Soil history as a primary control on plant invasion in abandoned agricultural fields


Andrew Kulmatiski, Forest, Range and Wildlife Sciences Department and Ecology Center, Utah State University, Logan, UT 84322–5230, USA (fax + 435 797 3796; e-mail


  • 1Abandoned agricultural (AA) fields are often invaded by exotic plants. This observation has been difficult to explain because agricultural practices change nearly every aspect of an ecosystem. Restoring native plants to AA fields is likely to require a prioritized understanding of the many mechanisms through which agriculture encourages exotic and discourages native plant growth.
  • 2Using 660 experimental plots in three sites in Methow Valley, Washington, USA, we determined the relative role of neighbour removal, propagule addition, plant–soil feedback, soil disturbance and fungal restriction to explain why exotics cover 38% of the ground in AA fields and 3% of the ground in non-agricultural (NA) fields.
  • 3After three growing seasons, neighbour removal improved exotic growth from 3% to 11% cover in NA fields but had no effect in AA fields. Propagule addition did not increase exotic growth above natural recruitment. Differences in soil history, a proxy for plant–soil feedback, explained an increase in exotic growth from 9% in NA fields to 39% in AA fields. Soil disturbance improved exotic growth from 9% to 16% cover in NA fields but had no effect in AA fields. Fungicide reduced exotic growth from 39% to 28% cover in AA soils but had no effect on exotic growth in NA soils. Native plant growth never differed by more than 5% cover among treatments.
  • 4Soil carbon, nitrogen (organic, inorganic and mineralization), phosphorus concentrations and fungal biomass were better associated with plant type (exotic or native) than agricultural history, suggesting that exotics facilitated their own growth by maintaining small beneficial fungal populations and fast nutrient cycling rates.
  • 5 Synthesis and application. Soil history was more important than neighbour removal in determining exotic and native plant distributions. Where exotics rely on plant–soil feedback or legacies of agricultural disturbance, native plant restoration may require soil-based management. In these cases, changing mycorrhizal fungal abundance, increasing soil pathogen loading and slowing nutrient cycling rates may help restore native plants to invaded fields.


Weed growth in rangelands is estimated to cost $2 billion annually in the USA (DiTomaso 2000). Most of these weedy species are exotic and their distributions are often positively correlated with disturbance (Hobbs & Huenneke 1992; Mack et al. 2000). Many mechanisms, including resource competition (Davis, Grime & Thompson 2000; Kennedy et al. 2002; McKane et al. 2002) and plant–soil feedback (Klironomos 2002), partially explain why exotics grow in disturbed sites. Understanding the relative importance of these mechanisms, however, has been difficult because disturbances can simultaneously affect soil traits as well as resource availability and plant species composition (Petryna et al. 2002). It is likely that restoring native plants to disturbed and invaded sites will require an understanding of the relative importance of competition and plant–soil feedback in disturbed sites (Levine et al. 2003).

Agricultural disturbances can alter nearly every aspect of the original ecosystem. Agricultural sites are cleared of native plant competitors, soils are tilled and fertilized, and exotic plants are grown in monoculture. Of these changes, the role of competition among plants has been emphasized as a catalyst for the invasion of exotic weeds (Elton 1958; Symstad 2000). The plant competition hypothesis explains that removal of native plants allows species suited to high resource availability to succeed because resources become abundant (Davis, Grime & Thompson 2000). The hypothesis predicts that early successional and well-dispersed exotic plants should invade abandoned agricultural (AA) fields. However, many exotic invaders are not early successional species and many invaded communities are not in the early stages of succession (Stylinski & Allen 1999; Kulmatiski 2006). Thus, mechanisms in addition to competition are required to explain the success of exotic plants in disturbed areas.

Recent research suggests that soils may not be a passive medium for resource competition but rather a critical interactive component of a plant–soil system (Klironomos 2002; Callaway et al. 2004b). The plant–soil feedback hypothesis proposes that legacies of past plant growth, in the form of specific microbial communities or nutrient availability, can provide a selective advantage to one plant species over another (Bever 1994; Bever, Westover & Antonovics 1997; Miki & Kondoh 2002; Reynolds et al. 2003). In this hypothesis, resource competition can be of secondary importance because soil conditions, such as the presence or absence of a symbiont, determine plant growth.

