1We investigated the role of a native generalist soil pathogen through which a non-native invasive plant species may suppress naturalized/native plant species.
2We found that rhizosphere soils of Chromolaena odorata, one of the world's most destructive tropical invasive weeds, accumulate high concentrations of the generalist soil borne fungi, Fusarium (tentatively identified as F. semitectum), thus creating a negative feedback for native plant species.
3Soils collected beneath Chromolaena in the Western Ghats of India inhibited naturalized/native species and contained over 25 times more spores of the pathogenic fungi Fusarium semitectum than soils collected at the same locations beneath neighbouring native species that were at least 20 m from any Chromolaena plant. Sterilization of these soils eliminated their inhibitory effect. Chromolaena root leachate experimentally added to uninvaded soils increased Fusarium spore density by over an order of magnitude, and increased the inhibitory effect of the soils.
4The positive effect of Chromolaena root leachates on Fusarium spores was attenuated by activated carbon, suggesting a biochemical basis for how the invader stimulated the pathogen.
5Synthesis. Invasive plants have been shown to escape inhibitory soil biota in their native range and to inhibit soil biota in their invaded range, but our results indicate that the impacts of Chromolaena are due to the exacerbation of biotic interactions among native plants and native soil biota, which is to our knowledge a new invasive pathway.
Hypotheses for the transformation of some plant species into overwhelming dominants when introduced by humans to a new part of the world include rapid genetic changes in invasive populations in response to new selection pressures in novel environments, escape from specialist enemies, and the production of phytotoxic, antimicrobial or defence biochemicals to which naive native species are not adapted (reviewed by Inderjit et al. 2005). The latter two hypotheses suggest that shared evolutionary trajectories within natural communities in different regions may mediate coexistence where species have been exposed to each other for a long time. In contrast, perhaps when humans introduce plant species to new regions, they force together species with different coevolutionary trajectories, which may ultimately lead to disruption of communities and competitive exclusion. Evidence for such invasive disruption has been found in studies of novel phytotoxic interactions among plants (Callaway & Aschehoug 2000; Bais et al. 2003; Vivanco et al. 2004; Hallett 2006; Inderjit et al. 2006), the novel biochemical effects of an invader on mutualist fungi (Stinson et al. 2006), and the release of invaders from coevolved predators (Keane & Crawley 2002).
In a similar evolutionary context, soil pathogens play an important role in many exotic plant invasions (Reinhart & Callaway 2006). Beckstead & Parker (2003) found significant inhibitory effects of soil-borne pathogens, particularly nematodes and fungi, on germination and early growth of invasive Ammophila arenaria. In parallel experiments, Beckstead & Parker (2003) observed a significant growth inhibition of A. arenaria in non-sterilized soil from the invaded range (California) compared to sterilized soil. They found partial escape from enemies (because pathogenic nematodes were absent in the invaded range), but suggested the importance of an unknown mechanism besides enemy release. Recently, Eppinga et al. (2006) suggested a new hypothesis; that A. arenaria accumulates high concentrations of local pathogens that inhibit local plant species more than the invader itself, because the local pathogens are not adapted to the invader. The hypothesis predicts that accumulation of local pathogens by A. arenaria in its invaded range could result in the elimination of native plants thus enhancing its dominance and rate of spread. Here, we test this hypothesis in a different system.
We have explored the hypothesis that an invader might accumulate local soil pathogens that harm native plant species. Chromolaena odorata (L.) King and Robinson (siam weed, Asteraceae; hereafter referred as Chromolaena) is a perennial forb, native to Central and South America, that has invaded croplands, natural forests and plantations throughout the Old World tropics and subtropics (Muniappan & Viraktamath 1993). In its invaded range Chromolaena forms dense thickets, 2–3 m high, and can eliminate almost all other vegetation (McFadyen 2003). In the forests of Western Ghats, one of 26 biodiversity ‘hot spots’ in the world (Gurevitch et al. 2006), Chromolaena invasion has suppressed regeneration in natural forests and plantations, reducing the productivity of arable lands, and has been identified as one of the most ecologically destructive invaders in the region (Muniappan & Viraktamath 1993; Inderjit & Drake 2006).
