Integrating the study of non-native plant invasions across spatial scales Anı´bal Pauchard 1, * & Katriona Shea 2 1 Facultad de Ciencias Forestales, Universidad de Concepcio ´n, Casilla, 160-C, Concepcio ´n, Chile; 2 Department of Biology and IGDP in Ecology, The Pennsylvania State University, 208 Mueller Laboratory, University Park, PA 16802, USA; *Author for correspondence (e-mail: [email protected]; fax: +56-41-255764) Key words: alien, dispersal, disturbance, emerging properties, multi-scale approach, propagule pressure Abstract Non-native (alien, exotic) plant invasions are affecting ecological processes and threatening biodiversity worldwide. Patterns of plant invasions, and the ecological processes which generate these patterns, vary across spatial scales. Thus, consideration of spatial scale may help to illuminate the mechanisms driving biological invasions, and offer insight into potential management strategies. We review the processes driving movement of non-native plants to new locations, and the patterns and processes at the new locations, as they are variously affected by spatial scale. Dispersal is greatly influenced by scale, with different mechanisms controlling global, regional and local dispersal. Patterns of invasion are rarely doc- umented across multiple spatial scales, but research using multi-scale approaches has generated interesting new insights into the invasion process. The ecological effects of plant invasions are also scale-dependent, ranging from altered local community diversity and homogenization of the global flora, to modified bio- geochemical cycles and disturbance regimes at regional or global scales. Therefore, the study and control of invasions would benefit from documenting invasion processes at multiple scales. Introduction One of the great challenges facing ecology is to understand the interaction of scale and ecological processes, explicitly recognizing the spatiotempo- ral context of natural phenomena (Allen and Starr 1982; Levin 1992; O’Neill and King 1998; Thomp- son et al. 2001; Willis and Whittaker 2001). Scale not only influences the patterns that we observe; ecological processes and mechanisms also differ at different spatial scales. The underlying cause of these scale-dependent relationships is that envi- ronmental heterogeneity changes across scales (Milne 1991), defining which processes dominate as the scale of observation changes (Levin 1992). Non-native plant invasion patterns and pro- cesses show scale-dependent properties (Table 1). Several sequential stages occur in plant invasions: movement to the new location and establishment at the new location are essential for an invasion to occur; spread and impact follow in many cases (Williamson 1999; Richardson et al. 2000). At any of these stages, from seed dispersal to production of new propagules, non-native plants face diverse ecological constraints that are scale- dependent. A comprehensive approach to capturing the dynamic process of non-native invasion across multiple spatial scales may contribute to our understanding of its ecological causes and effects, and help us to identify more efficient and effec- tive control strategies (Mack 2000; Pauchard et al. 2003). However, most studies of inva- sions have focused on only one spatial setting.
15
Embed
Integrating the Study of Non-native Plant Invasions across Spatial Scales
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Integrating the study of non-native plant invasions across spatial scales
Anıbal Pauchard1,* & Katriona Shea21Facultad de Ciencias Forestales, Universidad de Concepcion, Casilla, 160-C, Concepcion, Chile; 2Departmentof Biology and IGDP in Ecology, The Pennsylvania State University, 208 Mueller Laboratory, UniversityPark, PA 16802, USA; *Author for correspondence (e-mail: [email protected]; fax: +56-41-255764)
Non-native (alien, exotic) plant invasions are affecting ecological processes and threatening biodiversityworldwide. Patterns of plant invasions, and the ecological processes which generate these patterns, varyacross spatial scales. Thus, consideration of spatial scale may help to illuminate the mechanisms drivingbiological invasions, and offer insight into potential management strategies. We review the processesdriving movement of non-native plants to new locations, and the patterns and processes at the newlocations, as they are variously affected by spatial scale. Dispersal is greatly influenced by scale, withdifferent mechanisms controlling global, regional and local dispersal. Patterns of invasion are rarely doc-umented across multiple spatial scales, but research using multi-scale approaches has generated interestingnew insights into the invasion process. The ecological effects of plant invasions are also scale-dependent,ranging from altered local community diversity and homogenization of the global flora, to modified bio-geochemical cycles and disturbance regimes at regional or global scales. Therefore, the study and control ofinvasions would benefit from documenting invasion processes at multiple scales.
Introduction
One of the great challenges facing ecology is tounderstand the interaction of scale and ecologicalprocesses, explicitly recognizing the spatiotempo-ral context of natural phenomena (Allen and Starr1982; Levin 1992; O’Neill and King 1998; Thomp-son et al. 2001; Willis and Whittaker 2001). Scalenot only influences the patterns that we observe;ecological processes and mechanisms also differ atdifferent spatial scales. The underlying cause ofthese scale-dependent relationships is that envi-ronmental heterogeneity changes across scales(Milne 1991), defining which processes dominateas the scale of observation changes (Levin 1992).
Non-native plant invasion patterns and pro-cesses show scale-dependent properties (Table 1).
Several sequential stages occur in plant invasions:movement to the new location and establishmentat the new location are essential for an invasionto occur; spread and impact follow in manycases (Williamson 1999; Richardson et al. 2000).At any of these stages, from seed dispersal toproduction of new propagules, non-native plantsface diverse ecological constraints that are scale-dependent.
A comprehensive approach to capturing thedynamic process of non-native invasion acrossmultiple spatial scales may contribute to ourunderstanding of its ecological causes and effects,and help us to identify more efficient and effec-tive control strategies (Mack 2000; Pauchardet al. 2003). However, most studies of inva-sions have focused on only one spatial setting.