Plant–soil feedback may be important to exotic success in AA fields for several reasons. First, tillage may remove or alter legacies of past native plant growth (Calderon et al. 2000) in a way that confers an advantage to exotics. Secondly, monocultures of cultivated species could remove or alter legacies of past native plant growth. Further, exotic plants that establish in AA fields may themselves promote positive plant–soil feedback (Klironomos 2002; Miki & Kondoh 2002; Callaway et al. 2004b). Thus, AA soils may accumulate changes in biology (Klironomos 2002), chemistry (Ehrenfeld 2003) and structure (Lutgen & Rillig 2004), resulting from interactions among plants, nutrients, microbes and disturbance (e.g. tillage), that facilitate exotic growth. We refer to these combined effects as soil history.

It is unlikely that exotic plant success in AA fields results only from the removal of native plant competitors or the development of facilitative plant–soil feedback, but little is known about the relative importance of these factors under field conditions. It was our objective to determine the relative importance of plant competition and soil history in determining exotic plant success. For both exotic and native plants, we assessed the strength of competitive interactions by performing neighbour removals. We partitioned this effect into component effects caused by active plant growth, standing dead vegetation and propagule pressure. We assessed the strength of soil history by comparing plant growth on soils that had been used for agriculture and subsequently dominated by exotics to plant growth on soils that had never been used for agriculture and had only been dominated by natives. We partitioned this effect into component effects caused by tillage and fungal growth. We also determined if soil bacterial and fungal biomass as well as nutrient concentrations differed with management history and historical plant growth. We predicted that soil history would be more important than competition in determining exotic and native plant distributions because exotic and native distributions observed on the landscape were delimited by agricultural boundaries (Kulmatiski 2006).


study sites

The research was conducted in the northern shrub-steppe ecotype, Methow Valley, Washington, USA (48°37′N, 107°10′W), on the Newbon soil series (coarse-loamy, mixed mesic typic haploxerolls). Mean annual precipitation between 1971 and 2000 was 380 mm. During this 3-year study (2001–04), 84%, 75% and 74% of 250 mm, 300 mm and 370 mm of annual precipitation occurred outside the growing season (October–March). In this semi-arid shrub–steppe community, total vegetative ground cover is commonly between 40% and 60%, with bare soil or leaf litter exposed on the remaining surface (Kulmatiski 2006).

We used aerial photographs to identify agricultural fields that had been abandoned between 1955 and 1998, following alfalfa and/or wheat cultivation. Agricultural practices in the study sites were of a low input type: fields were not irrigated, pesticides were not used, and inorganic fertilizers were not frequently applied. Abandoned agricultural fields were selected because these fields were dominated by an exotic plant community consisting mostly of diffuse knapweed Centaurea diffusa Lam. and cheatgrass Bromus tectorum L., with bulbous bluegrass Poa bulbosa L., white-top Cardaria draba Desv. and two tumble mustard species Sisymbrium spp. also common (Kulmatiski 2006). In these fields, exotic plants cover 38 ± 3% of the ground (absolute cover; mean ± SD) while native plants cover 4 ± 1% of the ground. Species were considered ‘exotic’ if listed as exotic in Hitchcock & Cronquest (1973). Fields never used for agriculture (NA fields) were dominated by a native plant community consisting mostly of bluebunch wheatgrass Pseudoroegneria spicata Pursh., but arrowleaf balsamroot Balsamorhiza saggitata (Pursh.) Nutt., three lupine species Lupinus spp. and bitterbrush Purshia tridentata Pursh. were also common. In these fields, native plants cover 43 ± 2% of the ground while exotic plants cover 3 ± 1% of the ground.

Three study sites were chosen where an AA field was directly adjacent to a NA field with the same slope, aspect and soil series. These AA and NA field pairs were separated only by apparently arbitrary boundaries so that pre-existing differences in soil biology, chemistry and structure were minimized. More specifically, one field was bisected by an arbitrary property boundary, one field was bisected by a road and in one field a glacial erratic boulder discouraged tilling in one corner of the field. The three sites were separated by 10–20 km and were located between 680 and 880 m in altitude.


In June 2001, plots were established in each site to test the effects of (i) soil history, (ii) neighbour removal, (iii) propagule addition, (iv) soil disturbance and (v) fungicide application on the growth of exotic and native plants. More specifically, 22 1-m2 plots, spaced at 1-m intervals were established in five transects in the AA and NA field at each site (n = 110 plots per field, 220 plots per site and 660 plots total). Plots were used in a blocked split-plot, two-way factorial experimental design with subsamples. The primary treatment at the whole-plot level was soil history (AA or NA). Experimental treatments at the subplot level were defined by combinations of competition (live, dead or none), seeding (exotics, natives or mixed), soil disturbance (hand-tilled or untilled), and fungal suppression [with or without fungicide (Benomyl, DuPont Agricultural Products, Wilmington, Delaware, USA)]. This experimental design, with the use of planned contrasts, allowed measurements from some plots to be used in multiple experiments. For example, the ‘untilled’ plots in the tillage experiment were the same as the ‘without Benomyl’ plots in the fungal suppression experiment.

