*Author to whom correspondence should be addresed: Roy H. A. van Grunsven. Tel.: +31 317 483659. Fax: +31 317 419000. E-mail: email@example.com.
1As a result of global warming, species may spread into previously cool regions. Species that disperse faster than their natural enemies may become released from top-down control. We investigated whether plants originating from southern Europe and recently established in north-western Europe experience less soil pathogen effects than native species.
2We selected three plant species originating from southern Europe that have immigrated into the Netherlands and three similar native Dutch species. All six plant species were grown in sterilized soils with a soil inoculum collected from the rhizospheres of field populations. As a control we grew a series of all six plant species with a sterilized rhizosphere inoculum.
3We harvested the plants, added the conditioned soil to sterilized soil and grew a second generation of all six plant species in order to test for each plant pair feedback effects from the conditioned soil communities to conspecifics and heterospecifics.
4The effect of the soil community is dependent on plant species, and is dependent on soil fertility in only one of the three pairs.
5Soil conditioning caused less biomass reduction to exotic plant species than to native species, suggesting that exotic immigrants are less exposed to soil pathogens than similar native plant species.
6Our results suggest that plant species that expand their range as a result of climate change may become released from soil pathogenic activity. Whether the exotics are released from soil pathogens, or whether they experience enhanced benefit from mutualistic symbionts remains to be studied. We conclude that range expansion may result in enemy release patterns that are similar to artificially introduced invasive exotic plant species.
7The escape from enemies through range shifts changes key biotic interactions and complicates predictions of future distribution and dominance.
Global climate change affects the distribution and phenology of many species (Parmesan et al. 1999; Tamis et al. 2005; Hickling et al. 2006). Thus far, climate change studies have mainly described and analysed species distribution patterns (Hickling et al. 2006; Parmesan & Yohe 2003). Predictions of future ranges of species have been based predominantly on environmental requirements that limit a species range, so-called ‘climate envelopes’ (Bakkenes et al. 2002; Thomas et al. 2004). The envelope approach assumes the environmental conditions of the future range will be the same as in the current range. However, environmental requirements of species are also influenced by interactions with other trophic levels, such as natural enemies, the enemies of their enemies, and mutualistic symbionts (Van der Putten et al. 2004). While studies on phenological shifts have focused on biotic interactions (Visser & Both 2005), these interactions have been neglected in studies that analysed altered distribution patterns. In order to develop a more complete understanding of species distribution patterns and abundances in a warmer world, climate envelopes need to include biotic interactions within and between trophic layers. New introductions may affect biotic interactions (Liu & Stiling 2006), but whether this is also true for gradual shifts in distribution as a result of climate change is currently unknown.
As some species disperse faster than others (Hickling et al. 2006), range shifts due to climate change may lead to a disruption of trophic interactions in the original range, and to an assemblage of novel interactions in the new range. During trophic re-assemblage, species may become released from natural enemies and they may also be exposed to novel enemies. This process is similar to enemy release and biotic resistance, which is known from exotic invasive species that have artificially crossed geographical barriers (Elton 1958; Maron & Vila 2001; Keane & Crawley 2002). However, in contrast to exotic plants that have been released from many of their enemies by crossing geographical barriers (Hinz & Schwarzlaender 2004; Wolfe 2002; Reinhart et al. 2003; Van der Putten et al. 2005), range expansion might result in partial or temporary enemy release, depending on the response of the natural enemies to the factor that drives host range expansion. To our knowledge, enemy release has not yet been studied for plant species that immigrate into previously cold (poleward) ranges.
Soil pathogens and root herbivores generally are less capable of actively dispersing than above-ground enemies (Van der Putten et al. 2001). Therefore, we assumed that plant range shifts could result in release from adverse soil organisms. Soil organisms are known to influence succession, plant abundance, plant competition and community composition (Van der Putten et al. 1993; Bever et al. 1997; Van der Putten & Peters 1997; Klironomos 2002; Kardol et al. 2006). They can influence plant performance negatively, through herbivorous or pathogenic activities, or they can enhance plant performance through mutualistic interactions or improved plant nutrition by decomposition of soil organic matter (Wardle et al. 2002). Plants, in turn, can also influence soil organisms, e.g. via input of organic matter, or rhizodeposition (Ehrenfeld et al. 2005). The net effects of all positive and negative interactions between plants and soil organisms is called plant–soil feedback (Bever et al. 1997). Separating positive and negative effects in plant–soil feedback is difficult, because individual interactions can cause chain effects, and some interacting biota can have positive or negative effects depending on genotype and environmental conditions (Borowicz 2001; Klironomos 2002; Wolfe et al. 2006). The effects of isolated biota can be different from their effects when occurring in a rhizosphere community. This advocates a focus on the net effect of the soil community rather than attempting to separate individual positive and negative components. Due to enemy release, plant–soil feedback effects could be more negative for local plants than for species that colonize a formerly cold range (so-called thermophilic neophytes; Tamis et al. 2005). When these exotic species have a less negative net feedback from the local soil community than most native species this would give them an advantage in interspecific interactions, ultimately resulting in greater dominance than in their original habitat (Klironomos 2002).