Recently, with the increasing interest in land-scape ecology and long-term studies, researchersare trying to better describe and explain the pro-cess and implications of plant invasion across arange of scales. The studies that have addressedthe importance of scale in biological invasions(e.g. DeFerrari and Naiman 1994; Levine andD’Antonio 1999; Stohlgren et al. 1999; Pauchardet al. 2003; Sax and Gaines 2003; Allen andShea, this issue) have generated exciting and ini-tially unexpected results that corroborate theexistence of emergent properties of invasions.
In this paper, we discuss how spatial issuesaffect movement of organisms to a new location,and how they affect patterns and processes at thenew location. In the case of dispersal, we canfocus almost exclusively on the mechanismsinvolved, as that is what is most easily observedand traceable. When we consider the invader inits new environment, however, we usually havemore information on patterns rather than mecha-nisms, and must delve more deeply to infer theunderlying processes. As observed patterns ofinvasion in an environment depend both on dis-persal to that location and on processes specificto the location itself, untangling the underlyingprocesses is a significant challenge.
We then discuss how the type and relativeimportance of invasion impacts change acrossscales. The emergent properties of invasion im-pacts makes it necessary to adopt a multi-scaleapproach when assessing the risk posed by plantinvasions to local and global biodiversity. Inaddition, we discuss how management of theseinvasions may be more efficient and successful ifwe can target key processes at different scalesusing a holistic approach rather than single-scaled and localized initiatives. Finally, wediscuss some suggestions for future research, inorder to take advantage of multi-scale approachesboth for theoretical and practical purposes.
Dispersal mechanisms of non-native plants at
different spatial scales
Dispersal is the process of movement of organ-isms between locations, and plant species exhibita stunning array of adaptations for dispersal(Ridley 1930; van der Pijl 1982). Dispersal is nowrecognized as one of the most important pro-cesses determining invasion success, mainlybecause evidence suggests that limited propagulepressure may be one of the most significant
Table 1. Conceptual framework for understanding the role of scale on plant invasion processes.
Element/scale Global Regional-landscape Local
Dispersal Intercontinental introductions Range expansion, source-sink
flows
Infilling of infected areas, patch
expansion
Pattern:
e.g. relation
to diversity
Decrease in global diversity
(homogenization)
Increases in regional species
richness
Decreases or increases in diversity,
depending on local extinction
Study and
monitoring
Species lists, voucher specimens,
first records. Search for ‘‘expected’’
non-natives
Georeference new invaded areas
and monitor shifts (e.g. counties,
other political boundaries). Deter-
mine infection centers, corridors
and new patches, investigate chan-
ges in biochemical and disturbance
cycles
Determine changes in plant com-
munities, conduct population
studies, including control, disease
and insect interactions
Impacts Homogenization of the global flora Changes in biochemical cycles and
disturbance regimes, losses in agri-
cultural production. Regional im-
pacts concentrated over specific
landscape elements (e.g. reserves,
riparian zones)
Changes in community composi-
tion, competitive relationships and
displacement of natives
Control Limit new introductions, interna-
tional trade oversight. Early
detection and rapid response
Concentrate efforts on rapid
expansion fronts, watch lists.
Adapt human uses to diminish the
expansion of invaders
Direct control over non-native
populations. Control new foci and
local dispersal mechanisms
bottlenecks in the invasion process (D’Antonioet al. 2001).
Non-native plant species move into new envi-ronments with a wide set of scale-dependent dis-persal mechanisms, from global dispersal, oftenhuman-mediated, to local short distance dis-persal, by mostly natural mechanisms such aswind or animals (Nathan and Muller-Landau2000). However, the ultimate result of dispersalat any scale is basically the same; propagulesarrive in a new environment where a new popula-tion may establish. The success of a non-nativeplant is greatly constrained by dispersal at eachof the scales at which this phenomenon occurs,from the local to the global scale. Recent defini-tions suggest that the major difference betweennon-natives and their native counterparts residesin the requirement for at least one human-caused(accidental or intentional) long distance dispersalevent that goes beyond the natural range of theirdistribution (Hodkinson and Thompson 1997;Richardson et al. 2000; McNeely et al. 2001;Pysek et al. 2004).
Global long distance dispersal
Plant species moved around the globe long be-fore humans became an important dispersalagent. Colonization of new islands in the PacificOcean by continental plants caught ecologists’attention as a quantifiable natural process oflong distance dispersal, and a natural experimentto improve the understanding of dispersal. Forexample, the successful dispersal and establish-ment of one plant every 7900 years wouldexplain the accumulation of species in theGalapagos Islands flora; and for the HawaiianIslands one successful event every 20,000–30,000 years would be required (see review inFenner 1985). Under natural conditions, mostlong distance dispersal is mediated by birds(internal and external transport), as well as oceandrift and wind dispersal. Transoceanic and conti-nental global dispersal are highly constrained byphysical and biological barriers, which have con-tributed to an increase in overall global biodiver-sity by allowing the evolution of unique florasand faunas in isolated regions of the world.
Natural and human global non-native plantdispersal processes differ in rate, intensity, mech-
anisms and scale. Human dispersal mechanismsof non-native species at a global scale are muchmore frequent, efficient and effective than naturalmechanisms (Mack et al. 2000). Though humanshave served as dispersal vectors for non-nativeplants since the first human migrations and thebeginning of agriculture and livestock domestica-tion, the rate and distances of long-distancedispersal have increased greatly with the inter-vention of modern humans (Hodkinson andThompson 1997; Mack et al. 2000; Mack andLonsdale 2001; Rossman 2001).