Treatment effects on native and exotic community cover were assessed in June 2004. Treatment effects were measured as the percentage of intersections at which a plant species was found in a 1-m2 plot, using a regularly spaced 25-point grid (e.g. 5 of 25 intersections = 20% cover). Species observed but not found at a grid intersection were assigned a cover value of 1%.

Soil history

In June 2001, 40 plots in each AA and NA field were sprayed with 30 mL Roundup® herbicide (0·5 kg active ingredient ha−1) (Roundup®, Monsonto, St Louis, MO, USA). Follow-up spot spraying was performed as necessary to eliminate existing plant growth. Herbicide application was expected to be effective against exotics because herbicide was applied early in the growing season to reduce propagule pressure from exotic annuals and because many of the exotics are biennial or perennial. In October 2001, half of the plots were broadcast seeded with an exotic seed mixture and half with a native seed mixture. Seeding was repeated in October 2002. In both seeding years, seeds were planted at a rate thought to saturate germination space (Sheley, Olson & Larson 1997). The exotic seed mix included: 3000 C. diffusa (6 g), 50 B. tectorum (0·2 g), 50 P. bulbosa (0·4 g) and 300 Sisymbrium loeselii L. (0·1 g) seeds. The native seed mix included: 2600 P. spicata (6 g), 150 Lupinus spp. (2·9 g), 600 Stipa comata Trin. and Rupr. (2·6 g), 160 Helianthus annuus L. (0·8 g), 45 Lomatium dissectum Nutt. (0·8 g) and 30 Bromus sagittata Balsemorhiza sagittata (0·2 g) seeds. Native hand-collected seeds were mixed in a 1 : 1 ratio by mass with seeds harvested by Rainier Seed Inc. (Davenport, WA). Germination rates under laboratory conditions were greater than 90% for each species.

The effect of soil history on exotic plant growth was calculated as a plant–soil feedback. For example, the feedback strength for exotic species was calculated as the difference in exotic cover between plots in AA fields and plots in NA fields. The feedback strength for native species was calculated as the difference in native cover between plots in NA fields and plots in AA fields (Bever, Westover & Antonovics 1997; Klironomos 2002).

Neighbour removal

Forty plots in each AA and NA field were treated with 30 mL herbicide as described previously. An additional 20 plots in each AA and NA field were treated with 30 mL water. Standing dead vegetation was left in place in half the herbicide-treated plots and removed from the remaining half. Standing dead vegetation was removed by clipping vegetation at ground level by hand. Half of the plots were seeded with exotics and half were seeded with natives. This resulted in three treatments, each with 10 plots in each AA and NA field: live vegetation (live), standing dead vegetation (dead) and no vegetation (none). The effect of active plant competition from native plants on exotic plants was determined in NA fields by comparing exotic plant cover in 10 ‘live’ plots to exotic plant cover in 10 ‘dead’ plots. The effect of physical interference from standing dead vegetation was determined by comparing exotic plant cover in dead plots to exotic plant cover in 10 none plots. The same approach was used to determine the effect of active plant competition from exotic plants on native plants in AA fields. It was not possible to assess the effects of competition from exotic plants in NA fields or competition from native plants in AA fields because exotic plants were not common enough in NA fields and native plants were not common enough in AA fields.

Propagule addition

Thirty plots in each AA and NA field were treated with herbicide. In each AA and NA field, 10 plots were seeded with exotics, 10 were seeded with natives and 10 were seeded with an additive mix of both exotics and natives. The effect of propagule pressure from natives on exotics was determined by comparing exotic plant cover in 10 exotic-seeded plots to exotic plant cover in 10 mixed-seeded plots in each AA and NA field. The same approach was used to determine the effect of propagule pressure from exotics on natives.

