and rich-ness of annual forbs and the spatialheterogeneity of biomass and coverare higher in sites with bison than insites with cattle No dramatic differ-ences have been detected howeverbetween cattle- and bison-grazed sitesin cover of the dominant C4 grass Agerardii or the dominant forb Am-brosia psilostachya total plant spe-cies richness is also not dramaticallydifferent (E Gene Towne and DavidC Hartnett unpublished data)

Results at Konza Prairie are con-sistent with previous assessments(eg Schwartz and Ellis [1981] VanVuren and Bray [1983] and Plumband Dodd [1993] in mixed and short-grass prairie) which noted that bothbison and cattle are generalist herbi-vores that graze preferentially ongraminoids Nevertheless some dif-ferences in the foraging patterns ofbison and cattle have been docu-mented that may have long-term im-plications for grasslands For examplebison have a higher proportion ofgraminoids in their diet than docattle consequently forb and browsespecies are more common in cattlediets (Van Vuren and Bray 1983Hartnett et al 1997) Also bisonspend less time grazing than cattleand more time in nonfeeding activi-ties (Plumb and Dodd 1993) andbison strongly prefer open grasslandareas for grazing whereas cattle use

wooded and grassland habitats op-portunistically (Hartnett et al 1997)

Studies that have focused exclu-sively on cattle generally concur thattheir grazing activities increase spa-tial heterogeneity and enhance plantspecies diversity so long as stockingdensity is not too high (Collins 1987Hartnett et al 1996) Because bisongrazing in tallgrass prairie has a simi-lar effect one could conclude thateither herbivore can alter resourceavailability and heterogeneity andreduce the cover of the dominantgrasses sufficiently to enhance thesuccess of the subdominant speciesPerhaps of greater importance thandifferences in foraging patterns be-tween bison and cattle however arethe number of nongrazing activitiessuch as wallowing and horning (ierubbing on trees) that are associatedexclusively with bison (Coppedgeand Shaw 1997 Hartnett et al 1997)These activities when combined withthe spatial redistribution of nutri-ents and selective consumption ofthe dominant grasses may furtherincrease plant species richness andresource heterogeneity particularlyat the landscape scale

Nevertheless it is likely that be-cause bison and cattle are function-ally similar as large grass-feedingherbivores management strategies(stocking intensity and duration) willhave a greater influence on the de-gree of ecological equivalencyachieved than inherent differences inthese ungulates (Hartnett et al1997) Clearly the degree of over-lap in diet and foraging patterns isgreater between bison and cattle thanbetween cattle and other historicallyimportant native herbivores (Hart-nett et al 1997) such as antelope(Antilocapra americana) deer (Odo-coileus virginianus) and elk (Cervuscanadensis) Indeed the loss of ante-lope and elk from the tallgrass prai-rie coupled with dramatic increasesin deer populations presents addi-tional challenges for managing theseecosystems

Conservation implicationsConserving small and moderate-sizedtracts of once-vast biomes such asthe tallgrass prairie presents a uniqueset of problems that are distinct fromthose associated with spatially re-

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

44 BioScience Vol 49 No 1

Given that bison prefer grazingpatches that are initially dominatedby C4 grasses but that their selectiveforaging and reselection habits con-vert these patches to sites with agreater abundance of nonforage spe-cies it is likely that patch locationsare spatially dynamic across the land-scape Patches can ldquomoverdquo by twomechanisms patch abandonmentfollowed by selection of a new patchor patch migration wherein portionsof a patch are abandoned and thepatch expands into adjacent areasOver a 3-year period Catchpole(1996) found the rate of patch aban-donment to be approximately 6ndash7per year in both burned and un-burned watersheds thus at leastportions of previously establishedgrazing patches were reselected at ahigh rate (Figure 4) However whengrazing patches were mapped andcompared across years the extent ofspatial reselectionmdashalthough highlyvariable due to differences in totalburned area available to bison in anyone yearmdashaveraged approximately50 per year Thus grazing patchesin both burned and unburned water-sheds appear to migrate significantlyfrom year to year This local migra-tion permits periodic release of por-tions of the grassland from grazingpressures (Figure 4) and provides amechanism for recovery of below-ground carbohydrate storage reservesand production potential

At the watershed and landscapescales the long-term consequencesof bison activities include a reductionin cover dominance and productivityof grasses the competitive release ofmany subdominant species resultingin an increase in the abundance offorbs an overall increase in plantspecies richness and diversity andincreased spatial heterogeneity (Fig-ure 5 Hartnett et al 1996) Althoughalterations in plant community com-position can be attributed in largepart to the direct effects of grazingby bison increased plant species rich-ness is also likely to be a product ofincreased microsite diversity gener-ated by nongrazing activities suchas dung and urine deposition tram-pling and wallowing These andother bison activities contribute sig-nificantly to the increase in spatialheterogeneity that is characteristicof grazed tallgrass prairie (Figure 5)

Other impacts of bisonin tallgrass prairie

Effects of ungulates especially graz-ing on plant community composi-tion and structure have been studiedin many grasslands worldwide(McNaughton 1984 Milchunas etal 1988 Frank and McNaughton1992 Milchunas and Lauenroth1993) Ungulate activities howeveraffect many other aspects of grass-land structure and function includingthe physical structure of the environ-ment and the rates of a number ofecosystem-level processes (McNaugh-ton 1993 Frank and Evans 1997McNaughton et al 1997) There areseveral other important mechanismsby which bison alter ecosystem-levelprocesses and physical habitat struc-ture in tallgrass prairie

Nutrient redistribution and cyclingBison can substantially alter nutri-ent cycling processes and patterns ofnutrient availability in tallgrass prai-rie Their effects on nitrogen cyclingare critical because nitrogen avail-ability often limits plant productiv-ity in these grasslands (Seastedt et al1991 Blair 1997 Turner et al 1997)and influences plant species compo-sition (Gibson et al 1993 Wedinand Tilman 1993) Simulation mod-els of tallgrass prairie responses tograzing (Risser and Parton 1982)and studies of grazers in other grass-lands (Frank and Evans 1997McNaughton et al 1997) have dem-onstrated a disproportionate influ-ence of ungulates including bisonon the regulation of nitrogen cyclingprocesses Preliminary data fromKonza Prairie suggest that bison aresimilarly important in controllingnitrogen cycling in tallgrass prairie

Bison influence nitrogen cyclingconservation and availability intallgrass prairie ecosystems by alter-ing several soil and plant processesUngulates in grasslands consumerelatively recalcitrant plant biomassand return labile forms of nitrogen(ie urine) to soils (Ruess andMcNaughton 1988) thus bypassingthe otherwise slow mineralization ofnitrogen in plant litter Nitrogen inbison urine is largely urea which canbe hydrolyzed to ammonium in amatter of days (Ruess and McNaugh-ton 1988) Indeed application of

synthetic bison urine increased con-centrations of ammonium and ni-trate in Konza Prairie soils over 130-fold and 30-fold respectively 8 daysafter application (J R Matchett andLoretta C Johnson unpublished data)

Bison grazing can decrease theexport of nitrogen from tallgrassprairie by altering the magnitudes oftwo major pathways of nitrogenlossmdashcombustion and ammoniavolatilization Fire is the major path-way of nitrogen loss from ungrazedtallgrass prairie (Dodds et al 1996Blair 1997) nitrogen loss from burn-ing averages 1ndash4 gmiddotmndash2middotyrndash1 (Blair etal 1998) Grazing lowers combus-tion losses of nitrogen in tallgrassprairie by reducing the abovegroundplant detritus and increasing thepatchiness of a prairie fire (Figure 2Hobbs et al 1991) Although vola-tilization of ammonia can be in-creased by grazing in other types ofgrassland (Detling 1988) Hobbs etal (1991) suggested that any increasein ammonia volatilization in tallgrassprairie will be more than compen-sated for by a reduction in combus-tion losses of nitrogen