The arrival of Europeans in the Americas cre-ated a new scenario for both deliberate and acci-dental non-native plant invasion. Along withcrop seed, came a number of non-native speciescapable of colonizing this geologically isolatedcontinent. In the last 200–500 years, a significantnumber of species have been introduced, deliber-ately or accidentally, to the Americas. Speciesfrom the Americas have also become naturalizedin Europe, Africa and Asia (Williamson and Fit-ter 1996). However, Eurasia is recognized as themain source of non-native species for all othercontinents. With only 4.4% of the total flora ofthe world, Eurasia contributes with 58.9% ofnaturalized non-native species (Pysek 1998). Thisis partly attributable to human efforts, as Euro-peans transported domesticated plants and theirassociated weeds all over the globe, but also mayindicate a competitive advantage of Europeanspecies in disturbed environments (Sax andBrown 2000).
The continuous propagule flow of vascularplants across countries and continents has no par-allel in evolutionary history and contributes greatlyto the homogenization of the global flora(McKinney and Lockwood 1999; Mack et al.2000). The number of propagules of new speciesthat are introduced (both how often and howmany) drives non-native success; the larger thenumber of propagules, the higher the probability aspecies will establish (Veltman et al. 1996; Mem-mott et al. 1998; Shea and Possingham 2000;D’Antonio et al. 2001; Kolar and Lodge 2001).
Regional long distance dispersal
Once a population of a non-native plant species isestablished in a new continent or region, regional
dispersal mechanisms come into play, allowing thespecies to expand its distribution. Bromus tecto-rum (cheatgrass), a European annual grass, is anaggressive invader of more than 200,000 km2 inthe United States Intermountain West and hasextended to other areas of the United States.B. tectorum was first recorded in the United Statesin 1859 in Pennsylvania (Novak and Mack 2001).Additionally, it had presumably entered throughmultiple ports on the West Coast by 1875. By1930, B. tectorum had already reached its limits inthe western United States. Using genetic markers,Novak and Mack (2001) found that B. tectorumpopulations are the result of various introductionsand consequent complex pattern of terrestrialtransport of seeds, from both the east and the westcoasts across the continent.
The invasion of B. tectorum shows how regionallong distance non-native plant dispersal is drivenby different mechanisms than global dispersal. Inregional dispersal, humans continue to be a maindispersal agent, but the complexity of dispersalpathways increases with landscape heterogeneityand the interaction of natural processes and thenew invader. At the regional scale, propagulemovement tends to follow landscape corridorssuch as roads and rivers, a process that is intensi-fied by human mobility (Hodkinson and Thomp-son 1997; Parendes and Jones 2000; Trombulakand Frissell 2000; Pauchard and Alaback 2004). Inaddition, other natural factors such as wind andwild animals may differentially enhance dispersalsuccess (e.g. Parendes and Jones 2000).
Another well-studied example of regionaldispersal is the invasion of Tamarix species inriparian habitats of the western United States.This species was introduced as an ornamentaland erosion control agent in the early 1900s.Now, it occupies a large portion of riparian cor-ridors, invading more than 370,000 ha in 15states (Zavaleta 2000). Once established, Tamarixcan disperse downstream and, in disturbed floodregimes such as reservoirs and dams, may evendisperse upstream, affecting the integrity of ripar-ian habitats (Lesica and Miles 2001).
Local dispersal mechanisms
For a new population to establish, at least oneindividual must succeed in completing its life
cycle (Richardson et al. 2000). Short distance dis-persal of this individuals’ propagules is the initialstage in the development of a new population. Atthe local scale, dispersal mechanisms are highlyinfluenced by the interaction with the newenvironment. In most cases, humans are not theprimary agent of short distance dispersal, butthey can modify the abiotic and biotic conditionsenhancing invaders dispersal and the environ-mental conditions for survival. The intensity andfrequency of propagules reaching new habitat ishigher primarily because the propagule source iscloser to the invasible habitat (Nathan and Mul-ler-Landau 2000). This is the case for most non-native plant infestations, which after an initialintroduction increase their density and extent byshort distance dispersal of the propagules gener-ated in the nucleus population (Sakai et al.2001).
Non-native plant species, as their native coun-terparts, disperse their seed following a dispersalcurve, usually having the peak in seed dispersalat a short distance from the maternal plant (Fen-ner 1985; Nathan and Muller-Landau 2000).Wind dispersal tends to dominate invasive dis-persal mechanisms because most invasive plantshave evolved in early successional habitats, whereanimals are scarce and a large amount of seed isrequired to rapidly colonize the disturbed envi-ronments (Sax and Brown 2000).
Asexual reproduction is another common andsuccessful mechanism of local dispersal for inva-sive plants (Bazzaz 1996; Kolar and Lodge 2001).A large proportion of noxious invaders haveasexual reproduction, a strategy that increasesthe chance of long-term survival for the newpopulation even under harsh conditions (e.g.Linaria vulgaris in Pauchard et al. 2003).
Each of the scales at which invasion occurs iscrucial for invasion success (Table 1). The failureof a dispersal mechanism at short, long or globalscales may prevent a species from becoming inva-sive in a given environment. However, the threespatial scales of dispersal discussed above clearlydo not have well-defined boundaries. In fact, anygiven dispersal event may comprise a series ofdifferent processes, and movement over differentscales. This idea is well encapsulated by the con-cept of stratified dispersal (Shigesada et al. 1995;Shigesada and Kawasaki 1997). Where smaller
scale processes dominate, an invader will spreadout from the source like a wave. Such move-ment can be modeled using diffusion models (e.g.Skellam 1951; Okubo 1980) and may be mea-sured using relatively straightforward techniques(Greene and Calogeropoulos 2002). However, thesame invader may also undergo occasionallong distance dispersal events, which start newsources. Successful long-distance dispersal eventsare much harder to quantify (Nathan et al.2003). The main infestation spreads like a waveacross space, but later infestations coalesce witheach other and the main wave front, so that thespread is less smooth than the simple spatialspread models would predict.
