Disturbance, Diversity, And Invasion: Implications For Conservation

326Disturbance,Dive O ,and InvasionLowDisturbance frequency/intensityHobbs&Huem ekeHighFigure 1. The intermediate disturbance hypothesis,which indicates that species d i m i t y within a givenpatch should be highest at intermediate frequenciesor intensities of disturbance (after Connell 1978).function (see Fox 1979; Huston 1979; Sousa 1984).These arguments are often based on the fact that only afew species (ruderals) can persist in the face of frequent, severe disturbance; only a few species ( t h e longest-lived, best competitors, and those able to regeneratewithout disturbance) can persist over the long term inthe absence of disturbance; but many species (includingsome representatives of each of these, plus intermediates) can find some place to survive in a region comprising patches in various stages of recovery, arising atsome intermediate frequency.H o w is "intermediate" defined? It is perhaps easiest torelate the frequency of discrete events to the longevityof major species in the system. Approximately half thelifespan of the dominant species has b e e n used as oneestimation of intermediate disturbance frequency( H o b b s et al. 1984). Definitions of intermediate intensity may have fewer external referents, however; intensity can be evaluated in terms of percentage of individuals killed, or the d e g r e e of structural or r e s o u r c ealteration caused.The above discussion concentrates on within-patchdiversity (alpha diversity), but disturbance is also important for creating or maintaining diversity b e t w e e npatches or at the landscape level (beta diversity). Bycreating p a t c h e s of different ages and successionalstages, disturbance affects structural and habitat diversity as well as overall species diversity. While w e concentrate o n within-patch diversity in this review, therole of disturbance in creating landscape mosaics shouldalso be n o t e d (see Turner 1987).While disturbance is important for maintaining diversity both within communities and at a landscape level, ithas b e c o m e increasingly recognized that disturbancemay also have undesirable effects. Particularly importantis the recognition that disturbance may act to increaseConservation BiologyVolume 6, No. 3, September 1992the likelihood of invasion of a community. For invasionto o c c u r there must be available propagules of an invasive species capable of dispersing into a given plantcommunity, and there then has to b e a suitable microsite for germination and establishment to occur. That is,there has to be a suitable invasion "window" (Johnstone1986). Disturbance usually acts primarily by affectingthe availability of suitable microsites, although s o m eforms of disturbance may affect the availability of invasive propagules. For instance, non-native herbivoresmay bring seed into an area either on their coats or infeces. Here w e will discuss primarily the effect of disturbance on microsite availability.The spatial and temporal distribution of disturbancein a region or an ecosystem gives rise to the disturbancemosaic of an area. Pickett and T h o m p s o n ( 1 9 7 8 )pointed out that the r e c u r r e n c e of disturbances necessitates the preservation of a " m i n i m u m dynamic area,"or an area large enough to contain within it multiplepatches in various stages of disturbance or recoverysuch that internal recolonization can contribute to themaintenance of the overall ecosystem. The dynamics ofp a t c h d i s t u r b a n c e and o f biotic e x c h a n g e s a m o n gpatches, which determine the pattern of recovery, are ofmajor concern in defining the m i n i m u m critical size ofecosystems (see Lewin 1984), the size required to maintain characteristic species composition and system function. With increasing fragmentation of natural areas, it islikely that these minimum areas are n o w to be foundonly within the largest conservation units, and disturbance regimes and biotic exchanges b e t w e e n patchesare liable to be significantly altered in smaller r e m n a n tareas ( H o b b s 1987; Saunders et al. 1991). In particular,invasions are likely to b e c o m e m o r e important. H o wshall m a n a g e r s r e s p o n d to or c o m p e n s a t e for thechanged nature of disturbance? W e approach this question by surveying the major types of disturbance andreviewing their effects on plant species diversity andinvasions. We mostly consider grasslands, but w e alsoinclude illustrative e x a m p l e s f r o m o t h e r vegetationtypes.EmpiricalEvidence1. FireThe central role of fire in maintaining the o p e n nature ofthe vegetation has b e e n acknowledged for m a n y grasslands, particularly in mesic regions. Further, researchhas d o c u m e n t e d that fire can stimulate or maintain highp r i m a r y productivity. In tall-grass prairies of N o r t hAmerica, fire enhances productivity by r e m o v i n g thethick litter layer and altering the microclimate and nutrient content of surface soil (see Knapp & Seastedt1986). Fire also influences species diversity and thecharacteristic structure of these prairie communities.Classical w o r k on fire ecology of prairies (Kucera &