Finally bison grazing affects theamount and quality of plant litter re-turned to soils Grazing increases plantuptake of nutrients (Ruess 1984) andshoot nitrogen content in many grass-lands (Holland and Detling 1990Milchunas et al 1995) includingtallgrass prairie (Turner et al 1993)However the effects of grazers onroot growth and chemistry varyamong grasslands (Milchunas andLauenroth 1993) On Konza Prairieroot productivity and root biomasswere 30 and 20 lower respec-tively in bison grazing lawns than inungrazed exclosures In addition thenitrogen concentration of new rootgrowth in bison grazing lawns atKonza Prairie increased significantlyfrom 06 to 09 and the CN ratioof roots decreased A lower CN ratioreduces microbial immobilization andenhances nitrogen availability withingrazed areas Indeed recent studieson Konza Prairie indicated that netnitrogen mineralization in bison graz-ing lawns was 153 greater and netnitrification 126 greater than inungrazed prairie (Figure 6) Further-more net nitrogen mineralization rateswere proportional to the intensity ofbison use of a given area Thus the

January 1999 45

net effect of bison grazing appears tobe increased rates of nitrogen cyclingcoupled with a significant increase inspatial heterogeneity in nitrogen avail-ability together these effects can alterpatterns of plant productivity and spe-cies composition in tallgrass prairie(Figure 5 Steinauer and Collins 1995)

Wallowing One aspect of bison be-havior that differs from that of cattleand is primarily a physical activityis wallowing Wallows in Flint Hillstallgrass prairie which are estab-lished primarily in level upland orlowland sites dramatically alter thepatch structure of this prairie Bisonwallows develop as the animals pawthe ground and roll in the exposedsoil Continued use of wallows bybulls cows and calves creates a soildepression of 3ndash5 m in diameter (and10ndash30 cm in depth) that is devoid ofvegetation These denuded patcheseither gradually revegetate or remainas bare soil depending on the fre-quency of revisitation by bison Withthe vast numbers of bison that onceoccupied the Great Plains these soildepressions were probably abundantand widespread features of the land-scape (England and DeVos 1969)For example a number of relic wal-lows had to be filled to level the play-ing field for the first University ofOklahoma home football game in 1895(University of Oklahoma Athletic De-partment 1986) Relic wallows stillexist in many areas where bison havenot occurred in the past 125 years

Environmental conditions in relicand newly established wallowsstrongly influence prairie patch dy-namics (Polley and Collins 1984Polley and Wallace 1986) Becauseof soil compaction wallows oftenretain rainwater in the spring creat-ing localized habitats that are suit-able for ephemeral wetland speciessimilar to vernal pools in California(Holland and Jain 1981 Uno 1989)In the summer however the samewallows support only plants that cantolerate severe drought Vegetationcomposition and structure in wal-lows is different from that in thesurrounding prairie (Polley andCollins 1984) and these differencesare enhanced by fire (Collins andUno 1983) which may not spreadthrough wallows because of low fuelloads Consequently at larger spa-

tial scales grazed prairie that con-tains bison wallows has higher plantspecies diversity than grazed prairiewithout wallows (Collins and Bar-ber 1985) Thus bison can physi-cally alter grasslands in ways thatincrease environmental heterogeneityand enhance both local and regionalbiodiversity (Hartnett et al 1997)

Bison carcasses Bison not only af-fect vegetation patterns and soil pro-cesses through their grazing activi-ties but also have profound andlasting localized effects after theydie Although legal requirements andmanagement practices dictate theremoval of carcasses of domesticherbivores from public and privategrasslands native herbivores rou-tinely die of natural causes and theirbodies remain in situ As part of theminimal management strategy atKonza Prairie bison that die on siteare not removed As a result thesecarcasses create unique local distur-bances (Figure 7) that are the focusof studies to assess their effects onsoil nutrients and vegetation re-sponses in tallgrass prairie

When an individual bison diescopious quantities of fluids (withhigh nitrogen concentration) are re-leased during decomposition Adultbison can weigh more than 800 kgand these carcasses typically kill un-derlying and adjacent plants creat-ing a denuded zone of 4ndash6 m2 (Figure7) Although the fluids that are ini-tially released are toxic to vegeta-tion these sites eventually becomezones of high fertility For examplesoil cores extracted from the centerof carcass sites on Konza Prairie 3years after death had inorganic ni-trogen concentrations that were twoto three times higher than the sur-rounding prairie This nutrient enrich-ment may extend up to 25 m awayfrom the original carcass site andresults in patches dominated initiallyby early successional species Theaboveground primary production inthese patches is two to three timeshigher than in undisturbed prairie

Although disturbances created bybison carcasses are sporadic and lo-calized on Konza Prairie they pro-vide nutrient pulses that exceed allother natural processes even urineand fecal deposits Overall we canonly speculate about historical rates

of bison mortality Given the enor-mous size of the bison populationbefore their widespread slaughter inthe 1800s annual mortality wasprobably high High death rateswould have been especially commonduring droughts when there wouldbe the potential for large numbers ofcarcasses to occur across the land-scape Even though predators andscavengers may have consumed andrelocated many of these carcassesdecomposition of the remaining andpartial carcasses would still have re-sulted in patches of locally high nu-trient concentration Thus althoughit was variable bison mortality wouldhave led to a continual cycle of dis-turbance and recovery of thesepatches in presettlement grasslands

Are bison keystone speciesThe net effects of selective bison graz-ing activities at the landscape patchand individual plant level includeshifts in plant species compositionalterations of the physical and chemi-cal environment and increased spa-tial and temporal heterogeneity invegetation structure soil resourceavailability and a variety of ecosys-tem processes (Figures 4 5 and 6)Before bison reintroduction at KonzaPrairie the long-term burning ex-periments produced clear patterns ofresponse in the vegetation As firefrequency increased the dominanceof C4 grasses increased and the coverof C3 grasses forbs and woody spe-cies decreased (Figure 4 Gibson andHulbert 1987) Overall plant spe-cies diversity declined as fire fre-quency increased in ungrazed tall-grass prairie (Collins et al 1995)

These patterns in community struc-ture which had developed over 20years of burning treatments at KonzaPrairie are being rapidly and dra-matically altered by the grazing ac-tivity of the reintroduced bison Inparticular grazing by bison has low-ered the abundance of the dominant C4

grasses increased the abundance ofthe subdominant C3 grasses and forbsand markedly increased plant speciesdiversity (by 23) richness (by 38)and community heterogeneity (by13) relative to ungrazed sites evenunder annual burning conditions(Hartnett et al 1996 Collins andSteinauer 1998 Collins et al 1998)

46 BioScience Vol 49 No 1

Because of the multiple and dra-matic effects of bison on this land-scape we believe that bison are key-stone species in the tallgrass prairieOther authors have noted the poten-tial of large grazing mammals to actas ldquokeystone herbivoresrdquo capable ofmaintaining open grassland vegeta-tion that would otherwise undergosuccession to shrubland or wood-land (Owen-Smith 1987) Indeed thedisappearance of a grazing mega-fauna at the end of the Pleistocenemay have played a major role in thewidespread transition from steppe totundra at that time (Zimov et al 1995)However the concept of keystonespecies has been controversial since itsinception (Power et al 1996) One ofthe problems with this concept hasbeen the variable interpretation of cri-teria by which species are determinedto be keystone Power et al (1996)consider a keystone species to be ldquoonewhose impact on its community orecosystem is large and disproportion-ately large relative to its abundancerdquoTo make this definition operationalthese authors proposed a measure ofcommunity importance (CI) to be usedas an index of the strength of theimpact of a given species