Soil disturbance

Forty plots in each AA and NA field were treated with herbicide, and the resulting standing dead vegetation was removed. Half of these plots were then tilled with a pickaxe to 15 cm. Half of the hand-tilled plots and half of the untilled plots were seeded with exotics and the remaining plots were seeded with natives. The effect of tillage on exotic plants was determined by comparing the exotic cover in 10 hand-tilled exotic-seeded plots to exotic cover in 10 untilled exotic-seeded plots in both AA and NA fields. The same approach was used to determine the effect of tillage on native plants.

Soil fungi

Forty plots in each AA and NA field were treated with herbicide, and the standing dead vegetation was removed. Half of the plots in each AA and NA field were then treated with fungicide (Benomyl® at a rate of 9·4 kg ha−1) in October 2001 and April 2002. Half of the herbicide-treated plots and half of the herbicide-fungicide-treated plots were seeded with exotics. The remaining plots were seeded with natives. The effect of fungicide on exotic plant growth was determined by comparing exotic plant cover in 10 herbicide-treated plots with exotic cover in 10 herbicide- and fungicide-treated plots in each AA and NA field. The same approach was used to determine the effect of fungal activity on native plants.

landscape-level soil traits

We predicted that exotic and native plants would be associated with specific soil chemical [carbon (C), nitrogen (N) and phosphorus (P) availability] and biological (bacterial and fungal biomass) traits across the landscape. Because AA fields used in the experiment had been used for agriculture and dominated by exotic plants since abandonment, whereas NA fields had not, it was important to separate the effects of historical plant growth from the effects of historical agriculture. Occasionally exotic species could be found growing in NA fields, and native species could be found growing in AA fields. Of 50 study sites identified with aerial photos, an inspection on the ground revealed that eight contained all four combinations of disturbance (AA or NA) and plant cover type (exotic or native). We studied soils in these eight sites to determine which soil properties were associated with historical agricultural use and which were associated with exotic or native plant cover.

In May 2003, 10 4-cm diameter cores were collected from the top 15 cm of soil in each of the four agricultural history–plant cover combinations. Half of the samples were placed in thin plastic bags and buried for 1 month to determine net N mineralization rates (n = 5 per agricultural history × plant cover × site combination). From each sample, two 10-g dry-weight equivalent subsamples were taken. One subsample was extracted in 100 mL of 2·0 m KCl for the determination of extractable inorganic N (inline image and inline image) by colorimetric analysis using a Lachat autoanalyser (Lachat Chemicals, Mequon, WI). The second subsample was extracted in 100 mL of 0·5 m K2SO4 for the determination of organic N using persulphate oxidation and colorimetric analysis (Robertson et al. 1999). Extractable organic C in K2SO4 subsamples was determined on a Dohrman TOC analyser (Rosemount Analytical Inc., Santa Clara, CA). Two additional subsamples were used: one to determine the Olsen bicarbonate-extractable P, and one to determine soil organic matter concentration (SOM; Robertson et al. 1999). The concentration of SOM in the fine fraction (< 2 mm) was estimated based on mass loss at combustion at 500 °C for 12 h.

In May 2003, 0–15 cm soil cores from each agricultural history × plant cover × site combination were analysed within 48 h of collection for total and active fungal and bacterial biomass using direct counts of flourescein diacetate-stained cells (Soil Foodweb Inc., Corvallis, OR, USA; Robertson et al. 1999). Mycorrhizal infection rates were determined from the roots of 20 C. diffusa plants that were grown for 6 weeks in a greenhouse in 750 mL Spencer-Lemaire Rootrainers® (Edmonton, Alberta, CA), containing homogenized field soils from AA and NA fields (n = 10 per soil type). Plant roots from each pot were washed, stained with trypan blue and cut into 20 2-cm long segments. Infection was determined as the proportion of hyphae intersecting 100 random points using a crosshair ocular at 200× magnification (McGonigle et al. 1990).

data analysis

Random-effects factors in the two-way factorial, blocked split-plot model were site (block), the site × field interaction (whole plot error), the site × field × treatment interaction (subplot error) and residual error (subsampling variability). Fixed-effects factors were soil history (whole plot factor), tested using site × field interaction as the error term, and treatment (subplot factor), tested using site × field × treatment interaction as the error term. To meet model assumptions of normality and homogeneity of variances, data were transformed using arcsine[square root (percentage cover/100)]. The overall test of treatments was of little interest; instead, we constructed planned contrasts using proc mixed in SAS v.9 for Windows (SAS Institute, Cary, NC).

