Small-scale experiments (Knops et al. 1997; Tilman1999; Naeem et al. 2000; Kennedy et al. 2002) andmathematical models (Turelli 1981; Post and Pimm 1983;Rummel et al. 1983; Case 1990) have generally supporteda long-held ecological paradigm that, in natural areas,habitats of low plant diversity are more vulnerable toplant invasions than areas of high diversity (Elton 1958).This theory contends that, through the process of compet-itive exclusion (Grime 1973), species-rich areas are“immunized” against invasion by foreign plants throughthe preemption of resources by native species (Tilman1999). One recent study by Kennedy et al. (2002) foundthat “diverse communities will probably require minimalmaintenance and monitoring because they are generallyeffective at excluding undesirable invaders”.It would be comforting to believe that areas with manyplant species are less prone to invasion than those with fewerspecies. Botanical hotspots such as wetlands, riparian zones,Mediterranean environments, subtropical coastal areas, andtallgrass prairies might repel the frequent arrival of plantsfrom other regions or countries. Mechanical, chemical, andbiological control techniques might be limited to heavilyinvaded, species-poor areas, with little danger of compromis-ing unique assemblages of native plant species.Recently, some ecologists have begun to question thisperspective (Stohlgren et al. 1997, 1999; Levine andD’Antonio 1999; Levine 2000). Casual observationshave shown highly invasive plant species, includingtamarisk (Tamarix spp.), Russian olive (Elaeagnus angusti-folia), purple loostrife (Lythrum salicaria), and Chinesetallow (Sapium sebiferum), becoming widely established inspecies-rich riparian zones and wetlands. Until now, how-ever, large, carefully collected data sets from natural land-scapes have been unavailable. These are useful for com-paring local, landscape, regional, and national patterns tothose observed under carefully controlled conditions,such as heavily manipulated, small-scale, experimentalplots protected from disturbance (Knops et al. 1997;Naeem et al. 2000; Kennedy et al. 2002).MethodsWe evaluated two large independent data sets on the dis-tribution of native and non-native plant species. The firstset was gathered from 316 large vegetation-monitoringplots in eight states, which are part of the USDA ForestService’s Forest Health Monitoring Program (Cline et al.1995). The plots are systematically spaced throughoutthe US (one every 63 942 ha), and the numbers vary bystate: Colorado (33), Delaware (39), Michigan (71),Oregon (44), Pennsylvania (81), Virginia (15),Washington (12), and Wyoming (21). Each plot consistsof four 168-m2subplots, with three 1-m2quadrats in eachsubplot. Between 1997 and 2001, all the plots were sam-pled once every summer. Data were collected on the pres-ence and cover of native and non-native species in eachquadrat and species presence in the subplots.The second data set was gathered over the past 20 yearsby the Biota of North America Program (www.bonap.org)at the University of North Carolina, Chapel Hill. Datawere available for 44 of 50 US states, and were based onover 229 000 records of native and non-native plantspecies occurrences by county. The taxonomic accuracyand completeness of this data set have made it the stan-dard reference for many governmental and non-govern-mental agencies.Correlations were evaluated on log-transformed(log10[X + 1]) values of both native and non-native plantspecies at each spatial scale. Significant values in all cases11© The Ecological Society of America www.frontiersinecology.orgRESEARCH COMMUNICATIONS RESEARCH COMMUNICATIONSThe rich get richer: patterns of plantinvasions in the United StatesThomas J Stohlgren,1David T Barnett,2and John T Kartesz3Observations from islands, small-scale experiments, and mathematical models have generally supported theparadigm that habitats of low plant diversity are more vulnerable to plant invasions than areas of high plantdiversity. We summarize two independent data sets to show exactly the opposite pattern at multiple spatialscales. More significant, and alarming, is that hotspots of native plant diversity have been far more heavilyinvaded than areas of low plant diversity in most parts of the United States when considered at larger spatialscales. Our findings suggest that we cannot expect such hotspots to repel invasions, and that the threat ofinvasion is significant and predictably greatest in these areas.Front Ecol Environ 2003; 1(1): 11–141National Institute of Invasive Species Science, US GeologicalSurvey, Fort Collins Science Center, Fort Collins, CO 80526([email protected]); 2Natural Resource Ecology Labora-tory, Colorado State University, Fort Collins, CO 80523; 3Biota ofNorth America Program, University of North Carolina, ChapelHill, NC 27599Plant invasion patterns TJ Stohlgren et al.were based on  = 0.05 using Version 10 (2000) of theSYSTAT statistical software. County-scale correlationswere not evaluated for states without sufficient data.State data not maintained by BONAP were not includedin the state-scale correlations.Results and discussionPlant neighborhood and plot scalesData from 1-m2quadrats in natural landscapes might pro-vide insights on the success of establishment (presence) andgrowth (foliar cover) of native and non-native plants inclose proximity – what might be called a plant’s neighbor-hood. Based on the results of small-scale experiments at sim-ilar scales (Knops et al. 1997; Naeem et al. 2000; Kennedy etal. 2002), we might expect a negative correlation betweennative and non-native species richness in natural areas.However, US Forest Service Forest Health Monitoring datafrom the 1-m2quadrats showed slopes with positive trends(though they may not be significantly different from zero)rather than negative slopes in the correlations of native tonon-native species richness in seven of eight states (Table1). They also showed a highly significant positive relation-ship between native and non-native species richness for the2958 quadrats in all states combined (Table 1). (Vegetationwas not measured in subplots without trees.)Cumulative species lists of native and non-native speciesfrom replicate plots might show the patterns of plant diver-sity and invasion in larger areas more clearly. Some fieldstudies have shown that at least some species-rich vegeta-tion types have been successfully invaded (Stohlgren et al.1999, 2002).