CI = [(tN ndash tD)tN][1pi]

where tN is a quantitative measure ofa trait (eg diversity) in an intactcommunity tD is the measure of thetrait when species i has been deletedand pi is the proportional abundance(biomass) of species i before it wasdeleted CI values ldquomuch greater than1rdquo indicate that a species is key-stone We estimate bison biomass atKonza Prairie to be approximately11ndash12 gmiddotm2 (Collins and Steinauer1998) which is approximately 1of the total vegetative biomass OnKonza Prairie diversity is signifi-cantly higher (10ndash33) on grazedsites than ungrazed sites (Hartnett etal 1996) and these values yield arange of CIs from 6 to 25 For thisreason and others we consider bisonto be keystone species in tallgrassprairie ecosystems

Are bison and cattle functionalequivalents in tallgrass prairieThe historical presence of immenseherds of large ungulates in Great

Plains grasslands is undisputed(McHugh 1972) and we have em-phasized the keystone role that bisonplayed in determining the structureand function of tallgrass prairies atmultiple spatial and temporal scalesWith the replacement of native bisonby domesticated cattle in the remain-ing grasslands an obvious issue isthe degree of similarity between thesetwo ungulates with respect to theireffects on tallgrass prairie In otherwords can bison and cattle be con-sidered ecological equivalents

There have been several previousattempts to answer this questionbut the results have been equivocalat best (Plumb and Dodd 1993Hartnett et al 1997) The primarybarrier to resolving this issue restswith a lack of comparative studies inwhich management is held constantand the type of grazer is varied Suchstudies have recently been initiatedat Konza Prairie Results after 3 yearsindicate that the abundance and rich-ness of annual forbs and the spatialheterogeneity of biomass and coverare higher in sites with bison than insites with cattle No dramatic differ-ences have been detected howeverbetween cattle- and bison-grazed sitesin cover of the dominant C4 grass Agerardii or the dominant forb Am-brosia psilostachya total plant spe-cies richness is also not dramaticallydifferent (E Gene Towne and DavidC Hartnett unpublished data)

Results at Konza Prairie are con-sistent with previous assessments(eg Schwartz and Ellis [1981] VanVuren and Bray [1983] and Plumband Dodd [1993] in mixed and short-grass prairie) which noted that bothbison and cattle are generalist herbi-vores that graze preferentially ongraminoids Nevertheless some dif-ferences in the foraging patterns ofbison and cattle have been docu-mented that may have long-term im-plications for grasslands For examplebison have a higher proportion ofgraminoids in their diet than docattle consequently forb and browsespecies are more common in cattlediets (Van Vuren and Bray 1983Hartnett et al 1997) Also bisonspend less time grazing than cattleand more time in nonfeeding activi-ties (Plumb and Dodd 1993) andbison strongly prefer open grasslandareas for grazing whereas cattle use

wooded and grassland habitats op-portunistically (Hartnett et al 1997)

Studies that have focused exclu-sively on cattle generally concur thattheir grazing activities increase spa-tial heterogeneity and enhance plantspecies diversity so long as stockingdensity is not too high (Collins 1987Hartnett et al 1996) Because bisongrazing in tallgrass prairie has a simi-lar effect one could conclude thateither herbivore can alter resourceavailability and heterogeneity andreduce the cover of the dominantgrasses sufficiently to enhance thesuccess of the subdominant speciesPerhaps of greater importance thandifferences in foraging patterns be-tween bison and cattle however arethe number of nongrazing activitiessuch as wallowing and horning (ierubbing on trees) that are associatedexclusively with bison (Coppedgeand Shaw 1997 Hartnett et al 1997)These activities when combined withthe spatial redistribution of nutri-ents and selective consumption ofthe dominant grasses may furtherincrease plant species richness andresource heterogeneity particularlyat the landscape scale

Nevertheless it is likely that be-cause bison and cattle are function-ally similar as large grass-feedingherbivores management strategies(stocking intensity and duration) willhave a greater influence on the de-gree of ecological equivalencyachieved than inherent differences inthese ungulates (Hartnett et al1997) Clearly the degree of over-lap in diet and foraging patterns isgreater between bison and cattle thanbetween cattle and other historicallyimportant native herbivores (Hart-nett et al 1997) such as antelope(Antilocapra americana) deer (Odo-coileus virginianus) and elk (Cervuscanadensis) Indeed the loss of ante-lope and elk from the tallgrass prai-rie coupled with dramatic increasesin deer populations presents addi-tional challenges for managing theseecosystems

Conservation implicationsConserving small and moderate-sizedtracts of once-vast biomes such asthe tallgrass prairie presents a uniqueset of problems that are distinct fromthose associated with spatially re-

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

January 1999 45

net effect of bison grazing appears tobe increased rates of nitrogen cyclingcoupled with a significant increase inspatial heterogeneity in nitrogen avail-ability together these effects can alterpatterns of plant productivity and spe-cies composition in tallgrass prairie(Figure 5 Steinauer and Collins 1995)

Wallowing One aspect of bison be-havior that differs from that of cattleand is primarily a physical activityis wallowing Wallows in Flint Hillstallgrass prairie which are estab-lished primarily in level upland orlowland sites dramatically alter thepatch structure of this prairie Bisonwallows develop as the animals pawthe ground and roll in the exposedsoil Continued use of wallows bybulls cows and calves creates a soildepression of 3ndash5 m in diameter (and10ndash30 cm in depth) that is devoid ofvegetation These denuded patcheseither gradually revegetate or remainas bare soil depending on the fre-quency of revisitation by bison Withthe vast numbers of bison that onceoccupied the Great Plains these soildepressions were probably abundantand widespread features of the land-scape (England and DeVos 1969)For example a number of relic wal-lows had to be filled to level the play-ing field for the first University ofOklahoma home football game in 1895(University of Oklahoma Athletic De-partment 1986) Relic wallows stillexist in many areas where bison havenot occurred in the past 125 years

Environmental conditions in relicand newly established wallowsstrongly influence prairie patch dy-namics (Polley and Collins 1984Polley and Wallace 1986) Becauseof soil compaction wallows oftenretain rainwater in the spring creat-ing localized habitats that are suit-able for ephemeral wetland speciessimilar to vernal pools in California(Holland and Jain 1981 Uno 1989)In the summer however the samewallows support only plants that cantolerate severe drought Vegetationcomposition and structure in wal-lows is different from that in thesurrounding prairie (Polley andCollins 1984) and these differencesare enhanced by fire (Collins andUno 1983) which may not spreadthrough wallows because of low fuelloads Consequently at larger spa-

tial scales grazed prairie that con-tains bison wallows has higher plantspecies diversity than grazed prairiewithout wallows (Collins and Bar-ber 1985) Thus bison can physi-cally alter grasslands in ways thatincrease environmental heterogeneityand enhance both local and regionalbiodiversity (Hartnett et al 1997)

Bison carcasses Bison not only af-fect vegetation patterns and soil pro-cesses through their grazing activi-ties but also have profound andlasting localized effects after theydie Although legal requirements andmanagement practices dictate theremoval of carcasses of domesticherbivores from public and privategrasslands native herbivores rou-tinely die of natural causes and theirbodies remain in situ As part of theminimal management strategy atKonza Prairie bison that die on siteare not removed As a result thesecarcasses create unique local distur-bances (Figure 7) that are the focusof studies to assess their effects onsoil nutrients and vegetation re-sponses in tallgrass prairie