Dispersal processes across spatial scales:non-native plants in protected areas
Invasions into protected areas illustrate theimportance of multiple scales in the dispersalprocess. For a non-native species to invade intoa protected area it is necessary that the speciesfirst overcome long-distance intercontinental dis-persal barriers. However, this first stage has usu-ally occurred long before the actual invasion ofthe reserve, because most non-native species usu-ally are naturalized in adjacent heavily disturbedareas under human land use types (DeFerrariand Naiman 1994; Pauchard et al. 2003; Pau-chard and Alaback 2004). For example, manyspecies in the northwest of the United Stateswere introduced into agricultural or urban land-scapes in the late 1800s and early 1900s (Toneyet al. 1998). However, these species have takenlonger to establish in parks and reserves due totheir relative isolation at the regional scale.
An interesting example of how multi-scale pro-cesses drive the invasion of a natural area is thecase of Linaria vulgaris (yellow toadflax), aperennial noxious weed, that was first recordedin the Pacific Northwest of the US in 1880(Saner et al. 1995; Rice 2003). After its introduction,the species quickly advanced into inland counties(Figure 1). By 1950, the species had alreadyreached the Rocky Mountains. In the presentday, Linaria has expanded its distribution tomost counties in the Pacific Northwest and itappears that there is no physical or biologicalconstraint to limit its expansion. In the area of
West Yellowstone, near Yellowstone NationalPark, the species invades mostly disturbed soilsand areas adjacent to roads (Pauchard et al.2003). Its aggregated pattern, at the landscapescale, indicates that limited propagule dispersalinfluences its spread into less disturbed or lessaccessible areas. However, at the stand scale,L. vulgaris patches in clearcuts tend to be ran-domly distributed in advanced infestations, whilein recent infestations it shows a cluster distribu-tion (Pauchard et al. 2003). At finer scales, it ispossible to detect that the cover of L. vulgaris in-creases significantly towards the center of thepatch, while the cover of other plants signifi-cantly decreases (Pauchard et al. 2003).
In the case of L. vulgaris there is a clearrelationship between spatial scale and temporalscale of invasion processes. Broad scale pro-cesses such as regional dispersal have occurredover more than a century, while landscape andstand scale changes are visible only after dec-ades. At fine scales, such as the clonal patch,these changes occur over years. Looking at thisbroad range of scales helps our understandingof the invasion and its causal factors muchbetter than would a single scale approach.Impact assessment and control of the speciescan, using this multi-scale approach, be muchimproved.
A similar phenomenon occurs in reserves ofsouthern Chile, where non-native species insidereserves are a sub-sample of those located in thesurrounding matrix (Pauchard and Alaback2004). In rare cases, where a large number oflong distance visitors enter a reserve, they alsomay act as vectors of intercontinental dispersal.This new wave of invasions may bring speciesthat are absent from the disturbed matrix aroundthe reserve.
The invasion of reserves is also constrained byrelatively lower human transportation and distur-bance, diminishing the chances of successfulestablishment. However, the presence of otherdispersal agents such as large herbivores (domes-tic or wild) may increase the rate of successfulintroductions, by acting as major seed vectorsthat move freely across the landscape. Lesica andAhlenslager (1993) found a significant correla-tion, for the period of 1910–1990, between theincrease in the number of visitors and the
y , ( p p , p // )
number of non-native species recorded in GlacierNational Park. Using data from 52 parks fromthe United States and South Africa, Lonsdale(1999) found a similar positive correlation be-tween the number of non-native species and thenumber of visitors, even after correcting for parksize. This relationship does not necessarily implycausality. Increasing visitor numbers are alsorelated to increased development of surroundingareas and increasing overall human activitiesinside the reserve (Liu et al. 2001). On the otherhand, for 77 protected areas in the United States,McKinney (2002) found that visitation was not asignificant variable in explaining non-native speciesrichness; instead, native species richness and histor-ical use were significant predictors. Similarly, inNew Zealand the number of non-native weeds inreserves is related to proximity to propagulesources such as towns, road and railroads, human
use, reserve shape and habitat diversity (Timminsand Williams 1991).
A species established in a reserve does not nec-essarily become a major problem; local scale dis-persal barriers and biotic and abiotic conditionsmay limit the invasion. For example, in Yellow-stone National Park, most non-native speciesthat invade adjacent lands are able to cross thepark boundary and establish on roadsides andother disturbed areas. However, only a fewbecome abundant or invade more pristine environ-ments, at least in the short term (Olliff et al.2001; Pauchard et al. 2003). Long-term invasionsuccess in natural environments is difficult to pre-dict and may depend heavily upon propaguleproduction, genetic adaptations to the new envi-ronment and how the species is able to use nicheopportunities in the new environment (Sakaiet al. 2001; Lee 2002; Shea and Chesson 2002).
Figure 1. Presence of Linaria vulgaris by county (in gray) in the states of Washington, Oregon, Idaho, Montana and Wyoming.
Series shows years 1900, 1950 and 2000. (Reproduced with permission from Rice 2003, http://www.invader.dbs.umt.edu.)
Invaders in their new environment: spatial scale
Once an invader has reached a new environment,its success will depend on how it responds to theniche opportunities available (Shea and Chesson2002). That is, invasion success depends on howthe invader, with its specific characteristics,responds to the resource opportunities, naturalenemy escape opportunities and the physicalenvironment in the new community (Shea andChesson 2002).