One approach to the study of enemy release of exotic plant species is to compare the response of a species in its native and its new range (Wolfe 2002; Hinz & Schwarzlaender 2004; Hierro et al. 2005). However, this approach is less effective when the precise origin of the exotic plants is unknown (Reinhart et al. 2005). An alternative is to make a phylogenetic, or ecological, comparison between exotic species and similar related native species (Agrawal et al. 2005; Carpenter & Cappuccino 2005). We took the latter approach and compared plant biomass production of three exotic and three related or ecologically similar native species. All plants were grown in sterilized soil to which an inoculum from the rhizosphere of natives and exotics in the invaded range was added. This resulted in a conditioned soil containing a soil community specific to that plant species, which was used to assess plant–soil feedback in the second part of the experiment. The confounding effects of nutrient release by sterilization (Troelstra et al. 2001; McNamara et al. 2003) were counteracted by using a relatively small inoculum in a sterilized bulk soil and addition of a nutrient treatment.
In the plant–soil feedback experiment, we tested the hypothesis that exotic plants would have less negative plant–soil feedback than the native species (Callaway et al. 2004; Reinhart et al. 2003). In the cross-inoculation experiment, we grew native species on soil inoculated with soil conditioned by the exotics, and vice versa. Here, we tested the hypothesis that the exotic plants may accumulate local pathogens, so that they could have indirect negative effects on the performance of related native species (Eppinga et al. 2006). We discuss our results in relation to the effects of climate change on plant release from soil-borne enemies and on possible indirect effects of the exotic plants on the performance of native plant species through soil feedback.
plant species selection
Three exotic plant species (Heracleum mantegazzianum, Tragopogon dubius and Eragrostis pilosa) were selected and compared with three native species (Heracleum sphondylium, Tragopogon pratensis and Poa annua, respectively) that naturally co-occur with the exotic species in the Netherlands. Two species pairs were of the same genus and one species pair involved species with similar ecology. Invading plant species may not necessarily be a random sample from the total species pool. In our selection of native species from the Dutch flora, we choose species that themselves have become invaders in other regions of the world (Clements et al. 1999; Cody et al. 2000; Ryan et al. 2003; Page et al. 2006).
Heracleum mantegazzianum Somm. & Lev. (Apiaceae) is a monocarpic perennial species that can grow to a height of 3 m, and is native to the Caucasus and south-west Asia. It was introduced as an ornamental into Western Europe in the 19th century, and it has spread northwards in the 20th century. Nowadays it is considered naturalized in the Netherlands (Weeda et al. 1987). Heracleum mantegazzianum is a relatively common plant of enriched disturbed habitats, such as road verges and public parks; it also occurs in forest edges (Weeda et al. 1991). Heracleum sphondylium L. is smaller (< 1.50 m) than H. mantegazzianum and is found in the same habitats. Both Heracleum species often co-occur and sometimes produce hybrids (Grace & Nelson 1981). Heracleum sphondylium was introduced and became established in Canada (Page et al. 2006).
Tragopogon dubius Scop. (Asteraceae) originates from central and southern Europe and has extended its range northwards in the 20th century. In the 1950s, T. dubius was first found in the Netherlands and in the 1980s it started spreading naturally along railways. Currently, T. dubius occurs throughout large parts of the Netherlands at sites with a warm microclimate (Weeda et al. 1991). The native T. pratensis L. is morphologically very similar to T. dubius and also occurs in road verges, on dikes and in semi-natural grasslands. Tragopogon pratensis is naturalized in Canada and northern USA (Clements et al. 1999).