Our preliminary studies suggest that accumulation of local fungal pathogens by Chromolaena might provide an explanation for its invasive success in the Western Ghats of India. We hypothesized that Chromolaena enhances the infection potential of a native generalist soil-borne fungal pathogen, thereby creating an indirect negative feedback to native plant species. We explored the role of soil pathogen accumulation in the invasive success of Chromolaena in the Western Ghats through field measurements of plants and the pathogen, experimental manipulation of Chromolaena root leachates, and experimental manipulation of the soil pathogen.
study area and assay species
We selected an area at Thrissur in the Western Ghats of India, in the state of Kerala, which is heavily infested with Chromolaena odorata (10.30° N; Long. 76.15° E). The study area lies in tropical warm humid monsoonal climate; mean temperature varies from 29 °C (summer) to around 26 °C (winter) and mean annual precipitation of 3800 mm in the north to 1800 mm in the extreme south. The soil in the state can be broadly grouped as lateritic red loam, coastal alluvium and riverine alluvium (Sankaran & Sreenivasan 2001).
We collected soils in two sites heavily infested with Chromolaena, ‘Peechi Dam’ and ‘Pallapilli’, in lateritic red loam soils. At each site, soil was collected from Chromolaena rhizospheres and from the rhizospheres of neighbouring species. Peechi Dam is located near the Kerala Forest Research Institute (KFRI) and Pallapilli (a substation of KFRI) is located c. 40 km from Peechi Dam.
Amaranthus spinosus L. (hereafter referred as Amaranthus) and Bambusa arundinacea (Retz.) Willd. (hereafter referred as Bambusa) were chosen to examine the effects of Chromolaena. Amaranthus is probably a native of South America (Mosyakin & Robertson 1996), but others have thought its origin to be the old world (Munz 1959). We do not know the native origin of Amaranthus in the Western Ghats. It is an economically important plant in this area, used for its medicinal properties (Kunwar & Adhikari 2005). Bambusa is a native of India and distributed widely in the Western Ghats (Kumar et al. 2001).
field analysis of chromolaena effects
We selected a 4-km track which begins in a disturbed forest–grassland mixture and runs into closed forest at Peechi Dam to examine the general effect of Chromolaena on co-occurring species. We chose this particular track because it contained highly variable habitats representing extreme cases of disturbance and resource availability. We considered the part of the transect that occurred outside the forest as a separate site, ‘Roadside’ and the part of the track occurring inside the forest as ‘Peechi Dam’. In January 2006 and November 2006, a total of 144 and 150 quadrats (1 m2) were located in the two sites, respectively. During the first survey (January 2006), 73 quadrats (Roadside, 37; Peechi Dam, 36) were located in vegetation dominated by Chromolaena, and 71 (Roadside, 41; Peechi Dam, 30) in nearby areas where Chromolaena has not yet occupied. The second survey (November 2006) includes 75 quadrats (Roadside, 37; Peechi Dam, 38) in Chromolaena dominated vegetation and 75 in vegetation free of Chromolaena. We measured the average species richness of all plant species present in all quadrats placed in microsites with Chromolaena and microsites without Chromolaena. We tested the difference in plant species richness in plots with Chromolaena vs. plots without Chromolaena using a three-way anova with date, location and treatment as factors (SPSS 15.0, 2006; SPSS, Chicago, Illinois, USA). We followed this analysis with independent t-test of plant species richness in plots with Chromolaena vs. plots without Chromolaena for each site and on each date (SPSS 15.0).
Fifteen kilograms of soil was collected from the rhizospheres of 8–10 Chromolaena plants at each of two sites (Peechi Dam and Pallapilli) and from the rhizospheres of native plants that were at least 20 m from any Chromolaena plant at the same sites. Soils collected from beneath Chromolaena shrubs at the two sites were not pooled, but soil from the rhizospheres of neighbouring plants was pooled and considered as ‘non-Chromolaena soil’. Soils were immediately subjected to slow air-drying, mimicking conditions that would occur during natural drying, and stored in paper bags. The same soil was used for all experiments described below.
In a first experiment, 10 seeds of Amaranthus and six seeds of Bambusa, each from a single population, were placed on the surface of 50 g of unsterilized soil, sterilized soil (triple autoclaving at 121 °C, 103 kPa pressure for 30 min) and soil modified with activated carbon (2%, w/w, Qualigens Fine Chemicals, India). Activated carbon has a high affinity for organic compounds such as potentially allelopathic chemicals, and a weak affinity for inorganic electrolytes such as found in nutrient solution (Cheremisinoff & Ellerbusch 1978), and has been shown to reduce the negative effects of root exudates from other species (Mahall & Callaway 1992; Callaway & Aschehoug 2000). For each species, there were 8–16 replicates for each treatment. Soil was placed in 85 cm3 pots and moistened with 15 mL of double distilled water within three weeks of soil and seed collection.