Hobbs & HuennekeKoelling 1964; Abrams et al. 1986) found that annualburning favored tall warm-season grasses and resulted inlow abundance of typical prairie forbs after 5 - 1 0 years.Biennial b u r n i n g r e s u l t e d in t h e h i g h e s t c o m m u nity diversity with m i x e d grasses and forbs. Areas withlong fire-free periods resembled u n b u r n e d areas in theirheavy litter accumulation and decline in grasses.Fires may favor the dominant "matrix" prairie grassesand thus can actually decrease diversity (Collins 1987).A p p a r e n t l y m o s t p r a i r i e fires s t i m u l a t e individualgrasses and do not kill them; few openings are createdfor the establishment of n e w individuals or species. Asw e have noted previously, however, species diversitycomprises two main components: species density or alpha diversity within a patch, and patch diversity or then u m b e r of types of different patches or microhabitats.Glenn-Lewin and ver H o e f ( 1 9 8 8 ) reported that grasslands vary in the degree to which these two contributeto overall diversity. In three grasslands, patch diversityrather than species density was the major contributor tooverall c o m m u n i t y diversity. Fire (and o t h e r disturbances) may create a heterogeneous patch structure,even if within patches it serves to decrease species density.Life history, of course, determines the vulnerabilityand response of plants to fire. In annual grasslands inCalifornia, fire had only t e m p o r a r y effects on botanicalcomposition (forbs increased and grass dominance decreased for a brief time). Here the restructuring of thec o m m u n i t y e a c h a u t u m n w i t h g e r m i n a t i o n quicklyswamps any t e m p o r a r y effect on the seed bank or ongermination conditions (Heady 1972).Suppression of fires in ecosystems dominated by fireadapted species can cause severe disruption of community and ecosystem processes, which may have implications for the conservation of native, fire-tolerant species.For example, Cowling et al. ( 1 9 8 6 ) found that fire suppression has b e e n responsible for the conversion of aSouth African open, grassy veld to a vegetation n o wdominated by undesirable non-native shrubs. They suggested frequent prescribed fires as the best mechanismfor restoring the original o p e n nature of the vegetationand for maintaining populations of the region's endemicgeophytes. Strang ( 1 9 7 3 ) similarly suggested that firewas an expensive but necessary part of reversing theconversion of moist grassland in south-central Africa tobrush. Fire can also b e used m o r e precisely to favor thep e r f o r m a n c e of one species over another. For example,in an attempt to restore prairie on the site of an abandoned agricultural site, fire was used successfully to create openings in a turf of non-native P o a species and toenhance the colonization and expansion of true prairiespecies (Curtis & Partch 1948).As early w o r k in tall-grass prairie confirmed, the overall fire regime rather than any single fire is the criticalDistmt ce, Diversi , and ln sion327factor in determining c o m m u n i t y response. Fires of diffeting intensity or occurring in different seasons arelikely to affect species diversity in a variety of ways b yaltering the potential of individual species to regenerate.Hobbs et al. ( 1 9 8 4 ) provide an example of h o w fireintervals can alter the diversity of species that are ableto regenerate in heathland, and h e n c e affect overallc o m m u n i t y diversity. An intermediate fire frequency resulted in the highest species diversity.Fire has b e e n discussed as a factor that can increasethe likelihood of invasions (Christensen & Burrows1986). Fire acts to r e m o v e m u c h of the plant canopyand usually has a short-term fertilizing effect on the soil;h e n c e both light and nutrient availability can b e increased temporarily. Zedier and Scheid ( 1 9 8 8 ) discussthe invasion of coastal chaparral by Carpobrotus edulisfollowing fire. There is clear evidence, however, thatnot all fires result in increased invasion and that variations in fire regime can affect the extent of invasion.Hobbs and Atldns ( 1 9 9 0 ) have illustrated h o w invasionof Banksia woodlands differs b e t w e e n fires burned inspring versus autumn. In some cases, fire p e r se does notaffect the degree of invasion, or will do so only w h e nc o m b i n e d with some other type of disturbance, such asmechanical disturbance of the soil or nutrient input. Forinstance, Hester and Hobbs ( 1 9 9 2 ) studied b u r n e d andunburned shrubland patches within an area of r e m a n tvegetation in the Western Australian w h e a t b e l t andfound that invasion by non-native annuals was restrictedto the remant edges, even following burning. In adjacentwoodland, the abundance of non-native species actuallydeclined following the fire. Following another fire in thesame area, this time in heathland vegetation, invasionincreased only w h e r e the fire impinged on a roadvergethat had been subject to prior disturbance during roadgrading. This interaction is important w h e n managem e n t of roadside vegetation corridors is considered(see Loney & Hobbs, 1991; Panetta & Hopkins 1991).Because species vary in their response to fires, firemay favor one set of species o v e r another; these relationships can explain the balance b e t w e e n native andnon-native s p