Differences in soil characteristics among the four agricultural history × plant cover combinations sampled in the soil survey were determined using a blocked two-factor anova, where the eight sites were blocks and agricultural history and plant cover were the factors (each with two levels). Log transformations were used to meet model assumptions of normality and homogeneity of variances, when necessary. Values reported are means ± 1SE) unless otherwise indicated. In all cases, differences were considered significant when P < 0·05.


soil history

Exotic plant cover in herbicide-treated seeded plots was greater in AA fields than in NA fields, and native plant cover in herbicide-treated seeded plots was less in AA fields than in NA fields (Fig. 1a; F1,2 = 160, P = 0·006; F1,2 = 22, P = 0·04, respectively). Therefore exotics performed better in communities dominated by exotic species while natives performed best where they themselves dominated, but the magnitude of the effect was greater for exotics (Fig. 1a, inset; F1,2 = 270, P = 0·0037).

Figure 1.

Treatment effects on plant canopy cover (percentage + SE) three growing seasons after broadcast seeding. (a) Growth responses of native and exotic plants on AA and NA soils, and their resulting plant–soil feedback (inset). (b) Effect of removal of extant native plant neighbours on cover of exotics in NA fields and from extant exotic plant neighbours on native plants in AA fields; live, live vegetation; dead, standing dead vegetation; none, no vegetation. (c) Exotic plant cover on tilled and untilled soils in AA and NA fields. (d) Exotic plant cover on control and fungicide-treated plots in AA and NA fields. Bars represent mean percentage plant cover + SE. *Means are significantly different (P < 0.05). ns, no significant effect observed; AA, abandoned agricultural field, NA; never an agricultural field.

neighbour removal

In NA fields, exotic plant cover in non-herbicide-treated plots (live; 4%) was lower than in herbicide-treated plots where standing dead vegetation was either left (dead; 10%) or removed (none; 11%; Fig. 1b; F1,92 = 2·64, P = 0·0098; F1,92 = 2·28, P = 0·025). Exotic plant cover did not differ between plots with or without standing dead native vegetation in NA fields (F1,92 = 0·63, P = 0·53). In AA fields, native plant cover in non-herbicide-treated plots (live; 5%) and in herbicide-treated plots where dead vegetation was left standing (dead; 5%) was lower than in herbicide-treated plots where standing dead vegetation was removed (none; 8%; Fig. 1b; F1,92 = 2·90, P = 0·0035; F1,92 = 2·23, P = 0·028).

propagule addition

Seeding improved native plant cover from 9 ± 1% in unseeded control plots to 11 ± 1% in native seeded plots (F1,92 = 2·15, P = 0·034). Native plant cover was not reduced by the addition of exotic seed. Native plant cover was 11 ± 1% in native-seeded plots and 9 ± 2% in native + exotic-seeded plots (F1,92 = 1·08, P = 0·28). Seeding did not improve exotic plant cover, which was 27 ± 2% in unseeded control plots and 25 ± 2% in exotic-seeded plots. Exotic plant cover was not reduced by the addition of native seed. Exotic plant cover was 25 ± 1% in exotic-seeded plots and 21 ± 2% in native + exotic-seeded plots (F1,92 = 0·19, P = 0·85).

soil disturbance

In AA fields, exotic plant cover did not differ between hand-tilled and untilled plots (F1,92 = 0·60, P = 0·53). In NA fields, exotic plant cover was greater on hand-tilled than untilled plots (Fig. 1c; F1,92 = 2·67, P = 0·0089). Hand-tillage did not affect native plant cover in either AA or NA fields (F1,92 = 1·56, P = 0·12; F1,92 = 0·93, P = 0·36, respectively).

soil fungicide application

In AA fields, fungicide application reduced exotic plant cover (Fig. 1d; F1,92 = 2·13, P = 0·033), but in NA fields exotic plant cover did not differ between fungicide-treated and non-fungicide-treated plots (F1,92 = 0·33, P = 0·74). Fungicide application did not change native plant cover in AA or NA fields (F1,92 = 0·92, P = 0·36; F1,92 = 0·07, P = 0·95, respectively).