Amaranthus seeds germinate 3–4 days after seeding and Bambusa seeds germinate 5–7 days after seeding. Growth experiments for Amaranthus were carried out during in February and March 2006 because Amaranthus seeds require an optimum temperature of c. 20 °C for germination. Bambusa growth experiments were carried out during September and October 2006 because its seeds need an optimum temperature of c. 28–30 °C for germination. Shoot height of Amaranthus and Bambusa was measured after 5 and 14 days of seed sowing, respectively.
Our initial studies on the effect of Chromolaena on Fusarium spore counts show much higher accumulation of Fusarium semitectum (hereafter referred as Fusarium) in Chromolaena rhizospheres and non-Chromolaena soil treated with Chromolaena root leachates than in soil from the rhizospheres of neighbouring native plants or soil not treated with Chromolaena root leachates. Based on the size (36–40 × 3–6 µm), shape (curved with a wedge-shaped basal cell; 3–7 septate) of macroconidia and scarcity or absence of microconidia, the fungus was tentatively identified as F. semitectum. However, molecular studies are needed to confirm the identification of Fusarium as F. semitectum. Fusarium semitectum and other Fusarium species are common fungi in soils in the Western Ghats, and are considered to be generalists (Booth 1971; Allen et al. 2005).
We used the same soils described above to compare Fusarium spore counts in soil from the rhizosphere of Chromolaena vs. the rhizospheres of neighbouring species. Five days after planting Amaranthus and Bambusa, 1 g of soil was collected from each of the 8–16 replicate pots and suspended in 100 mL of water. Four replicates were made from each soil sample and 10 subsamples were taken from each 100-mL soil suspension. These replicates were averaged to achieve a single count per pot. Approximately 0.4 mL from each of the 10 subsamples was introduced to one of the wells of the Improved Neubauer Hemocytometer (Rohem, India), with the subsample covering the mirrored surface. The charged haemocytometer was then placed under a microscope (model no. 186012, Nikon, Japan). One entire grid on a haemocytometer with Neubauer rulings could be observed under 40× magnification. Each square has 25 grids with a total surface area of 1 mm2, and the depth of the chamber was 0.1 mm. Fusarium spores in the introduced samples were counted in all 25 grids. An average of 40 observations was taken for each soil sample, which quantifies spore counts in units of 0.1 mm3 or 10−4 cm3 or 10−4 mL. We have presented Fusarium spore counts per gram soil.
We analysed the results of this experiment by first conducting an anova in which site (random factor) and treatment (fixed factor) were tested using the mean heights in each pot as the variable. As for the experiments in which we grew Amaranthus or Bambusa in these soils, we analysed spore counts by first conducting an anova in which site (random factor) and treatment (fixed factor) were tested using the mean spore count in each pot as the variable. We then conducted a single anova in which each of the six soil-treatment combinations was considered a separate factor in the anova. Following this anova, differences among each of the six soil-treatment combinations was tested using a post-anova Tukey HSD test. Note that Tukey tests are not appropriate to compare individual treatments when several factors are simultaneously manipulated. We therefore used Tukey's test for an overall comparison of the treatments.
In a following experiment, soil from the rhizospheres of neighbouring plants was treated with 0, 33, 60, 150 and 300 µL of Chromolaena root leachate per g soil, which was prepared by soaking 15 g of Chromolaena roots in 100 mL of distilled water for 48 h. One half of the soils from the rhizosphere of neighbouring plants were treated by triple autoclaving to kill microbes. Another half of the soils was treated with activated carbon (2% w/w) to adsorb organic molecules in Chromolaena root leachates. Fifty grams of these soils were placed in 85 cm3 pots, 8–16 replicates per species–treatment combination, and were treated with 0, 33, 60, 150 and 300 µL of Chromolaena root leachate per g soil. For Amaranthus, 10 seeds were planted in each pot, grown for 5 days after germination, and measured for shoot height. For Bambusa, six seeds were planted in each pot, grown for 14 days and measured for shoot height. Amaranthus seedlings were very fragile in our experimental conditions, making it difficult to prolong the experiment; we thus recorded ‘final’ shoot height 5 days after germination.