When an individual bison diescopious quantities of fluids (withhigh nitrogen concentration) are re-leased during decomposition Adultbison can weigh more than 800 kgand these carcasses typically kill un-derlying and adjacent plants creat-ing a denuded zone of 4ndash6 m2 (Figure7) Although the fluids that are ini-tially released are toxic to vegeta-tion these sites eventually becomezones of high fertility For examplesoil cores extracted from the centerof carcass sites on Konza Prairie 3years after death had inorganic ni-trogen concentrations that were twoto three times higher than the sur-rounding prairie This nutrient enrich-ment may extend up to 25 m awayfrom the original carcass site andresults in patches dominated initiallyby early successional species Theaboveground primary production inthese patches is two to three timeshigher than in undisturbed prairie

Although disturbances created bybison carcasses are sporadic and lo-calized on Konza Prairie they pro-vide nutrient pulses that exceed allother natural processes even urineand fecal deposits Overall we canonly speculate about historical rates

of bison mortality Given the enor-mous size of the bison populationbefore their widespread slaughter inthe 1800s annual mortality wasprobably high High death rateswould have been especially commonduring droughts when there wouldbe the potential for large numbers ofcarcasses to occur across the land-scape Even though predators andscavengers may have consumed andrelocated many of these carcassesdecomposition of the remaining andpartial carcasses would still have re-sulted in patches of locally high nu-trient concentration Thus althoughit was variable bison mortality wouldhave led to a continual cycle of dis-turbance and recovery of thesepatches in presettlement grasslands

Are bison keystone speciesThe net effects of selective bison graz-ing activities at the landscape patchand individual plant level includeshifts in plant species compositionalterations of the physical and chemi-cal environment and increased spa-tial and temporal heterogeneity invegetation structure soil resourceavailability and a variety of ecosys-tem processes (Figures 4 5 and 6)Before bison reintroduction at KonzaPrairie the long-term burning ex-periments produced clear patterns ofresponse in the vegetation As firefrequency increased the dominanceof C4 grasses increased and the coverof C3 grasses forbs and woody spe-cies decreased (Figure 4 Gibson andHulbert 1987) Overall plant spe-cies diversity declined as fire fre-quency increased in ungrazed tall-grass prairie (Collins et al 1995)

These patterns in community struc-ture which had developed over 20years of burning treatments at KonzaPrairie are being rapidly and dra-matically altered by the grazing ac-tivity of the reintroduced bison Inparticular grazing by bison has low-ered the abundance of the dominant C4

grasses increased the abundance ofthe subdominant C3 grasses and forbsand markedly increased plant speciesdiversity (by 23) richness (by 38)and community heterogeneity (by13) relative to ungrazed sites evenunder annual burning conditions(Hartnett et al 1996 Collins andSteinauer 1998 Collins et al 1998)

46 BioScience Vol 49 No 1

Because of the multiple and dra-matic effects of bison on this land-scape we believe that bison are key-stone species in the tallgrass prairieOther authors have noted the poten-tial of large grazing mammals to actas ldquokeystone herbivoresrdquo capable ofmaintaining open grassland vegeta-tion that would otherwise undergosuccession to shrubland or wood-land (Owen-Smith 1987) Indeed thedisappearance of a grazing mega-fauna at the end of the Pleistocenemay have played a major role in thewidespread transition from steppe totundra at that time (Zimov et al 1995)However the concept of keystonespecies has been controversial since itsinception (Power et al 1996) One ofthe problems with this concept hasbeen the variable interpretation of cri-teria by which species are determinedto be keystone Power et al (1996)consider a keystone species to be ldquoonewhose impact on its community orecosystem is large and disproportion-ately large relative to its abundancerdquoTo make this definition operationalthese authors proposed a measure ofcommunity importance (CI) to be usedas an index of the strength of theimpact of a given species

CI = [(tN ndash tD)tN][1pi]

where tN is a quantitative measure ofa trait (eg diversity) in an intactcommunity tD is the measure of thetrait when species i has been deletedand pi is the proportional abundance(biomass) of species i before it wasdeleted CI values ldquomuch greater than1rdquo indicate that a species is key-stone We estimate bison biomass atKonza Prairie to be approximately11ndash12 gmiddotm2 (Collins and Steinauer1998) which is approximately 1of the total vegetative biomass OnKonza Prairie diversity is signifi-cantly higher (10ndash33) on grazedsites than ungrazed sites (Hartnett etal 1996) and these values yield arange of CIs from 6 to 25 For thisreason and others we consider bisonto be keystone species in tallgrassprairie ecosystems

Are bison and cattle functionalequivalents in tallgrass prairieThe historical presence of immenseherds of large ungulates in Great

Plains grasslands is undisputed(McHugh 1972) and we have em-phasized the keystone role that bisonplayed in determining the structureand function of tallgrass prairies atmultiple spatial and temporal scalesWith the replacement of native bisonby domesticated cattle in the remain-ing grasslands an obvious issue isthe degree of similarity between thesetwo ungulates with respect to theireffects on tallgrass prairie In otherwords can bison and cattle be con-sidered ecological equivalents

There have been several previousattempts to answer this questionbut the results have been equivocalat best (Plumb and Dodd 1993Hartnett et al 1997) The primarybarrier to resolving this issue restswith a lack of comparative studies inwhich management is held constantand the type of grazer is varied Suchstudies have recently been initiatedat Konza Prairie Results after 3 yearsindicate that the abundance and rich-ness of annual forbs and the spatialheterogeneity of biomass and coverare higher in sites with bison than insites with cattle No dramatic differ-ences have been detected howeverbetween cattle- and bison-grazed sitesin cover of the dominant C4 grass Agerardii or the dominant forb Am-brosia psilostachya total plant spe-cies richness is also not dramaticallydifferent (E Gene Towne and DavidC Hartnett unpublished data)

Results at Konza Prairie are con-sistent with previous assessments(eg Schwartz and Ellis [1981] VanVuren and Bray [1983] and Plumband Dodd [1993] in mixed and short-grass prairie) which noted that bothbison and cattle are generalist herbi-vores that graze preferentially ongraminoids Nevertheless some dif-ferences in the foraging patterns ofbison and cattle have been docu-mented that may have long-term im-plications for grasslands For examplebison have a higher proportion ofgraminoids in their diet than docattle consequently forb and browsespecies are more common in cattlediets (Van Vuren and Bray 1983Hartnett et al 1997) Also bisonspend less time grazing than cattleand more time in nonfeeding activi-ties (Plumb and Dodd 1993) andbison strongly prefer open grasslandareas for grazing whereas cattle use

wooded and grassland habitats op-portunistically (Hartnett et al 1997)

Studies that have focused exclu-sively on cattle generally concur thattheir grazing activities increase spa-tial heterogeneity and enhance plantspecies diversity so long as stockingdensity is not too high (Collins 1987Hartnett et al 1996) Because bisongrazing in tallgrass prairie has a simi-lar effect one could conclude thateither herbivore can alter resourceavailability and heterogeneity andreduce the cover of the dominantgrasses sufficiently to enhance thesuccess of the subdominant speciesPerhaps of greater importance thandifferences in foraging patterns be-tween bison and cattle however arethe number of nongrazing activitiessuch as wallowing and horning (ierubbing on trees) that are associatedexclusively with bison (Coppedgeand Shaw 1997 Hartnett et al 1997)These activities when combined withthe spatial redistribution of nutri-ents and selective consumption ofthe dominant grasses may furtherincrease plant species richness andresource heterogeneity particularlyat the landscape scale