landscape-level soil traits

There was an interaction between the effect of agricultural history and plant cover type on SOM concentrations (F1,144 = 8·78, P = 0·0036). Concentrations of SOM were greater under native plant cover, but only in NA soils (Table 1). There was also an interaction between agricultural history and plant cover on extractable organic C concentrations (F1,155 = 6·17, P = 0·014). Organic C concentrations were greater under exotic cover, but only in NA soils. Concentrations of extractable organic N and inorganic N were lower under exotic cover than under native cover (F1,155 = 32, P < 0·001; F1,155 = 35, P < 0·001, respectively) but did not differ between AA and NA soils (F1,155 = 1·47, P = 0·23; F1,155 = 0·19, P = 0·67, respectively). Concentrations of extractable P were less under exotic than under native cover (F1,155 = 5·51, P < 0·020) and less in AA than NA soils (F1,155 = 9·63, P = 0·002); there was no interaction between treatment effects (F1,155 = 0·66, P = 0·44). Rates of net N mineralization were faster under exotic plant cover (F1,137 = 5·29, P = 0·02) but did not differ between AA and NA soils (F1,137 = 0·17, P = 0·68) and there was no interaction between treatments (F1,137 = 0·28, P = 0·60).

Table 1.  Soil characteristics (0–15 cm layer) associated with exotic and native plant cover on previously agricultural and non-agricultural soils in Methow Valley, Washington, USA
Soil characteristic*AgriculturalNever agricultural
Exotic coverNative coverExotic coverNative cover
  • *

    SOM, soil organic matter; Corg and Norg, K2SO4-extractable organic carbon and nitrogen; Ninorg, KCl-extractable inorganic N; P, olsen bicarbonate-extractable P; N-min, net nitrogen mineralization rate.

  • Values in the same row followed by the same letter were not different at the 0·05 level in a Fisher's LSD test.

SOM (g kg−1)  52 ± 3 b  48 ± 2 b  53 ± 3 b  64 ± 3 a
Corg (mg kg−1) 234 ± 15 b 234 ± 17 b 368 ± 30 a 266 ± 17 b
Norg (mg kg−1)  22 ± 1 b  31 ± 2 a  21 ± 1 b  28 ± 2 a
Ninorg (mg kg−1)0·91 ± 0·05 b1·25 ± 0·06 a0·95 ± 0·04 b1·25 ± 0·05 a
P (mg kg−1)  20 ± 1 b  22 ± 1 b  23 ± 1 b  27 ± 2 a
N-min (mg m−2 day−1) 271 ± 41 a 180 ± 25 b 267 ± 26 a 210 ± 31 b
Microbial biomass
Total bacterial (mg kg−1) 222 ± 15 a 234 ± 17 a 189 ± 14 a 223 ± 16 a
Active bacterial (mg kg−1)  26 ± 4 a  29 ± 4 a  27 ± 6 a  27 ± 4 a
Total fungal (mg kg−1) 109 ± 6 a 107 ± 6 a 118 ± 6 a 112 ± 8 a
Active fungal (mg kg−1) 1·3 ± 0·2 b 1·5 ± 0·2 b 1·9 ± 0·2 b 2·8 ± 0·4 a

There was an interaction between agricultural history and cover type on active fungal biomass (F1,155 = 8·45, P = 0·004), which was greater under native cover on NA soils (Table 1). Other measures of microbial biomass from the field did not differ among treatments. In the greenhouse bioassay, the percentage mycorrhizal infection of C. diffusa roots was greater in AA than NA soils (28 ± 2% vs. 19 ± 2%, F1,18 = 7·98, P = 0·011).


After three growing seasons, in plots cleared of vegetation and planted with exotic propagules, exotics covered 39% of the ground in AA soils and 9% of the ground in NA soils (Fig. 1). The exotic plant growth response to differences in soil history therefore explained most of the difference in exotic plant abundance observed on the landscape (i.e. exotics covered 38% of the ground in AA fields and 3% of the ground in NA fields). In contrast, the exotic plant growth response to neighbour removal was small. Exotics covered 11% of the ground in plots cleared of native vegetation and 3% of the ground in plots where native vegetation was left undisturbed. The presence of standing dead native vegetation and increased native propagule pressure had no effect on exotic cover. Thus the effect of neighbour removal on increased exotic growth was solely the result of active native plant growth (e.g. resource competition). These results demonstrated that short-term competitive interactions, whether between established plants, standing dead vegetation or germinating seeds, were less important to exotic plant growth than differences in soil history.

Native plant cover was less on AA than NA soils. Either AA soils suppressed native growth (e.g. persistent allelochemicals) or NA soils facilitated native growth (e.g. mycorrhizal populations; Vivanco et al. 2004). Soils were not tested for the allelochemicals released by C. diffusa (i.e. 8-hydroxyquinoline) or species-specific mycorrhizae. However, even if allelochemicals or native-specific mycorrhizae were found, their effect on native plants would probably not explain the distribution of plants on the landscape. Native plant treatment responses were never greater than 5% ground cover but natives covered 43% of the ground in NA fields and 4% of the ground in AA fields.