All of these components vary in space and,indeed, may vary differently at different spatialscales. For example, resource competitionbetween plants tends to occur at local spatialscales, while apparent competition (competitionfor enemy free space) may occur at larger spatialscales, relevant to the movement of the naturalenemies in question. In such situations, both pro-cesses could contribute to observed patterns ofnon-natives. Furthermore, the resolution atwhich patterns are documented will affect theconclusions that can be drawn. For example,Carduus thistles are present in Pennsylvaniaand all surrounding states, yet studies at finerresolution show distinct within-species aggrega-tion patterns at smaller scales that are completelyobscured by larger scale records (Allen and Sheathis issue; see also Table 1).
In this section, we consider the importance ofspatial scale in determining the outcome of anon-native plant invasion in relation to nicheopportunities in the invaded community.
The relationship between native and exoticdiversity at different spatial scales
The pattern of native and exotic species richnessis currently an exciting and rapidly moving areaof research (Shea and Chesson 2002; Espinosa-Garcıa et al. 2004). Elton (1958) proposed thatcommunities with higher species richness aremore ‘‘stable’’ and less susceptible to invaders.This hypothesis is based on the premise thatmore niches are used and fewer niches are avail-able for invaders in diverse communities (Levineand D’Antonio 1999; Mack et al. 2000), and hasbeen supported by models and small-scale experi-ments (Knops et al. 1999; Stachowicz et al. 1999;Naeem et al. 2000). For example, Tilman (1997),
using experimental manipulations in grasslands,found that communities with higher species rich-ness were more resistant to invasion (1 m2 scale).He hypothesized that more empty niches wereavailable for invasion in the low diversity com-munities. In a different environment, Centaureasolstitialis grown in experimental microcosmplots of 315 cm2 is less likely to invade morefunctionally diverse communities, being morecapable of dominating and suppressing diversityin species-poor communities (Dukes 2001).
In contrast, larger scale studies have shownthat more exotic species occur in more diversecommunities (Lonsdale 1999; Stohlgren et al.1999, 2003; Stadler et al. 2000; Sax 2002; Deut-schewitz et al. 2003; Espinosa-Garcıa et al. 2004).Stohlgren et al. (1999), looking at natural com-munities in a 1 m2 scale, found that CentralGreat Plains prairies confirm the hypothesis thatmore diverse sites are less invasible, while forestand meadows sites in the Rocky Mountains con-tradict this pattern. However, when sampled at alarger scale (1000 m2), all forests and grasslandssites showed a positive correlation between spe-cies richness and susceptibility to invasion. Stohl-gren et al. (1999) concluded that invasibility maybe more related to resource availability (e.g.nitrogen) than to species richness. Similar resultshave been found by Brown and Peet (2003). At alarger regional scale, Stohlgren et al. (2003)found that the most diverse sites in terms of vas-cular plants, in the United States, contain therichest sets of non-native species, contradictingthe common notion that hotspots of diversity areless susceptible to invasion. Deutschewitz et al.(2003) found that both non-native and nativespecies richness increase with temporal andspatial heterogeneity at the regional scale inGermany.
Shea and Chesson (2002) discuss a possibleexplanation for the changes in responses of non-native diversity to native diversity across scales.A negative pattern of non-native richness as afunction of native diversity may be obtainedunder similar extrinsic conditions (e.g. soil, climate).Under these constant conditions, a more diversecommunity would be less susceptible to invasion.However, at broader scales where physicalfactors dominate, the combination of differentdatasets of negative relationships may result in a
positive relationship between non-native and na-tive diversity. At these larger scales, extrinsic fac-tors vary and those factors that favor nativediversity also favor non-native diversity (e.g. lati-tudinal and elevation climate variation).
This hypothesis has recently been supported bya study of competition models in which differencesin resources (niche opportunities) between com-munities generates such a pattern of negative cor-relations at small scales and positive correlationsat larger scales (Byers and Noonburg 2003). In thesame vein, Levine (2000) found that propagulepressure was more important than communitydiversity in the success of invaders in controlledtussocks of 350 cm2, where he manipulated diver-sity and added seed of three invasive plants. Thus,if an area receives more invader propagules perunit area, it may be more invaded just becauseinvaders have had more opportunity to invade.Such patterns may explain higher exotic diversityin popular protected areas, where visitors bring ina higher pressure of invaders (Lonsdale 1999).
An important component of this idea relatesto the role played by spatial and/or temporalvariation in resource (Davis et al. 2000; Shea andChesson 2002) and natural enemy escape oppor-tunities (Shea and Chesson 2002). Davies et al.(in press) studying a grassland data set in Cali-fornia, found the same pattern of a negative rela-tionship between exotic and native diversity atsmall spatial scales with a positive relationship atlarger spatial scales. However, a detailed analysisof the data suggests that, rather than extrinsicconditions per se driving the pattern, it is in factheterogeneity or variation in extrinsic conditionsthat is responsible for the outcome. In this sys-tem at least, resource heterogeneity drives thispattern. Such ideas again come back to the roleof niche differentiation, as in the case of theempty niche hypothesis (Simberloff 1995) or withlimiting similarity (MacArthur and Levins 1967).In communities with high heterogeneity, theremay be more unused niche opportunities for a gi-ven resident species richness. Thus, both averagefactors and the variance in those factors mayplay a role in determining observed covariancepatterns between native and non-native diversity,with one or other aspect dominating under differ-ent circumstances (Davies et al. 2005). This iscertainly an exciting avenue for future research.