Nevertheless it is likely that be-cause bison and cattle are function-ally similar as large grass-feedingherbivores management strategies(stocking intensity and duration) willhave a greater influence on the de-gree of ecological equivalencyachieved than inherent differences inthese ungulates (Hartnett et al1997) Clearly the degree of over-lap in diet and foraging patterns isgreater between bison and cattle thanbetween cattle and other historicallyimportant native herbivores (Hart-nett et al 1997) such as antelope(Antilocapra americana) deer (Odo-coileus virginianus) and elk (Cervuscanadensis) Indeed the loss of ante-lope and elk from the tallgrass prai-rie coupled with dramatic increasesin deer populations presents addi-tional challenges for managing theseecosystems

Conservation implicationsConserving small and moderate-sizedtracts of once-vast biomes such asthe tallgrass prairie presents a uniqueset of problems that are distinct fromthose associated with spatially re-

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

46 BioScience Vol 49 No 1

Because of the multiple and dra-matic effects of bison on this land-scape we believe that bison are key-stone species in the tallgrass prairieOther authors have noted the poten-tial of large grazing mammals to actas ldquokeystone herbivoresrdquo capable ofmaintaining open grassland vegeta-tion that would otherwise undergosuccession to shrubland or wood-land (Owen-Smith 1987) Indeed thedisappearance of a grazing mega-fauna at the end of the Pleistocenemay have played a major role in thewidespread transition from steppe totundra at that time (Zimov et al 1995)However the concept of keystonespecies has been controversial since itsinception (Power et al 1996) One ofthe problems with this concept hasbeen the variable interpretation of cri-teria by which species are determinedto be keystone Power et al (1996)consider a keystone species to be ldquoonewhose impact on its community orecosystem is large and disproportion-ately large relative to its abundancerdquoTo make this definition operationalthese authors proposed a measure ofcommunity importance (CI) to be usedas an index of the strength of theimpact of a given species

CI = [(tN ndash tD)tN][1pi]

where tN is a quantitative measure ofa trait (eg diversity) in an intactcommunity tD is the measure of thetrait when species i has been deletedand pi is the proportional abundance(biomass) of species i before it wasdeleted CI values ldquomuch greater than1rdquo indicate that a species is key-stone We estimate bison biomass atKonza Prairie to be approximately11ndash12 gmiddotm2 (Collins and Steinauer1998) which is approximately 1of the total vegetative biomass OnKonza Prairie diversity is signifi-cantly higher (10ndash33) on grazedsites than ungrazed sites (Hartnett etal 1996) and these values yield arange of CIs from 6 to 25 For thisreason and others we consider bisonto be keystone species in tallgrassprairie ecosystems

Are bison and cattle functionalequivalents in tallgrass prairieThe historical presence of immenseherds of large ungulates in Great

Plains grasslands is undisputed(McHugh 1972) and we have em-phasized the keystone role that bisonplayed in determining the structureand function of tallgrass prairies atmultiple spatial and temporal scalesWith the replacement of native bisonby domesticated cattle in the remain-ing grasslands an obvious issue isthe degree of similarity between thesetwo ungulates with respect to theireffects on tallgrass prairie In otherwords can bison and cattle be con-sidered ecological equivalents

There have been several previousattempts to answer this questionbut the results have been equivocalat best (Plumb and Dodd 1993Hartnett et al 1997) The primarybarrier to resolving this issue restswith a lack of comparative studies inwhich management is held constantand the type of grazer is varied Suchstudies have recently been initiatedat Konza Prairie Results after 3 yearsindicate that the abundance and rich-ness of annual forbs and the spatialheterogeneity of biomass and coverare higher in sites with bison than insites with cattle No dramatic differ-ences have been detected howeverbetween cattle- and bison-grazed sitesin cover of the dominant C4 grass Agerardii or the dominant forb Am-brosia psilostachya total plant spe-cies richness is also not dramaticallydifferent (E Gene Towne and DavidC Hartnett unpublished data)

Results at Konza Prairie are con-sistent with previous assessments(eg Schwartz and Ellis [1981] VanVuren and Bray [1983] and Plumband Dodd [1993] in mixed and short-grass prairie) which noted that bothbison and cattle are generalist herbi-vores that graze preferentially ongraminoids Nevertheless some dif-ferences in the foraging patterns ofbison and cattle have been docu-mented that may have long-term im-plications for grasslands For examplebison have a higher proportion ofgraminoids in their diet than docattle consequently forb and browsespecies are more common in cattlediets (Van Vuren and Bray 1983Hartnett et al 1997) Also bisonspend less time grazing than cattleand more time in nonfeeding activi-ties (Plumb and Dodd 1993) andbison strongly prefer open grasslandareas for grazing whereas cattle use

wooded and grassland habitats op-portunistically (Hartnett et al 1997)

Studies that have focused exclu-sively on cattle generally concur thattheir grazing activities increase spa-tial heterogeneity and enhance plantspecies diversity so long as stockingdensity is not too high (Collins 1987Hartnett et al 1996) Because bisongrazing in tallgrass prairie has a simi-lar effect one could conclude thateither herbivore can alter resourceavailability and heterogeneity andreduce the cover of the dominantgrasses sufficiently to enhance thesuccess of the subdominant speciesPerhaps of greater importance thandifferences in foraging patterns be-tween bison and cattle however arethe number of nongrazing activitiessuch as wallowing and horning (ierubbing on trees) that are associatedexclusively with bison (Coppedgeand Shaw 1997 Hartnett et al 1997)These activities when combined withthe spatial redistribution of nutri-ents and selective consumption ofthe dominant grasses may furtherincrease plant species richness andresource heterogeneity particularlyat the landscape scale

Nevertheless it is likely that be-cause bison and cattle are function-ally similar as large grass-feedingherbivores management strategies(stocking intensity and duration) willhave a greater influence on the de-gree of ecological equivalencyachieved than inherent differences inthese ungulates (Hartnett et al1997) Clearly the degree of over-lap in diet and foraging patterns isgreater between bison and cattle thanbetween cattle and other historicallyimportant native herbivores (Hart-nett et al 1997) such as antelope(Antilocapra americana) deer (Odo-coileus virginianus) and elk (Cervuscanadensis) Indeed the loss of ante-lope and elk from the tallgrass prai-rie coupled with dramatic increasesin deer populations presents addi-tional challenges for managing theseecosystems

Conservation implicationsConserving small and moderate-sizedtracts of once-vast biomes such asthe tallgrass prairie presents a uniqueset of problems that are distinct fromthose associated with spatially re-

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

January 1999 47

stricted ecosystems because many ofthe defining forces that historicallywere important in structuring thesesystems occurred at spatial scalesthat no longer exist For example inpre-1900s grasslands fires were notplot-level or even watershed-levelevents but operated at spatial scalesencompassing thousands of hectaresThis large spatial scale resulted in po-tentially high fire frequenciesthroughout the tallgrass prairie be-cause any point of ignition in thisldquoinland sea of grassrdquo could affect grass-lands hundreds of kilometers distant

Today the fragmentation of GreatPlains grasslands is recognized as akey factor in reducing the frequencyof fire which in turn contributes tospecies loss (Leach and Givnish1996) Indeed the primary manage-ment strategy for small prairie pre-serves which are most prone to in-vasion by woody vegetation andexotic species is to burn them asfrequently as possible to suppressinvasion by undesirable plants (Leachand Givnish 1996) Unfortunatelyfrequent (annual or biannual) springfire maintains dominance by C4grasses but reduces plant species di-versity relative to grasslands that areburned infrequently One alterna-tive is to conduct burns at differenttimes of the year (Howe 1994) butsummer fires for example may notprevent invasion or reduce the abun-dance of woody vegetation (Adamset al 1982) In addition burning inlate summer may be difficult becauseof other considerations includingreduced ability to control the fireunder dry windy conditions Notonly are prairies threatened by frag-mentation and invasion by undesir-able species but grasslands through-out the Great Plains are now affectedby increased atmospheric nitrogendeposition (Wedin and Tilman 1996)Thus remnant grasslands are sub-jected to a variety of anthropogenicfactors that can reduce the diversityof native prairie species