Five days after planting Amaranthus and Bambusa, 1 g of soil was collected from each of the 8–16 replicate pots and suspended in 100 mL of water. Four replicates were made from each soil sample and 10 subsamples were taken from each 100-mL soil suspension. These replicates were averaged to achieve a single count per pot. Fusarium spores were counted as described above.
Because the crucial comparisons in this experiment were among leachate treatments within a soil treatment, we analysed these data by conducting a single anova for each soil treatment using each of the five leachate amounts as factors. We then followed each anova with a Tukey HSD test to determine differences among each of the five leachate treatments. We also conducted three-way anovas (species, treatment and leachate amount as factors) to compare the general effects of Chromolaena leachate between specific soil treatments.
At both sites, we also examined the effect of soil randomly collected from 0, 10, 20, 30 and 40-cm distances away from the outer edges of six Chromolaena canopies on the shoot height of Bambusa. Fifty grams of each soil was placed in 85-cm3 pots (n = 6 for each distance) and six seeds were added to each pot and moistened with 15 mL of distilled water. Shoot height was measured 14 days after starting the experiment and analysed as described above.
At both sites, Pallapilli and Peechi Dam, we also counted the number of Fusarium spores in the soil collected at 0, 10, 20, 30 and 40-cm distances away from the outer edges of Chromolaena canopies, described above for growth bioassays of Bambusa. Five days after collecting soils, 1 g of soil was soaked in 100 mL of distilled water and Fusarium spores were counted as described above.
Root leachate could contribute phenolic compounds either directly through leaching or indirectly through their microbial degradation, which have been shown to exert allelopathic and other ecological effects in many different ecosystems (Inderjit 1996; Inderjit & Weston 2003). We collected soil at 0, 10, 20, 30 and 40-cm away from the outer edges of Chromolaena canopies (see above, n = 6 Chromolaena plants), and analysed these soils for total phenolics. Phenolics were analysed by soaking 5 g of each soil sample in 25 mL of water, filtering the sample through Whatman no. 44 filter paper, and measuring for total phenolic concentration using Folin and Ciocalteu's reagent method (Swain & Hillis 1959).
We statistically tested these data using separate single anovas using each distance as factor. After each anova, differences among all distances were compared using a Tukey HSD test.
To examine the effects of Chromolaena root exudates on phenolics in soils, first we collected another 5 g of soil from open areas c. 20 m away from Chromolaena and applied Chromolaena root leachate at 0, 33, 60, 150 and 300 µL root leachate per g soil. Second, we collected soil from Chromolaena rhizospheres and from the rhizospheres of other species at Peechi Dam and Pallapilli. A composite sample of soil collected from the rhizosphere of other species at Pallapilli and Peechi Dam was made. Total phenolic concentration was determined by Folin and Ciocalteu's reagent method as described above (Swain & Hillis 1959). All analyses were made using six replicates. We analysed these results using one-way anova, with each distance as a factor, followed by Tukey HSD tests (P = 0.01).
Phenolics have been reported to influence the levels of available phosphorus (Appel 1993). For example, Thorpe et al. (2006) found that Centaurea maculosa, an aggressive Eurasian invader in North America, increases phosphorus availability in soil, apparently through the exudation of (+/–)-catechin. The phenolic component of Chromolaena root leachate may therefore influence available phosphate-P levels, which then might have an impact on fungal growth. We therefore analysed available phosphate-P in the soils collected beneath Chromolaena and soils collected where no Chromolaena occurred, and in non-Chromolaena soils to which Chromolaena leachates had been added – soils from the two experiments (experiment 1 and 2) described above. To determine plant-available P, exchangeable P, 5 g of each soil was soaked with 25 mL of 2.5% acetic acid, shaken for 30 min followed by filtration. Exchangeable phosphate-P was estimated using the molybdenum blue method (Allen 1989). We tested for correlation between total phenolics and exchangeable P using the Pearson coefficient.