Native plant cover was also reduced in the presence of exotic plants (Fig. 1b). This effect was associated completely with the presence of standing dead exotic vegetation. It is unlikely that standing dead exotic vegetation reduced native cover by providing increased seed rain because experimental additions of exotic seeds failed to reduce native cover and because most vegetation was treated with herbicide before exotic seeds could mature. It is more likely that the small reduction in native cover under dead exotic vegetation resulted from physical effects, such as shading, or chemical effects, such as allelopathy or the addition of C-rich plant matter. While the response of natives or exotics to soil or competition could have explained plant distributions, we found that it was exotic plant responses to soil history that best explained plant distributions in the landscape.

potential soil history mechanisms

Soils in AA and NA fields may differ in several ways. Abandoned agricultural soils had been selected by farmers, tilled, fertilized and dominated by exotic plant species (both agricultural and post-agricultural). Farmers may have selected more fertile soils, tillage may have freed soil resources (Jackson et al. 2003) and fertilization may have increased soil nutrient concentrations. Each of these conditions may have increased soil resource availability and this could be expected to provide a growth advantage for fast-growing (e.g. exotic invasive) species (Davis, Grime & Thompson 2000). In addition to increasing resource availability, tillage may have altered soil biology or structure in a way that benefits exotic plant species (Jackson et al. 2003). Finally, agricultural plant monocultures or post-agricultural exotic species may have exercised plant–soil feedback that altered soils in ways that facilitated the growth of exotic species.

We attempted to control for differences selected by farmers by using only NA and AA fields with similar slope and aspect that were separated by biologically arbitrary boundaries. That native plant growth was greater on NA than AA soils suggested that NA soils were not of lower quality for all species. We tested the effect of tillage by hand-tilling soils. In NA fields, exotic plants grew better in soils that had been hand-tilled (16% absolute cover) than in soils that had not been hand-tilled (9% absolute cover), an effect comparable in size to the effect of removing native plant competitors. This suggests that native plants grow in a soil that inhibits exotic plant growth and that disturbing this soil removes the inhibition. This soil-based inhibition, however, appears to account for only 7% of the 30% difference in exotic absolute cover observed between AA and NA soils. Hand-tillage treatments were also performed in AA fields. Hand-tillage in AA fields did not improve exotic plant growth, showing that, in terms of exotic plant growth, disturbed soils in AA fields were different from soils in NA fields.

Abandoned agricultural fields had been fertilized while under agricultural use. Fertilized soils could be expected to benefit fast-growing, early successional exotic species (Davis, Grime & Thompson 2000; Davis & Pelsor 2001; Ehrenfeld 2003). While we did observe interactions between agricultural history and plant cover type on some measures of soil chemistry, these interactions did not suggest the presence of a fertilization effect in AA relative to NA fields. For example, extractable P concentrations were lower in AA than NA soils. Furthermore, differences in extractable C, N and P availability were at least as large between samples stratified by vegetation history as between samples stratified by agricultural history (Table 1). Our results indicate that agricultural practices in these fields removed more nutrients (e.g. by exporting crops) than were replaced (e.g. with fertilizer).

Tillage, but not fertilization, therefore provided a partial explanation for the success of exotic plants in AA fields. It is likely, then, that changes induced in soils by the growth of exotic plants themselves explained why exotics grew well in AA but not NA soils. Exotic plants in AA fields could be expected to change the seed bank, soil biology or soil chemistry. We attempted to control for the effect of the seed bank by adding propagules from one plant community into the other, but it was impossible to completely replicate the propagule pressure of one field in the other. As a result, differences in propagule pressure could be expected to explain partially exotic plant growth. We do not, however, believe this provides a strong explanation for differences in plant growth between AA and NA fields for two reasons. First, experimental seed additions did not increase exotic growth above background recruitment. This result was anticipated because experimental plots in NA fields were placed as close to AA fields as possible and because exotic growth on the landscape is small in NA fields even where NA fields are located downwind and downhill of AA fields that have been dominated by exotic species for decades (Kulmatiski 2006). Secondly, tillage of exotic seeds can be expected to reduce seed bank effects, yet tilling exotic soils did not reduce exotic growth. This result suggests that the seed bank was not a dominant factor determining exotic plant growth in AA fields.