e c i e s in s o m e f i r e - i m p a c t e d systems.Where native species are sensitive to fire (because fuelloads w e r e such that fires in the native ecosystem w e r eof low frequency and intensity), fire can enhance theinvasion of non-native fire-tolerant species. W h e n thesefire-tolerant species contribute to increased fuel loadsand inflammability, the d i s t u r b a n c e r e g i m e can b eshifted toward m o r e frequent and intense fires; thesefires further enhance the dominance of non-native o v e rnative species. Just such a cycle has enhanced invasionof w o o d y species in South African Mediterranean systems, and of annual grasses into other Mediterraneanclimate regions (MacDonald et al. 1989). Similarly, invasion of fire-tolerant grasses in dry Hawaiian lowlandsConservation BiologyVolume 6, No. 3, Septonber 1992

328Disturbance,Diversi ,and Invasionhas had severe effects on native species (Hughes et al.1991).2. GrazingGrazing animals are conspicuous and important featuresof many grasslands; it has long been known that someplants are tolerant of grazing while others are not, andthat grazing alters the appearance, productivity, andcomposition of grasslands. Milchunas et al. ( 1 9 8 8 ) havereviewed the effects of grazing by large herbivores ondiffering types of grassland and relate these to the intermediate disturbance hypothesis. They suggest that grazing constitutes a disturbance only where the evolutionary history of grazing is short. This has also b e e ndiscussed by Naveh and Whittaker ( 1 9 8 0 ) and Peet etal. (1983). We suggest, however, that in any situation asignificant change in grazing regime will constitute adisturbance. Thus, imposition of grazing animals (or different herbivores) on a system not previously subject tothat type or level of grazing will constitute a disturbance. So, too, will the removal of grazing from a systemwith a long grazing history. Species diversity will beaffected by the direction of change in grazing regimerelative to the historical regime (Ranwell 1960; White1961; van der Maarel 1971; Milchunas et al. 1990; Dolman & Sutherland 1991). Numerous authors have reported maximum species diversity under intermediatelevels of grazing (Zeevalking & Fresco 1977; Milchunaset al. 1988; Puerto et al. 1990).The most detailed understanding of h o w grazing affects community structure comes from the chalk grasslands of Britain and northern Europe; these infertile sitessupport a diverse mixture of grasses and forbs, withspecies adapted to openings of different kinds and scales(Grubb 1976). These communities, although admittedlyartifacts of human activity (clearing, fires, or grazing),have long been prominent features of the landscape;today they are of major conservation value both fortheir diversity and for particular rare species. Repeatedly it has been demonstrated that grazing is an important factor in the maintenance of chalk grassland diversity; the cessation of grazing leads to dominance of a fewgrasses, and even to incursions by shrubs or otherw o o d y species (Wells 1969). Entire components of theflora may be lost; for example, During and Willems( 1 9 8 6 ) blamed the loss of most lichens and the impoverishment of the bryophte flora in Dutch chalk grasslands on the absence of grazing.Grazing maintains high species diversity in othergrasslands, as well. Grazing management is an importantand successful technique for preserving diversity andconservation value of old grasslands and pastures in England (Hopkins & Wainwright 1989). Sykora et al.( 1 9 9 0 ) found that grassland on embankments in theConservation BiologyVolume 6, No. 3, September1992Hobbs & HuennekeNetherlands was converted to w o o d y scrub in the absence of grazing; under light grazing, a species-poorgrassland resulted from competition from a few competitive grasses. Under more intensive grazing, thosegrasses did not dominate, and a more diverse grasslandwas maintained. Mediterranean-climate grasslands mayrespond similarly to grazing management; in a Californiagrassland on serpentine substrate, cessation of livestockgrazing enhanced the dominance of non-native annualgrasses and led to a rapid decline in abundance of thediverse native annual forb flora (Huenneke et al., unpublished data).One straightforward effect of grazing is the elimination of trees and shrubs invading me,sic grasslands. Without grazing, many North American prairie sites havebeen converted to woodland. Similarly, there are alsodocumented cases of grazing preventing or reversingthe succession of African savanna to woodland. For example, Smart et al. ( 1 9 8 5 ) found that in Uganda theexclusion of elephants was even more important thanfire suppression in encouraging acacia invasion, leadingto the loss of many species including the original grassland dominants. In these regions, a long evolutionaryhistory of grazing has led to the dominance of grasslandplants adapted to and tolerant of grazing pressure. Caldwell's work (for example, Caldwell et al. 1981 ) has docu m e n t e d the many physiological traits that affect aplant's tolerance of grazing losses.In contrast, regions with no recent history of grazingare often dominated by plants that lack these tolerancemechanisms. Extreme examples are presented by oceanic islands with no native mammalian