To examine the effect of Fusarium on native/naturalized plant species, a spore suspension of Fusarium was prepared by transferring Fusarium cultures, grown on PDA plates, to a flask containing 500 mL sterilized distilled water and stirring this solution for 30 min to homogenize the suspension. The amount of Fusarium spores in the suspension was 134 ± 33 × 106 spores per g soil. One half of this Fusarium spore suspension was autoclaved to kill the Fusarium spores. Fifteen mL aliquots of the non-autoclaved suspension at different dilutions were then added to soils to achieve different concentrations of Fusarium spores. Aliquots were added to 50 g of soil in 85-cm3 pots to achieve 134 ± 33, 67 ± 17, 45 ± 11, 22 ± 6 spores × 106 g−1 soil, respectively. Soil was collected at least 20 m from Chromolaena plants. In a second treatment, 15 mL of autoclaved suspension was added to 50 g of the same soil. Soil treated with distilled water served as a second control. For each of the treatment concentrations and the control there were six replicate pots. Six seeds of Bambusa and 10 seeds of Amaranthus were placed in each pot and shoot height was measured after 14 and 6 days, respectively. These results were analysed using an anova for both species combined (with non-autoclaved leachate Bambusa germination was very low), with the mean height in each pot as the variable, and with treatment (live vs. autoclaved spore suspension) and the number of Fusarium spores added as factors.
field analysis of chromolaena effects
In the Western Ghats, we observed substantially lower species richness in plots dominated by Chromolaena (Fig. 1; FChromolaena vs. no Chromolaena = 134.98; d.f. = 1, 286; P < 0.0001).
Soil from the rhizosphere of Chromolaena, collected from Peechi Dam and Pallapilli, suppressed the shoot growth of Amaranthus and Bambusa (Fig. 2a,b). Adding activated carbon increased the growth of Amaranthus in both Chromolaena and non-Chromolaena soil (post-anova Tukey HSD, P < 0.001 for both comparisons), but decreased the growth of Bambusa in both soils (Fig. 2a,b; post-anova Tukey HSD, P < 0.001 for both comparisons). Importantly, autoclaving soils had no effect on the growth of Amaranthus in non-Chromolaena soil relative to non-autoclaved soil (post-anova Tukey HSD, P = 0.983), but greatly improved growth in Chromolaena soil (post-anova Tukey HSD, P < 0.001). Similarly, autoclaving did not significantly improve the growth of Bambusa in non-Chromolaena soil (post-anova Tukey HSD, P = 0.983), but did in Chromolaena soil (post-anova Tukey HSD, P < 0.001).
This basic pattern was mirrored in the abundance of Fusarium spores. Soils collected beneath Chromolaena had over 25 times more spores of the pathogenic fungi Fusarium than soils collected beneath neighbouring native species (Fig. 2c,d; Ftreatment Amaranthus = 342.4; d.f. = 5107; P < 0.001; Ftreatment Bambusa = 247.5; d.f. = 5,71; P < 0.001; post-anova Tukey HSD, P < 0.001 for both comparisons). Activated carbon decreased spore numbers in Chromolaena soils in which Amaranthus had been grown (post-anova Tukey HSD, P < 0.001), but not in non-Chromolaena soils (post-anova Tukey HSD, P = 0.990). In soils used for Bambusa, activated carbon had no effect on spore counts in either Chromolaena (post-anova Tukey HSD, P = 1.00) or non-Chromolaena soils (post-anova Tukey HSD, P = 0.833). Autoclaving soils used for Amaranthus had no effect on spore counts in non-Chromolaena soil (post-anova Tukey HSD, P = 1.00), but greatly reduced spore counts in Chromolaena soil (Fig. 2c; post-anova Tukey HSD, P < 0.001 for both autoclaved vs. untreated and autoclaved vs. activated carbon). Similarly, autoclaving had no effect on spore counts in non-Chromolaena soil used for Bambusa (post-anova Tukey HSD, P = 0.745), but reduced spore counts in Chromolaena soil (Fig. 2d; post-anova Tukey HSD, P < 0.001 for both autoclaved vs. untreated and autoclaved vs. activated carbon).
No inhibition was observed in the control where the Fusarium count was very low. This indicates that the seedlings were not affected by other diseases which could have made them more susceptible to Fusarium.