The vigorous growth of exotics in AA fields could result from positive plant–soil feedback. Positive plant–soil feedback has been shown previously for individual plant species and in potted experiments (Bever, Westover & Antonovics 1997; Klironomos 2002; Reinhart et al. 2003) but have not been shown for whole communities under field conditions. Feedback can be caused by two general mechanisms: facilitation and inhibition. Soils from AA fields could have facilitated exotic plant growth, or soils from NA fields could have inhibited exotic plant growth. Tillage improved exotic growth on NA soils by 7% cover, suggesting that tillage removes a form of soil-based exotic plant growth inhibition. The remainder of the positive plant–soil feedback (Fig. 1a, inset; i.e. 23% cover) may be explained by facilitation between exotic plants and their soils. Soil-based facilitation is often attributed to symbioses between plants and fungi (Klironomos 2002; Callaway et al. 2004a).

We found that fungicide addition in AA fields reduced exotic plant growth by 11% cover. Fungicide addition did not affect exotic plant growth in NA fields (Fig. 1d). This suggests that AA soils maintain a fungal community that improves exotic growth. Because the fungicide Benomyl has a greater affect on mycorrhizal than non-mycorrhizal fungi (Callaway et al. 2004a) and because the growth of mycorrhizal fungal species is tightly associated with the growth of their plant symbionts, it is more likely that the fungal communities in AA soils reflect plant growth than historical agricultural practices, although this could not be determined. A plant–soil fungal relationship was also evident across the landscape, although not as expected. Exotic plants tended to be associated with small active fungal populations (Table 1). Furthermore, mycorrhizal infection rates for the dominant exotic, C. diffusa, were higher on soils from AA fields than from NA fields. It is likely, then, that exotics promote a small but highly beneficial mycorrhizal fungal community: a mycorrhizal fungal community that is not abundant in NA soils. Further descriptions of microbial communities may greatly contribute to explaining invasive success in these sites.

Plant–fungal relationships explained a large proportion of the difference in plant growth observed between AA and NA fields, but did not fully explain this difference. We predicted that plant–soil nutrient feedback would provide an additional explanation for facilitation between exotic plants and their soils (Miki et al. 2002; Ehrenfeld 2003). Exotic plants were associated with higher rates of net N mineralization than native plants. Interestingly, the soils that supported exotics showed higher net N mineralization but smaller inorganic N pools than soils supporting natives (Table 1). Faster N cycling rates and smaller nutrient pool sizes could result because exotic plants are drawing down labile nutrient pools to maintain fast growth rates.

Under field conditions, exotic plant growth responses to differences in soil history were more than three times greater than exotic plant growth responses to neighbour removal: exotics could not grow well on NA soils. Removing native plants from NA soils improved exotic growth, as did tilling NA soils. The magnitude of these effects, however, appeared to be small relative to the effect of facilitation between exotic plants and their soils. This plant–soil relationship appeared to persist across multiple plant generations and to result from plant–soil microbial and plant–soil nutrient relationships.

Before a plant can germinate and grow, or compete with other plants, it must acquire soil-borne nutrients, avoid soil-borne pathogens and herbivores, and possibly develop symbioses with soil-borne organisms. We did not measure specific plant–soil organism interactions, but we did observe that plant growth responses to adjacent soils with different management histories were greater than plant growth responses to neighbour removal. Where exotic plants induce long-term changes in soil traits, the restoration of native plant communities will require either the production of soils that mimic native soils or the selection of native plants that are better able to respond to the growth conditions present in exotic soils. For example, sawdust and activated C additions can manipulate soil biology and chemistry and have been suggested as management tools for native plant restoration (Corbin & D’Antonio 2004; Kulmatiski & Beard 2006). Activated C addition, in particular, has demonstrated the potential to limit the growth of some exotic plant species. Many studies have found links between the soil biota and plant success (Thomashow 1996; Wilson & Hartnett 1997; van der Heijden et al. 1998; Borneman & Hartin 2000; Ronsheim & Anderson 2001; van Os & van Ginkel 2001), suggesting that continued study of specific plant–microbe interactions is likely to result in the development of novel species-specific management approaches for invaded plant communities (van Bruggen & Semenov 2000; Kulmatiski, Beard & Stark 2004; Wolfe & Klironomos 2005).


This research was funded by USDA-NRI (number 35320-13473), the Utah State Agricultural Experimental Station, and the Switzer Foundation. We thank the WA Department of Wildlife, especially J. Mountjoy, Rainier Seeds Inc. and G. P. Kyle for field assistance. We also thank Susan Durham for help with the statistical analyses.