Disturbance and scale
Given the importance of both mean extrinsicconditions and variation in extrinsic conditionsat different spatial scales to patterns of invadersin native communities, processes which suddenlyrelease nutrients or remove competitors or natu-ral enemies are an important component of inva-sion ecology. Such events are usually studiedunder the general heading of ‘‘disturbance’’.
Disturbances can be characterized by theirtype, intensity, extent, frequency and duration(Shea et al. 2004). Several of these characteristicshave scale-dependent properties. The most obvi-ous is the extent of a disturbance. Certain distur-bances only modify conditions on a very localscale (e.g. rodent burrows), while others haveeffects on the stand scale (e.g. windthrows, clear-cuts) and yet others over large landscape or re-gions (e.g. fire, volcanic eruptions, hurricanes).Large-scale disturbances modify niche opportuni-ties over a wider area and generally have a great-er effect on ecosystem processes than small-scaledisturbances (Foster et al. 1998). With limiteddispersal, new organisms may not reach portionsof the large disturbed areas for some time, andthe slower recovery rate of the ecosystem alsoopens a wider temporal window for non-nativeplants (e.g. Foster et al. 1998). As species areincreasingly moved around (for example byhumans) long distance dispersal is effectivelyincreased. Thus, recolonization following a firecomes not only from the regional species pool,but also from further afield (Hobbs and Hu-enneke 1992). The extent of disturbances alsoinfluences the array of non-native species that areable to colonize an area. Pioneer species, whichaccount for most non-native invaders, performbest in areas without competition and with highresource availability (Sax and Brown 2000).However, late successional species can invadeareas via small-scale disturbances like forestopenings (e.g. Hedera helix; Reichard 2000).
Despite the non-spatial nature of disturbancefrequency, if there is an interaction between fre-quency and extent, spatial scale issues may arise.Larger scale disturbances tend to be more intenseand less frequent that smaller scale disturbances.Ultimately, disturbance attributes interact witheach other, modifying the biotic and abiotic
conditions for non-native plant invasions. Thespecific spatial and temporal scale of each distur-bance creates a unique set of conditions that mayfavor a particular set of non-native species.
Impacts of invasions: from local to global scales
Biological invasions are considered a majorthreat to biodiversity (Sala et al. 2000), affect-ing ecological processes from the local-scale tothe global-scale (Table 1). The impact of anon-native plant depends on its range, abun-dance and effect, all of which will change withthe scale at which they are measured (Parkeret al. 1999). For plant invasions, changes inspecies richness can be summarized acrossscales as increases in local diversity due to thearrival of new species, and decreases at the glo-bal scale by homogenization of the biota (Saxand Gaines 2003). However, at the local scale,over-dominance of a non-native species mayproduce local extinctions and the subsequentreduction in species richness.
Small-scale impacts are related to changes innative plant population dynamics, communitystructure and diversity (Parker et al. 1999; Macket al. 2000). Changes in the diversity of invadedcommunities have been widely reported. How-ever, in most field ecology studies where manipu-lation and prior monitoring are absent, it isdifficult to isolate whether lower diversity is aninvasion effect or a factor promoting invasion.Allen and Knight (1984) found that cover, den-sity and richness per unit area of native speciesin sagebrush-grassland communities in Wyominghave been reduced by invasive annual non-nativespecies. Similar results were found for Califor-nian serpentine grassland, where invasive annualgrasses displace native forbs in fertilized plots(Huenneke et al. 1990).
At local scales, the demand for resources bynon-native plants limits the resources availablefor the native species, reducing the growth,reproductive outcome and population size ofnative species (Davis 2003). For instance, Centaureamaculosa, an aggressive weed of the northwesternUnited States, reduces Festuca idahoensis seedproduction and root and shoot growth (Callawayet al. 1999).
The ecological effects of invasions may includethe restriction of native populations to smallareas of undisturbed environments, where theirecological function will be badly affected (Mc-Kinney and Lockwood 1999). However, differ-ences in the set of non-native plants that invadea region may have a differentiating effect on theircommunity composition, especially for areas withfew non-native plants (McKinney 2004). Thismay reflect an early stage of the process wheremost invasive plants have not yet arrived at alllocalities, so that non-native species occurrenceis a differentiating factor rather than a homoge-nizing factor. This may reverse rapidly withincreased anthropogenic transport and disturbance.
Large-scale changes in ecosystem processes areinduced by non-native species that becomeinvasive (defined by Pysek et al. 2004). Initialchanges in vegetation diversity and structureproduced by invaders may directly or indirectlyalter ecosystem structure, disturbance regimesand biogeochemical cycles (Mack and D’Antonio1998; Parker et al. 1999; Mack et al. 2000) Forexample, Melaleuca quinquenervia (Australianpaperbark tree) has increased its range in Floridaat a rate of more than 20 ha per day, reachingabout 160,000 ha (see Schmitz et al. 1997 inMack et al. 2000). This non-native tree hasreplaced cypress, sawgrass and other native species,degrading habitat for native animals, using high-er amounts of water and intensifying the fireregime. Other similar cases include Mimosa pigrain Australia, Chromolaena odorata in Asia andAfrica, and Lantana camara in East Africa(Mack et al. 2000). Impacts of invasive specieson disturbance regimes may contribute to largerindirect effects on invaded ecosystems. Positivefeedback has been reported between disturbanceand abundance of invasive species (Mack andD’Antonio 1998). For example, invasion of Afri-can grasses in the Amazon has increased fire fre-quency and intensity, and eventually may causethe conversion of tropical forest into a savanna-like ecosystem (D’Antonio and Vitousek 1992).Non-natives may even impact processes outsidethe area they invade, for example if invaderschange riparian habitat dynamics by increasingrunoff erosion or stabilizing disturbed substrates,thus modifying geomorphological disturbances(D’Antonio et al. 1999).