Adding as little as 33 µL of Chromolaena root leachate per g of soil in which Chromolaena did not occur increased Fusarium spore counts by over 50 times in the Amaranthus experiment and over three times in the Bambusa experiment, and adding even more leachate increased spore counts significantly more (Fig. 3c,d). Soils with increased spore counts reduced the shoot height of Amaranthus and Bambusa significantly more than soils that did not receive root leachate (Fig. 3a,b). Adding activated carbon to the soils exposed to Chromolaena root leachate decreased the spore counts compared to soil without activated carbon, indicating that root leachates from Chromolaena stimulated Fusarium (Fig. 3c,d; three-way anova with activated carbon-unsterilized soil as one factor, and plant species and leachate amount as other factors, Factivated carbon= 285.12; d.f. = 1,79; P < 0.0001; Factivated carbon × leachate amount= 50.86; d.f. = 4,79; P < 0.0001). The activated carbon-stimulated decrease in spore density corresponded with an increase in plant height (Factivated carbon = 8.21; d.f. = 1136; P < 0.001; Fig. 3a,b). Autoclaving soil reduced the effect of leachate additions even more (three-way anova with activated carbon-autoclaving treatments as one factor, and plant species and leachate amount as other factors, Ftreatment= 64.06; d.f. = 1,83; P < 0.0001). Most importantly, autoclaving soil prior to adding Chromolaena leachate strongly increased plant height (Ftreatment = 124.88; d.f. = 1,83; P < 0.0001; Figs 3c,d).
The shoot height of Bambusa was significantly inhibited when grown in soil collected in Chromolaena rhizospheres in comparison to soil 20, 30 and 40 cm away from the Chromolaena rhizosphere (Fig. 4a; anova for Peechi Dam, Fdistance = 112.81; d.f. = 4,50; P < 0.001; anova for Pallapilli, Fdistance = 30.95; d.f. = 4,44; P < 0.001).
Correspondingly, the number of Fusarium spores declined in soil collected at distances of 30 and 40 cm away from Chromolaena rhizospheres (Fig. 4b; anova for Peechi Dam, Fdistance = 82.01; d.f. = 4,50; P < 0.001; anova for Pallapilli, Fdistance = 70.15; d.f. = 4,44; P < 0.001).
We observed a significant decline in total phenolic concentration in the soil collected from a distance of 30 and 40 cm away from the Chromolaena rhizosphere at both sites compared to soil from Chromolaena rhizospheres (Fig. 4c; anova for Peechi Dam, Fdistance = 66.61; d.f. = 4,50; P < 0.001; anova for Pallapilli, Fdistance = 47.82; d.f. = 4,44; P < 0.001).
We also found significantly higher levels of total phenolics in soil from the rhizospheres of Chromolaena at both sites when compared with combined soils from the rhizospheres of other species at both sites (Fig. 5a; one-way anova, F = 288.04; d.f. = 2,17; P < 0.001). Furthermore, adding Chromolaena root leachate to non-Chromolaena soil increased total soil phenolics (Fig. 5b; one-way anova, F = 54.33; d.f. = 4,38; P < 0.001).
Some of the strong stimulatory effect of Chromolaena root leachate on Fusarium may have been indirect. Exchangeable soil phosphate was almost two times higher in soil beneath Chromolaena (0.289 ± 0.06 and 0.299 ± 0.26 mg 100 g−1 soil vs. 0.152 ± 0.02 mg 100 g−1 soil) and experimental addition of Chromolaena root leachate increased exchangeable phosphate by similar amounts depending on the exudate dosage. A significant correlation (Pearson's coefficient = 0.948) was observed between total phenolics and exchangeable P in soil after the addition of Chromolaena root leachate.
We found a significant decrease in the shoot height of Amaranthus after adding the live Fusarium spore suspension (Fig. 6a; anova for both species combined with treatment [live vs. autoclaved spore suspension] and number of Fusarium added as factors; Ftreament = 58.36; d.f. = 1,56; P < 0.001). We combined species because seed germination of Bambusa was strongly inhibited (out of 36 Bambusa seeds only two germinated) after adding the Fusarium spore suspension. However, we measured no inhibition in the shoot growth of Amaranthus or Bambusa when grown in soil treated with autoclaved Fusarium spore suspension (Fig. 6) and autoclaving the suspension significantly increased the growth of the test plants (Ftreatment = 83.45; d.f. = 4,56; P < 0.001).