The spatial and temporal impactsof bison grazing activities caused bythe historically large and nomadicherds are also best characterized aslandscape-level forces These too aredifficult to replicate in todayrsquos frag-mented grassland remnants Yet justas some of the ecological character-istics of natural fires can be reintro-

duced to grasslands through pre-scribed fire the key elements of bi-son grazing activities can and shouldbe incorporated into conservationand restoration strategies for rem-nant prairies (Steuter 1997) Oneapproach to accomplish this goal isthe substitution of cattle for bisonPlumb and Dodd (1993) argued thatthe choice of whether to use cattle orbison as a management tool in grass-lands is scale and context dependentClearly reintroducing bison may notbe appropriate for small prairie rem-nants with public access and loweconomic resources But cattle man-aged for their ecological rather thantheir economic value may be suit-able in such cases

Alternatively mowing can be usedto reduce the dominance of the tallgrasses and to enhance species rich-ness (Gibson et al 1993 Collins andSteinauer 1998) Results from a long-term experiment at Konza Prairieincorporating annual fire nitrogenaddition and mowing (Collins et al1998) indicated that on annuallyburned and fertilized treatment plotsproductivity of the grasses washigher and plant species diversitylower than in control plots How-ever on burned fertilized plots thatwere mowed (with removal of thefoliage a rough substitute for graz-ing) plant species diversity was re-stored to levels similar to controlplots (Collins and Steinauer 1998)

Figure 5 Landscape-level changes in spatialheterogeneity in tall-grass prairie induced bybison grazing activitiesFalse-color compositeof Thematic Mapper(TM) data from an areaon Konza Prairie grazedby bison (upper left)and from a nearby areaprotected from grazing(upper right) Both siteswere burned and havesimilar soil types as-pect and slopes Redcolors represent areasof high productivityand blue colors corre-spond to areas of lowproductivity or bareground (lower panel)Percentage difference inspatial heterogeneity ofbiomass (estimatedfrom remotely sensedspectral reflectancedata) between areasgrazed by bison andadjacent ungrazed ar-eas on Konza PrairieBefore bison reintro-duction watershedsscheduled to remainungrazed appeared to have greater spatial heterogeneity than those scheduled to begrazed (negative values in 1987) After the reintroduction of bison in 1988 spatialheterogeneity in the grazed watersheds increased substantially Spatial heterogeneitywas assessed using the TEXTURE algorithm which involves passing a moving 3 acute 3pixel window through the images and determining the differences between the mini-mum and maximum values for each subset of pixels (Briggs and Nellis 1991) Pixelsrepresent derived Normalized Difference Vegetation Index (NDVI) values from TMdata (30 m2 spatial resolution) from 1988 to 1991 and from 1993 (TM data were notavailable for 1992) Previous studies have confirmed that the use of the TEXTUREalgorithm with NDVI data is useful for estimating patterns of spatial heterogeneity intallgrass prairie (Nellis and Briggs 1989 Briggs and Nellis 1991)

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

48 BioScience Vol 49 No 1

A combination of frequent springfire to maintain populations of thedesirable C4 prairie grasses decreasenitrogen availability (Blair 1997)and suppress growth of weedy annu-als and woody vegetation coupledwith mowing portions of the site toreduce the competitive dominanceof C4 grasses can enhance the abun-dance of forbs and maintain highplant species diversity in small rem-nant prairies Ultimately manage-ment designed to increase the spatialheterogeneity of resources in a man-ner analogous to that imposed byungulate activities is essential if sig-

nificant and sustainable biotic diver-sity in tallgrass prairie is a goal

ConclusionsDespite less than a decade of re-search at Konza Prairie on bisonndashtallgrass prairie interactions the key-stone role that bison must havehistorically played in this grasslandis clear Moreover much as fire isnow recognized as an essential com-ponent of tallgrass prairie manage-ment (because without fire this grass-land disappears) the need forreintroducing the forces of large un-

gulate herbivory to this grassland isevident Indeed it is the interactionof ungulate grazing activities andfire operating in a shifting mosaicacross the landscape that is key toconserving and restoring the bioticintegrity of the remaining tracts oftallgrass prairie

Before bison were reintroduced toKonza Prairie Knapp and Seastedt(1986) speculated that bison grazingand fire could act in similar ways byreducing the accumulation of detri-tus in this system It is primarily theblanketing effect of the accumula-tion of dead plant material aboveground that limits productivity inundisturbed tallgrass prairie Likefire bison grazing reduces above-ground standing dead biomass Butit is now clear that the unique spatialand temporal complexities of bisongrazing activities (Figure 5) are criti-cal to the successful maintenance ofbiotic diversity in this grassland Thisgrazing-induced heterogeneity con-trasts sharply with the spatial homo-geneity induced by fire in an ungrazedlandscape (Figure 6)

Tallgrass prairie by virtue of itsinherently variable climatic grazingand fire regimes is an ecosystem thatrequires long-term study to docu-ment patterns and quantify processes(Knapp et al 1998b) Through thepartnership of The Nature Conser-vancy the National Science Foun-dationrsquos LTER program and KansasState University ongoing studies atthis site will continue to explore theecological interactions of fire andgrazing in the tallgrass prairie land-scape Such research is timely be-cause conservation and managementissues have intensified in the remain-ing tracts of this once-vast biomeparticularly in response to predictedalterations in global climate and land-use changes Interdisciplinary eco-logical research such as that ongo-ing at Konza Prairie will provide thebasic information necessary for de-signing optimal conservation resto-ration and management strategiesin this and other grasslands

AcknowledgmentsResearch summarized here was sup-ported by the National Science Foun-dationrsquos Long-Term Studies Ecol-ogy and Ecosystems Programs the

Figure 6 Net nitrogenmineralization and netnitrification rates ingrazed and ungrazedareas of Konza PrairieMeasurements weremade in situ for a 1-month period in June1996 and May 1997using a modified bur-ied soil core technique(Raison et al 1987)Replicate study plotswere located on uplandareas of adjacentgrazed and ungrazedannually burned wa-tersheds Values are means plusmn 1 SE and are significantly different (P lt 005 n = 120per grazing treatment)

Figure 7 Bison carcass approximately 9 months after death in burned tallgrass prairieIn this spring (May) photo vegetation is completely lacking within the area in which theanimal died A zone of high fertility exists around the edge of this disturbance andwithin 1ndash2 years the site will be dominated by early successional (annual) species

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

January 1999 49

US Department of Agriculturersquos Na-tional Research Initiative Programthe National Aeronautics and SpaceAdministration The Nature Conser-vancy and the Kansas State Univer-sity Agricultural Experiment Station(99-156-J)

References citedAdams DE Anderson RC Collins SL 1982

Differential response of woody and her-baceous species to summer and winterburning in an Oklahoma grassland South-western Naturalist 27 55ndash61

Axelrod DI 1985 Rise of the grassland biomecentral North America The BotanicalReview 51 163ndash201

Berger J Cunningham C 1994 Bison Mat-ing and Conservation in Small PopulationsNew York Columbia University Press

Blair JM 1997 Fire N availability and plantresponse in grasslands A test of the tran-sient maxima hypothesis Ecology 782359ndash2368

Blair JM Seastedt TR Rice CW RamundoRA 1998 Terrestrial nutrient cycling intallgrass prairie Pages 222ndash243 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Briggs JM Knapp AK 1995 Interannualvariabili ty in primary production intallgrass prairie Climate soil moisturetopographic position and fire as determi-nants of aboveground biomass AmericanJournal of Botany 82 1024ndash1030