herbivores,where the introduction of livestock or other grazers hasusually been catastrophic in its effect on native vegetationmfor example, the effect of feral goats on islandfloras (Coblentz 1978) and of introduced herbivores inthe Galapagos on native vegetation (Hamann 1975,1979). A less obvious but still major impact has beenmade on regions with few native grazers (at least sincepost-Pleistocene time), such as the intermountain West(Mack & Thompson 1982).In semi-arid grasslands in the American Southwest,species diversity has declined and, in many cases, thephysiognomy of the vegetation has been altered fromperennial grassland to shrub-dominated desert scrub.The chief question of range management and ecology isthe determination of the p r o p e r utilization rate: whatlevel of grazing will maximize productivity and maintainthe grassland's general character? Unfortunately, it is notknown what utilization level maximizes plant speciesdiversity or productivity, or w h e t h e r the same levelmaximizes both. Westoby et al. ( 1 9 8 9 ) outlined the differences in grazing management that would result fromconsidering grazing in an equilibrial, successional context versus a nonequilibrial series of alternate states;

Hobbs & Huetmekeworking with the second mental model requires muchmore active management to "seize opportunities and toevade hazards."Given grazing's impact on community structure, it hasbeen used as a management tool in conservation applications. One example is a grassland restoration project,w h e r e an abandoned, species-poor pasture in Holiandwas being grazed by cattle; seed inputs from cattle feces(together with openings created by grazing) contributed significantly to increasing species diversity (Bulow-Olsen 1980). In another case, sheep w e r e used torestore abandoned fields (Gibson et al. 1987), again byimporting seeds and creating openings for recruitment.Several grasses in the Middle East, wild cereal ancestorsof conservation interest, are negatively affected byheavy grazing but also vulnerable to competition fromtall perennial grasses. Therefore the two sets of speciesalternate on lightly grazed or p r o t e c t e d sites (Noy-Meir1990). Upland British grassland species of conservationvalue vary in their response. Some benefit from removalof grazing, while others are negatively influenced by theresulting increase in grass (Rawes & Welch 1972).Wells ( 1 9 6 9 ) c o m m e n t e d that grazing (or mowing)during the season w h e n the dominant grass species isgrowing most rapidly is usually the most effective wayto maintain diversity in chalk grassland. He stated thatthe cessation of grazing is the major conservation problem in those grasslands, eliminating many forbs andcausing increases of litter and w o o d y species. He added,however, that grazing should be timed to avoid the sensitive phases in the life cycle of species vulnerable tograzing.This raises an important point: Effects of grazing arespecies-specific. That is, two species in the same community may vary in their response to grazing or to aspecific grazing regime. For example, in an English highelevation grassland on limestone, after sheep w e r e excluded from one site, several rare shrubs benefited fromprotection, but one species declined (Elkington 1981 ).The optimal design of grazing management may thus bedifficult. Vinther ( 1 9 8 3 ) found that a mesic meadowwas maintained as o p e n m e a d o w if it was heavilygrazed---because tree seedlings were killed by browsing----or if it was not grazed at all---because seedlingscouldn't establish in the dense herb layer. Intermediategrazing levels allowed w o o d y regeneration and loss ofthe meadow's o p e n character. Unfortunately, thesesame intermediate levels of grazing are those maximizing the richness of herbaceous species in the short term.An alternate means of p r e v e n t i n g w o o d y plant encroachment would then be necessary to allow continued management for maximum species diversity.Grazing's impact presents an interesting contrast tomowing, which is often suggested as an alternative tograzing management. Mowing can reduce the growth ofcompetitively dominant grasses, allowing the persis-Disturbance, Diverst , and Invasion329tence of less competitive species, but it does not createopenings for recruitment of seedlings as grazing does.Sykora et al. ( 1 9 9 0 ) emphasized the different results ofthe two, with grazing creating more microsites for establishment and greater heterogeneity, while providingseed dispersal in animal feces, hooves, and coats. As vanden Bos and Bakker ( 1 9 9 0 ) pointed out, grazers do notuse an entire area evenly but always prefer some spotsto others, so they create greater heterogeneity thandoes mowing. There is also a difference in the form inwhich nutrients are returned or retained in the system(Rizand et al. 1989). Grazing is thus an amalgam of different effects. Clearly, if mowing is to be used by managers in preference to grazing, more sophisticated

may bring seed into an area either on their coats or in feces. Here we will discuss primarily the effect of dis- turbance on microsite availability. The spatial and temporal distribution of disturbance in a region or an ecosystem gives rise to the disturbance mosaic of an area. Pickett and Thompson (1978)