Our results and many general observations indicate that one of the world's most destructive tropical invasive weeds, Chromolaena odorata, suppresses native plant species in the Western Ghats of India. It is possible that Chromolaena preferentially invades low-diversity plots rather than causing the low diversity, but the observations and measurements of a number of different scientists strongly indicate that the dramatic decrease in diversity associated with Chromolaena is caused by Chromolaena itself (Muniappan & Viraktamath 1993; McFadyen 2003; Inderjit & Drake 2006). We found that this inhibition of other species may be due to the weed's ability to accumulate high concentrations of the generalist, soil-borne fungi Fusarium (tentatively identified as F. semitectum), thus creating a negative feedback for native plant species (see Bever 2002, 2003). These results support an earlier general hypothesis based on theoretical modelling (Eppinga et al. 2006). Direct allelopathic effects of Chromolaena may occur, but appear to be minimal because activated carbon (Fig. 2a,b) did not ameliorate the inhibitory effect of Chromolaena soil on Amaranthus and Bambusa. Furthermore, root leachate added to sterilized non-Chromolaena soils inhibited neither Amaranthus nor Bambusa (Fig. 3a,b). Fusarium species are generalist phytopathogenic fungi (but they appear to have species preferences, see below) that may directly consume plant tissues or damage plants via mycotoxins (Nelson et al. 1993). Fusarium semitectum is widespread in soils from tropical and subtropical regions around the world (Booth 1971) and is commonly found in soils of the Western Ghats (Kumari et al. 2004); but typically at far lower abundance than we found in soils beneath Chromolaena. We did not observe spores of other fungal species in Chromolaena soils, but it is important to note that within-species variation in fungal pathogens can be extensive. Our results indicate that root leachate from Chromolaena alters the balance of soil microbes resulting in an increase in Fusarium. It is not clear how Fusarium inhibits native/naturalized species, and in particular Amaranthus and Bambusa, but direct tissue consumption by the pathogen is likely. Furthermore, the toxin fusapyrone has been isolated from Fusarium semitectum, which possesses antifungal activity against pathogenic and mycotoxigenic filamentous fungi (Altomare et al. 2000; Evidente et al. 2000; Zaccardelli et al. 2006), and this chemical may have phytotoxic effects.
Our results correspond with recent reports on an invasive plant species in China. Ageratina adenophora (also in the Asteraceae) is a highly aggressive invader in parts of Asia and generates feedback effects in its soils that are inhibitory to other native species, but much less so to itself (Niu et al. 2007). It is not clear what components of the soil biota drive these feedbacks for A. adenophora, but substantial shifts in soil bacterial and fungal composition were demonstrated.
We did not observe a proportional reduction in Amaranthus or Bambusa shoot height with decreasing abundance of Fusarium spores in non-Chromolaena soil modified with activated carbon when exposed to Chromolaena root leachate – compared to non-Chromolaena soil (Fig. 3). For example, adding 33 µL of leachate per g soil increased spore counts by 50 times, but decreased plant height by 23%. Increasing leachate ten times increased spore counts c. 50% and decreased plant height c. 50%. Thus it is likely that once a particular level of fungal spores is reached, further increases in concentration have minimal effects on shoot growth, i.e. pathogenic effects may reach an asymptote. However, we found that there was no inhibition of shoot height for either Amaranthus or Bambusa when these plants were grown in autoclaved soil from the rhizosphere of Chromolaena (Fig. 2a,b). The elimination of growth inhibition in autoclaved non-Chromolaena soil, but amended with Chromolaena root leachate (Fig. 3a,b), suggests that soil pathogens are involved. Our results suggest that Chromolaena may inhibit native/naturalized species by culturing a pathogenic Fusarium that is lethal to native plant species. The observed pattern showing an increase in the number of Fusarium spores and total phenolics and suppression of Bambusa shoot height in Chromolaena rhizosphere soil compared to soil 20, 30 and 40 cm away from Chromolaena rhizosphere (Fig. 4) further supports the hypothesis that greater accumulation of Fusarium in Chromolaena rhizospheres creates a negative feedback for native species. The suppression of Amaranthus growth and Bambusa seed germination in soil treated with a live (non-autoclaved) Fusarium suspension further supports our hypothesis that Fusarium can inhibit native/naturalized plant species and thereby be a mechanism by which Chromolaena exerts indirect negative effects on neighbours. Our study, however, does not preclude those other factors, such as root exudate-mediated changes in rhizosphere chemistry, that may be involved in Fusarium effects as well (Inderjit & Weston 2003).