Briggs JM Nellis MD 1991 Seasonal varia-tion of heterogeneity in the tallgrass prai-rie A quantitative measure using remotesensing Photogrammetric Engineering andRemote Sensing 57 407ndash411

Catchpole FB 1996 The dynamics of bison(Bos bison) grazing patches in tallgrassprairie Masterrsquos thesis Kansas State Uni-versity Manhattan KS

Collins SL 1987 Interaction of disturbancesin tallgrass prairie A field experimentEcology 68 1243ndash1250

______ 1992 Fire frequency and communityheterogeneity in tallgrass prairie vegeta-tion Ecology 73 2001ndash2006

Collins SL Barber SC 1985 Effects of dis-turbance on diversity in mixed-grass prai-rie Vegetatio 64 87ndash94

Collins SL Steinauer EM 1998 Disturbancediversity and species interactions intallgrass prairie Pages 140ndash156 in KnappAK Briggs JM Hartnett DC Collins SCeds Grassland Dynamics Long-Term Eco-logical Research in Tallgrass Prairie NewYork Oxford University Press

Collins SL Uno GE 1983 The effect of earlyspring burning on vegetation in buffalowallows Bulletin of the Torrey BotanicalClub 110 474ndash481

Collins SL Glenn SM Gibson DJ 1995Experimental analysis of intermediate dis-turbance and initial floristic compositionDecoupling cause and effect Ecology 76486ndash492

Collins SL Knapp AK Briggs JM Blair JMSteinauer EM 1998 Modulation of di-

versity by grazing and mowing in nativetallgrass prairie Science 280 745ndash747

Coppedge BR Shaw JH 1997 Effects ofhorning and rubbing by bison (Bison bi-son) on woody vegetation in a tallgrassprairie landscape American MidlandNaturalist 138 189ndash196

______ 1998 Bison grazing patterns on sea-sonally burned tallgrass prairie Journalof Range Management 51 258ndash264

Coppock DL Detling JK 1986 Alteration ofbison and black-tailed prairie dog grazinginteraction by prescribed burning Jour-nal of Wildlife Management 50 452ndash455

Coppock DL Detling JK Ellis JE Dyer MI1983 Plantndashherbivore interactions in aNorth American mixed-grass prairie IEffects of black-tailed prairie dogs onintraseasonal aboveground plant biomassand nutrient dynamics and plant speciesdiversity Oecologia 56 1ndash9

Damhoureyeh SA Hartnett DC 1997 Ef-fects of bison and cattle on growth repro-duction and abundances of five tallgrassprairie forbs American Journal of Botany84 1719ndash1728

Day TA Detling JK 1990 Grassland patchdynamics and herbivore grazing prefer-ence following urine deposition Ecology63180ndash188

Detling JK 1988 Grasslands and SavannasRegulation of energy flow and nutrientcycling by herbivores Pages 131ndash148 inPomeroy LR Alberts JJ eds Concepts ofEcosystems Ecology Ecological Studies67 New York Springer-Verlag

Dodds WK Blair JM Henebry GM KoellikerJK Ramundo R Tate CM 1996 Nitro-gen transport from tallgrass prairie bystreams Journal of Environmental Qual-ity 25 973ndash987

England RE DeVos A 1969 Influence ofanimals on pristine conditions on the Ca-nadian grasslands Journal of Range Man-agement 22 87ndash94

Fahnestock JT Knapp AK 1993 Water rela-tions and growth of tallgrass prairie forbsin response to selective herbivory by bi-son International Journal of Plant Sci-ence 154 432ndash440

______ 1994 Responses of forbs and grassesto selective grazing by bison Interactionsbetween herbivory and water stressVegetatio 115 123ndash131

Flores D 1991 Bison ecology and bisondiplomacy The southern plains from 1800to 1850 Journal of American History 78465ndash485

Frank DA Evans RD 1997 Effects of nativegrazer s on gras sland N cycl ing inYellowstone National Park Ecology 782238ndash2248

Frank DA McNaughton SJ 1992 The ecol-ogy of plants large mammalian herbi-vores and drought in Yellowstone Na-tional Park Ecology 73 2043ndash2058

Freeman CC 1998 The flora of Konza Prai-rie A historical review and contemporarypatterns Pages 69ndash80 in Knapp AK BriggsJM Hartnett DC Collins SC eds Grass-land Dynamics Long-Term EcologicalResearch in Tallgrass Prairie New YorkOxford University Press

Gibson DJ Hulbert LC 1987 Effects of firetopography and year-to-year climate varia-tion on species composition in tallgrass

prairie Vegetatio 72 175ndash185Gibson DJ Seastedt TR Briggs JM 1993

Management practices in tallgrass prai-rie Large- and small-scale experimentaleffects on species composition Journal ofApplied Ecology 30 247ndash255

Glenn SM Collins SL 1990 Patch structurein tallgrass prairies Dynamics of satellitespecies Oikos 57 229ndash236

Hartnett DC Hickman KR Fischer-WalterLE 1996 Effects of bison grazing fireand topography on floristic diversity intallgrass prairie Journal of Range Man-agement 49 413ndash420

Hartnett DC Steuter AA Hickman KR 1997Comparative ecology of native versus in-troduced ungulates Pages 72ndash101 inKnopf F Samson F eds Ecology andConservation of Great Plains VertebratesNew York Springer-Verlag

Hobbs NT Schimel DS Owensby CE OjimaDJ 1991 Fire and grazing in tallgrassprairie Contingent effects on nitrogenbudgets Ecology 72 1374ndash1382

Holland EA Detling JK 1990 Plant responsesto herbivory and belowground nitrogencycling Ecology 71 1040ndash1049

Holland RF Jain SK 1981 Insular biogeog-raphy of vernal pools in the Central Val-ley of California American Naturalist 11724ndash37

Howe HF 1994 Response of early- and late-flowering plants of fire season in experi-mental prairies Ecological Applications4 121ndash133

Jones JK Hoffmann RS Rice DW Jones CBaker RJ Engstrom MD 1992 Revisedchecklist of North American mammalsnorth of Mexico 1991 The MuseumTexas Tech University Occasional Papers146 1ndash23

Knapp AK 1985 Effect of fire and droughton the ecophysiology of Andropogongerardii and Panicum virgatum in atallgrass prairie Ecology 66 1309ndash1320

Knapp AK Seastedt TR 1986 Detritus lim-its productivity in tallgras s prairieBioScience 26 662ndash668

Knapp AK Fahnestock JT Hamburg SPStatland LJ Seastedt TR Schimel DS1993 Landscape patterns in soilndashwaterrelations and primary production intallgrass prairie Ecology 74 549ndash560

Knapp AK Briggs JM Blair JM Turner CL1998a Patterns and controls of above-ground net primary production in tallgrassprairie Pages 193ndash221 in Knapp AKBriggs JM Hartnett DC Collins SC edsGrassland Dynamics Long-Term Ecologi-cal Research in Tallgrass Prairie NewYork Oxford University Press

Knapp AK Briggs JM Hartnett DC CollinsSC 1998b Grassland Dynamics Long-Term Ecological Research in Tallgrass Prai-rie New York Oxford University Press

Leach MK Givnish TJ 1996 Ecological de-terminants of species loss in remnant prai-ries Science 273 1555ndash1561

McHugh T 1972 The Time of the BuffaloNew York Knopf

McNaughton SJ 1983 Compensatory plantgrowth as a response to herbivory Oikos40 329ndash336