At the global scale, species area curves havebeen used to model the effects of cross continen-tal dispersal by humans and the homogenizationof the global biota. Vitousek et al. (1997) pro-jected that reduction in dispersal barriers wouldsignificantly reduce world species richness. How-ever, it may be difficult to accurately predict suchchanges using existing data collected at the smal-ler scales of modern continents (Collins et al.2002).
The economic cost of non-native plant inva-sions is hard to estimate as most studies tendto focus on a single species at very localizedscales. However, Pimentel et al. (2000) esti-mated that, for the United States, non-nativeplants are responsible for $26.5 billion in dam-age and control of crops weeds and $6 billionmore in weeds in pastures. Additionally, $100million of costs are estimated for aquatic weedcontrol (Pimentel et al. 2000). Nonetheless, inestimating the cost of plant invasions, precau-tions should be taking to avoid scaling up theresults of local studies without considering theheterogeneity at multiple scales. A focus onareas with high concentrations of invaders maylead to over-estimation of the economic im-pacts of non-native invaders, particularly if na-tive weeds are not explicitly excluded.
The challenge for ecologists and managers isto determine the variation in ecological andeconomic impacts over space and time and topredict, based on that variation, the overalleffects of non-native invasions (Parker et al.1999). Small scale studies may miss spatio-tem-poral variation; therefore, more large-scalestudies are needed. However, large scale studiesalone may be confounded by spatial or tempo-ral gradients. Integrated multi-scale studies mayreduce this uncertainty by providing a morecomplete picture of the invasion. Ultimately,the effects of an invader should be judged bythe degree to which it ‘changes the rules of thegame’ (Vitousek 1990). Defining which speciesare affecting ecological functions, and how theyare doing it, will be critical to prioritizing con-trol efforts (Byers et al. 2002). As societyincreasingly values biodiversity and realizes theimpacts of non-native species, ‘more sophisti-cated and science based information on theecological impacts of invaders should play a
greater and greater role in practical decision-making’ (Parker et al. 1999).
Implications for management of invasives at
different spatial scales
Controlling invasive species requires an under-standing of the mechanisms underlying invasion.A multi-scale approach may improve the effec-tiveness and efficiency of the management ofnon-native plant invaders by, for example, identi-fying and targeting the driving processes thatcontribute to the success of the invader. Singlescale approaches may provide only a limited setof management actions, where a multi-scaleapproach may more closely tailor a managementresponse to the scale that dominates. Thus, amulti-scale approach may help to identify andtarget the limiting processes or the ‘‘Achillesheel’’ for a given invasion.
Most strategies for controlling invaders in nat-ural or semi-natural areas have been developedat the stand scale (e.g. herbicides, hand pulling)to limit the impacts of the invaders on nativecommunities and to diminish the potential prop-agule sources. However, increasing awareness ofthe complexity of invasion processes has resultedin the development of initiatives at larger scalessuch as the landscape (e.g. weed free hay, cleanboots, clean wheels) and global scales (e.g. limi-tations on the imports of the horticulture indus-try, a focus on local rather than non-nativenatural enemies for biological control, voluntarylimitation to new introductions). For example,the implementation and use of a global databaseon invasiveness of non-native species wouldimprove both the risk assessment for new inter-continental introductions and the managementstrategies at the local scale for already existingpopulations (White and Schwarz 1998; Ricciardiet al. 2000). Comprehensive multi-scale schemesfor controlling invasive species are especially use-ful when dealing with new species. In such cases,global scale efforts are as important as local con-trol initiatives in the process of early detectionand confinement of new populations (see Whiteand Schwarz 1998).
Looking at broader scales may help to identifythe factors responsible for a specific invasion
scenario. For example, two national parks mayhave similar patterns of invasion. In this hypo-thetical example, in one of the parks invasion hasoccurred because of intense propagule pressurefrom nearby nurseries (a dispersal problem) whilein the other, large fires have disturbed andopened areas up to the invasion of non-nativespecies better adapted to fire than native species(creation of niche opportunities). The outcome inboth situations is similar. However, looking atthe phenomena beyond the local scale (parkboundaries), it is possible to determine that thetwo invasion processes differ in the causal mech-anism. Effective control will require unique solu-tions for each case.
The goal should be to implement managementstrategies based on the integration of the multiplescales or to identify and target the most appropri-ate scale for a specific problem. An illuminatingexample of such tailored management strategies isthe work by Moore and Possingham (lamentablyunpublished, but discussed in Shea et al. 2002)which asks whether it is better to use limited re-sources to either (i) reduce the number of newpropagules issuing from a major infestation of anew weed or (ii) to stamp out new small infesta-tions of the weed. The optimal management strat-egy depends on the life history of the species inquestion. Species with high, longer distance dis-persal are more likely to be controlled with a focuson the source of propagules than on new out-breaks. Destruction of new, isolated populations ismore likely to be effective for species that spreadmore slowly. Further exploration of such rules ofthumb for the appropriate spatial scale of controlwould be incredibly useful.
The magnitude of non-native plant invasionsin natural ecosystems is not only related to theprocesses occurring in the specific area, but isalso greatly influenced by the broader scale pro-cesses of propagule flow and human-inducedchanges in the adjacent landscape. Therefore,acting only by controlling the populations ofnon-native plants will not have a sustainableeffect in diminishing the threat posed by plantinvasions. On the other hand, holistic approachesthat intend to understand the invasion process ina multi-scale setting may be more efficient andhave a more lasting effect on the protection ofnative diversity.