Invaders have been shown previously to escape soil pathogens (Klironomos 2002; Reinhart et al. 2003) and to cultivate soil biota that favour themselves (Callaway et al. 2004), but to our knowledge never have they been shown to cultivate soil biota that harms other plant species. Chromolaena may be physiologically resistant to this fungus, but the invader's life history may play a role in its resistance. In general, Fusarium species kill young seedlings rather than adult plants (Odunfa 1978; Ju et al. 2002), and in India Chromolaena primarily reproduces through below-ground vegetative propagation (Muniappan & Viraktamath 1993), perhaps allowing the invader to escape strong seedling regulation by the pathogen it stimulates. However, future experiments are needed to test whether the establishment and growth of older seedlings is also affected.
We do not know precisely how Chromolaena stimulates Fusarium, but the effect appears to be driven through root exudates. We did not observe Fusarium spores in the root leachate, indicating that leachate is not adding spores to the soil. To further check the presence of Fusarium spores in Chromolaena root leachate, we cultured agar plates after adding 1 mL of distilled water (served as control), Chromolaena root leachate with and without autoclaving, and Chromolaena root leachate after passing through activated carbon. We did not detect any Fusarium spores in either control or in any of the treatment (data not shown). This indirectly indicates that Fusarium spores are present in the Western Ghats soil (perhaps as generalists) and the number of Fusarium spores is significantly increased by Chromolaena root leachate to the level where it can suppress growth of native species.
There may be a biochemical basis for the stimulation of Fusarium spores, which is indirectly supported by a decline in the number of spores when soil was modified with activated carbon (Figs 2 and 3). Chromolaena may exude a particular biochemical that enhances the pathogen, or Chromolaena may release unusually large amounts of exudates that are consumed by many different kinds of soil fungi, but Fusarium benefits disproportionally from these substrate-rich conditions. Fusarium can consume the root exudates of other plants; for example, exudates of cowpea and sorghum enhance the germination of Fusarium conidia (Odunfa 1978) and phenolics from soybean root exudates have been shown to promote Fusarium growth in agricultural soils (Ju et al. 2002).
Our results raise some interesting questions about the biogeographical nature of plant–pathogen interactions. For example, how could Chromolaena stimulate such strong indirect pathogenic effects of Fusarium semitectum in India, but not stimulate the same pathogenic effects by the same Fusarium species in New World habitats? We did not make comparisons between native and invaded ranges of Chromolaena to test evolutionary hypotheses but we can think of two possibilities. First, if the two taxa truly belong to the same species, our results suggest that important regional traits may have evolved among strains of F. semitectum. Fusarium may be stimulated by Chromolaena in both its native and non-native ranges, but in the invaded range the local pathogen may have the strongest effects on species with which it has coexisted for a longer time, as reported by Niemi et al. (2006). Second, the effect of Chromolaena on Fusarium may be the same in both native and invaded ranges but the very high infestation of Chromolaena in the invaded range creates much larger stimulatory effects. While Chromolaena is identified as a serious pest in its invaded range, it is not categorized as a pest in its native range (Raimundo et al. 2007).
Our results are unusual in that they suggest the impacts of Chromolaena on other species in its invaded ranges are due to the exacerbation of biotic interactions that are naturally part of the native ecosystem rather than the introduction of novel interactions. Invasive plants may succeed because they escape inhibitory biotic interactions (DeWalt et al. 2004), possess novel biochemicals that are inhibitory to naive species (Callaway & Aschehoug 2000; Callaway & Ridenour 2004), and evolve to be better competitors (Siemann & Rogers 2003). Our results provide experimental evidence of yet another invasive pathway, the exceptionally powerful stimulation of a native generalist soil pathogen.
We thank Kurt Reinhart for valuable comments on the manuscript and K.G. Mukerji and Surinder Kaur for their assistance in identifying fungal species. We also thank Dr K.V. Sankaran for help in field study at KFRI. This research on Chromolaena odorata was funded by the Ministry of Environment and Forests (MoEF). R.M.C. was supported by The National Science Foundation, the USDA, The Aldo Leopold Wilderness Center, The US Forest Service Fire Laboratory, The Office of Sponsored Research at The University of Montana, US Department of Defense-SERDP, and the Civilian Research and Development Foundation. We sincerely thank Marcel van der Heijden, the Associate Editor and three anonymous referees for their constructive suggestions.