______ 1984 Grazing lawns Animals inherds plant form and coevolution Ameri-can Naturalist 124 863ndash868

50 BioScience Vol 49 No 1

______ 1993 Grasses grazers science andmanagement Ecological Applications 317ndash21

McNaughton SJ Banyikwa FF McNaughtonMM 1997 Promotion of the cycling ofdiet-enhancing nutrients by African graz-ers Science 278 1798ndash1800

Milchunas DG Lauenroth WK 1993 Quan-titative effects of grazing on vegetationand soils over a global range of environ-ments Ecological Monographs 63 327ndash366

Milchunas DG Sala OE Lauenroth WK1988 A generalized model of the effectsof grazing by large herbivores on grass-land community structure AmericanNaturalist 132 87ndash106

Milchunas DG Varnamkhasti AS LauenrothWK Goetz H 1995 Forage quality inrelation to long-term grazing history cur-rent-year defoliation and water resourceOecologia 101 366ndash374

Nellis MD Briggs JM 1989 The effects ofspatial scale on Konza landscape classifi-cation using textural analysis LandscapeEcology 2 93ndash100

______ 1997 Modelling impact of bison ontallgrass prairie Transactions of the Kan-sas Academy of Science 100 3ndash9

Nellis MD Bathgate J Briggs JM 1992Geographic information systems for mod-eling bison impact on Konza Prairie Kan-sas GISLIS Proceedings 2 618ndash623

Owen-Smith RN 1987 Pleistocene extinc-tions The pivotal role of megaherbivoresPaleobiology 13 351ndash362

Pearson SM Turner MG Wallace LL RommeWH 1995 Winter habitat use by largeungulates following fire in northernYellowstone National Park EcologicalApplications 5 744ndash755

Peden DG Van Dyne GM Rice RW HansenRM 1974 The trophic ecology of Bisonbison L on shortgrass plains Journal ofApplied Ecology 11 489ndash498

Pfeiffer KE Hartnett DC 1995 Bison selec-tivity and grazing responses of littlebluestem in tallgrass prairie Journal ofRange Management 48 26ndash31

Plumb GE Dodd JL 1993 Foraging ecologyof bison and cattle on a mixed prairieImplications for natural area managementEcological Applications 3 631ndash643

Polley HW Collins SL 1984 Relationshipsbetween vegetation and environment inbuffalo wallows American Midland Natu-ralist 112 178ndash186

Polley HW Wallace LL 1986 The relation-ship of plant species heterogeneity to soilvariation in buffalo wallows Southwest-ern Naturalist 31 493ndash501

Power ME Tilman D Estes JA Menge BABond WJ Mills LS Daily G Castilla JCLubchenco J Paine RT 1996 Challengesin the quest for keystones BioScience 46609ndash620

Pyne SJ 1982 Fire in America A CulturalHistory of Wildland and Rural FirePrinceton (NJ) Princeton University Press

Raison RJ Connell MJ Khanna PK 1987Methodology for studying fluxes of soilmineral-N in situ Soil Biology and Bio-chemistry 19 521ndash530

Risser PG Parton WJ 1982 Ecological analy-sis of a tallgrass prairie Nitrogen cycleEcology 63 1342ndash1351

Risser PG Birney CE Blocker HD May SWParton WJ Wiens JA 1981 The TruePrairie Ecosystem USIBP Synthesis Se-ries 16 Stroudsburg (PA) HutchinsonRoss Publishing Company

Ruess RW 1984 Nutrient movement andgrazing Experimental effects of clippingand nitrogen source on nutrient uptake inKyllinga nervosa Oikos 42 183ndash188

Ruess RW McNaughton SJ 1988 Ammoniavolatilization and the effects of large graz-ing mammals on nutrient loss from EastAfrican grasslands Oecologia 77 382ndash386

Samson F Knopf F 1994 Prairie conserva-tion in North America BioScience 44418ndash421

Schwartz CC Ellis JE 1981 Feeding ecologyand niche separation in some native anddomestic ungulates on the shortgrass prai-rie Journal of Applied Ecology 18 343ndash353

Seastedt TR Briggs JM Gibson DJ 1991Controls of nitrogen limitation in tallgrassprairie Oecologia 87 72ndash79

Senft RL Coughenour MB Baily DWRittenhouse RL Sala OE Swift DM 1987Large herbivore foraging and ecologicalhierarchies BioScience 37 789ndash799

Shaw JH Lee M 1997 Relative abundanceof bison elk and pronghorn on the south-ern plains 1806ndash1857 Plains Anthro-pologist 42 163ndash172

Stebbins GL 1981 Coevolution of grassesand herbivores Annals of the MissouriBotanical Garden 68 75ndash86

Steinauer EM 1994 Effects of urine deposi-tion on small-scale patch structure andvegetative patterns in prairie vegetationPhD dissertation University of Okla-homa Norman OK

Steinauer EM Collins SL 1995 Effects ofurine deposition on small-scale patchstructure in prairie vegetation Ecology76 1195ndash1205

Steuter AA 1997 Bison Pages 339ndash347 inPackard S Mutel CF eds The TallgrassRestoration Handbook For Prairies Sa-vannas and Woodlands Washington (DC)Island Press

Steuter AA Steinauer EM Hill GP BowersPA Tieszen LL 1995 Distribution anddiet of bison and pocket gophers in asandhills prairie Ecological Applications5 756ndash766

Turner CL Seastedt TR Dyer MI 1993

Maximization of aboveground grasslandproduction The role of defoliation fre-quency intensity and history EcologicalApplications 3 175ndash186

Turner CL Kneisler JR Knapp AK 1995Comparative gas exchange and nitrogenresponses of the dominant C4 grassAndropogon gerardii and five C3 forbs tofire and topographic position in tallgrassprairie during a wet year InternationalJournal of Plant Science 156 216ndash226

Turner CL Blair JM Schartz RJ Neel JC1997 Soil N availability and plant re-sponse in tallgrass prairie Effects of firetopography and supplemental N Ecology78 1832ndash1843

University of Oklahoma Athletic Department1986 Oklahoma Football Media GuideNorman (OK) University of OklahomaPress

Uno GE 1989 Dynamics of plants in buffalowallows Ephemeral pools in the GreatPlains Pages 431ndash444 in Bock JH LinhartYB eds The Evolutionary Ecology ofPlants Boulder (CO) Westview Press

Van Vuren D Bray MP 1983 Diets of bisonand cattle on a seeded range in southernUtah Journal of Range Management 36499ndash500

Vinton MA Hartnett DC 1992 Effects ofbison grazing on Andropogon gerardiiand Panicum virgatum in burned and un-burned tallgrass prairie Oecologia 90374ndash382

Vinton MA Hartnett DC Finck EJ BriggsJM 1993 The interactive effects of firebison (Bison bison) grazing and plant com-munity composition in tallgrass prairieAmerican Midland Naturalist 129 10ndash18

Wallace LL 1990 Comparative photosyn-thetic responses of big bluestem to clip-ping versus grazing Journal of RangeManagement 43 58ndash61

Wallace LL Turner MG Romme WH OrsquoNeillRV Wu Y 1995 Scale of heterogeneity offorage production and winter foraging byelk and bison Landscape Ecology 10 75ndash83

Wedel WR 1961 Prehistoric Man on theGreat Plains Norman (OK) University ofOklahoma Press

Wedin DA Tilman D 1993 Competitionamong grasses along a nitrogen gradientInitial conditions and mechanisms of com-petition Ecological Monographs 63 199ndash229

______ 1996 Influence of nitrogen loadingand species composition on the carbonbalance of grasslands Science 274 1720ndash1723

Zimov SA Chuprynin VI Oreshko APChapin FS Reynolds JF Chapin MC1995 Steppendashtundra transition A herbi-vore driven biome shift at the end of thePleistocene American Naturalist 146765ndash794