The next step: understanding invasions
across scales
To enhance our understanding of the mecha-nisms of non-native plant invasions scientistsmust consider a research approach that inte-grates the study of invasion across scales. Whilemany studies have explored small-scale mecha-nisms of invasive plant species, and a few haveexamined large-scale patterns, there is a criticalgap in integrating our understanding of invasionprocesses over a range of scales. This is not asimple matter of scaling up, because most pro-cesses controlling invasion are scale-dependent(Levin 1992). Processes at one scale tend to gen-erate patterns at another; to explain large scalepatterns may require studies at smaller scales, orsmall scale patterns may be constrained by largerscale factors (Levin 1992). Therefore, spatialecology may provide the tools to capture pat-terns and processes over multiple scales and theconceptual framework within which to analyzethe results. Without sufficient knowledge of inva-sion-driven processes at different scales it is diffi-cult to understand, and therefore manage, plantinvasions. Assessing invasion at multiple scalesmay help to better understand the dynamics ofinvasion and its implications for ecosystem pro-cesses.
Ideally, researchers could study a range ofscales, which should be broad enough to generateinsight about underlying mechanisms. This multi-scale approach would require scale-specificresearch methods and techniques (Bullock et al.2002; see, for example, Pauchard et al 2003;Allen and Shea 2006), which can be awkward toimplement. However, methodological complica-tions may be resolved by cooperation amongresearchers with expertise in such a range ofscales. Techniques for working at larger scalesare rapidly being developed (for example, Geo-graphic Information Systems (GIS)), and anincreasing number of databases are available toexplore larger spatial contexts. Herbarium data-bases, county records and floristic lists may pro-vide a useful source of information about largescale patterns of non-native species (e.g. Stohl-gren et al. 2006; Pysek 1998; Toney et al. 1998;McKinney 2002; Pysek et al. 2004; Pauchardet al. 2004).
Probably the most difficult task will be to inte-grate and properly interpret the combined multi-scale results. For that purpose, it is necessary toremember that most commonly we will find mul-tiple patterns and mechanisms at the differentscales and that generalities may be constrained toa specific scale. As Thompson et al. (2001) rec-ommend, we should look for the driving mecha-nisms in an explicit spatial context, because aswe move over multiple scales emergent mecha-nisms arise.
Scientists must also attempt to use a multi-scale approach to understand the dynamics ofthe systems exposed to plant invasions. This mayprovide a more useful ecological context for thestudy of the invader. Questions about the inter-actions between invasion and disturbance orcommunity invasibility will only be realisticallyanswered if there is sufficient understanding ofthe ecological processes affecting that system.For instance, more emphasis must be placed onthe mechanisms by which disturbance increasesnon-native species invasion, and the characteris-tics of disturbance that favor this process. Addi-tionally, interactions among the driving factorsbehind invasion should be addressed explicitly.For example, this has been proven to be particu-larly important for determining the specific roleof propagule pressure, or community aspects incommunity invasibility (e.g. Levine 2000).
Temporal context should also be more fullyincorporated into invasion research. Most studiesonly look at short periods of time (2–3 years),yet it is well known that invasions occur overmuch longer periods of time. Furthermore, chan-ges over time are rarely linear (Kowarik 1995),as shown in the historical reconstruction of inva-sive species detection (e.g. Toney et al. 1998).Monitoring systems need to be implementedusing a scientific design which recognizes thesemulti-scale relationships. For example, small andmedium size permanent plots, the classical ap-proach to temporal variation in vegetation, arenecessary to understand long-term changes inplant communities (Mack 2000). The advantageof georeferenced historical data is that invasionsmay be analyzed with respect to both temporaland spatial dimensions and their interactions.Monitoring invasion at multiple scales may helpto understand the mechanisms driving invasions,
while providing valuable information to optimizeefficiency in the control of invasive species.
Development of multi-scale conceptual frame-works for understanding evolutionary processes,disturbance processes, nutrient transport, andmost recently biodiversity patterns have providedecologists with a rich set of tools with which toaddress increasingly complex questions. Theenormous challenge of managing an increasingnumber of invasive plant populations should alsobenefit from an integrated multi-scale approach.
Acknowledgements
To Paul Alaback who inspired AP to thinkabout the importance of scale in ecological pro-cesses and who made helpful comments on ear-lier versions of this manuscript. To Tom DeLucafor his editing comments. To the President of theRepublic Scholarship of Chile for funding AP’sgraduate studies. KS thanks the Shea lab (espe-cially Olav Skarpaas) for comments and theShea-Mortenson dispersal discussion group atPSU for interesting discussion of these themes.Part of this work was supported by USDA-CSREES (Biology of Weedy and Invasive Plants)NRI Grant #2002-35320-12289, and by NSFGrant #DEB-0315860 to KS, and by Fondecyt1040528 to AP.
References
Allen EB and Knight DH (1984) The effects of introduced
annuals on secondary succession in sagebrush-grassland,
Wyoming. The Southwest Nature 29: 407–421
Allen MR and Shea K. (2006) Spatial segregation of congeneric
invaders in central Pennsylvania, USA. Biological Invasions
18: 525–537
Allen EB and Starr TB (1982) Hierarchy: Perspectives for
Ecological Complexity. University of Chicago Press, Chicago
Bazzaz FA (1996) Plants in Changing Environments: Linking
Physiological, Population, and Community Ecology. Cam-
bridge, UK, 320 pp.
Brown RL and Peet RK (2003) Diversity and invasibility of