Eragrostis pilosa (L.) P. Beauv. (Poacaeae) is a C4 grass of warm and temperate regions, including southern and central Europe. It was first found in the Netherlands in 1958 and started to spread naturally in the 1970s (Weeda et al. 1994). After an apparent lag phase, it is now a very common species in the Netherlands (Tamis & Van ‘t Zelfde 2003) occurring on sidewalks, parking lots and on highly disturbed road verges. This species is paired with Poa annua L. (Poacaeae), which is a C3 grass, and therefore less drought tolerant. It occurs in habitats similar to E. pilosa. Poa annua has become cosmopolitan by human dispersal (Ryan et al. 2003).
soil and seed collection
Soil from the rhizosphere of all six plant species was collected from populations in the Netherlands. Soil samples from Eragrostis, Poa and Heracleum spp. were collected from the centre of the Netherlands (51°58′ N, 5°39′ E), soil from Tragopogon pratensis was collected along the river Waal (51°48′ N, 5°19′ E) and Tragopogon dubius in the west of the Netherlands (52°10′ N, 4°30′ E). For each species, a total of 20 L of soil was collected from the root-zone of 20 plants. The soil of each plant species was sieved through a mesh of 5-mm to remove roots and coarse fragments. Soils were homogenized and split into two fractions, one of which was sterilized by gamma irradiation (> 45 kGray), which effectively eliminates all soil biota (McNamara et al. 2003). The sterilized and non-sterilized soils were used as inoculum and homogenized at a 1 : 6 w/w ratio with a sterilized mixture of sandy loam soil and river sand (1 : 5 w/w, sterilized by gamma irradiation > 45 kGray). Pots (diameter 14 cm, height 13 cm) were filled with 1100 g of the resulting soil mixture. Five samples were taken for nutrient analysis per soil type (see Appendix S1 in Supplementary material).
Seeds were collected from the same plant populations as used for soil collection, except for seeds of T. pratensis which were bought from a small company that collects seeds from wild Dutch populations. Seeds of both Heracleum species were stored at 7 °C for 6 weeks during which they were soaked in water once a week to break dormancy. All seeds were germinated on sterilized mineral sand.
In a conditioning experiment, all three plant pairs were grown in soil with a sterilized or non-sterilized inoculum of con- or heterospecific origin. Half the pots received additional nutrients. There were five replicates of each treatment, resulting in: 2 (inoculum origin) × 2 (sterilization) × 2 (nutrients) × 5 (replicates) = 40 pots per species, and 240 pots in total. Of each forb species one similar sized seedling was planted in each pot, whereas two seedlings were planted per pot in the case of grasses.
The pots with additional nutrients received 50 mL nutrient solution pot−1 week−1 (KNO3 12.5 mmol L−1; Ca(NO3)2 6.50 mmol L−1; MgSO4 3.75 mmol L−1; NH4H2PO4 7.50 mmol L−1; FeEDTA 0.21 mmol L−1) starting after 3 weeks and maintained until harvest. This resulted in 12.3 mmol N, 3.8 mmol P and 6.3 mmol K per plant, or an addition of approximately 1.5–2 times the available N and 2.5–4 times the available P and K at the start of the experiment. Nutrients were supplied during watering so that pots without nutrient addition received the same amount of water.
Pots were placed in five randomized complete blocks in a greenhouse with 70% relative humidity, 16/8 h light/dark at 20/15 °C. (Philips, Amsterdam, The Netherlands) SOL-T armatures were used for additional lighting, providing 80 µmol m−2 s−1 when sunlight was insufficient. Twice a week, soil moisture was reset to 20% using a Theta probe and weighing. In between, water was supplied to compensate for evapotranspiration. Twice a week ripe seeds of P. annua were removed and stored. Total seed biomass was added to the final dry weight (determined after 12 weeks when plants were harvested, separated into shoots and roots, dried at 70 °C for at least 48 h and weighed). Soil from each individual pot was stored separately in plastic bags.
Feedback responses were determined by growing every species in soil inoculated with soil conditioned by a conspecific to test plant–soil feedback (Reinhart et al. 2003; Callaway et al. 2004). To test for the effect of the species on each other through the soil community (Eppinga et al. 2006), plants were also grown in soil conditioned by the other species within the species pair. Prior to the feedback experiment, the replicates of the conditioned soils of each species were homogenized and split into two portions, one to be sterilized by autoclaving for 3 hours at 121.5 °C and the other to remain non-sterilized. The conditioned soils were used as inoculum and homogenized at a 1 : 6 w/w ratio with the same sterilized substrate as used for the conditioning phase. For each treatment there were five replicates. Pots were placed in five randomized complete blocks in a greenhouse with conditions described above, except that no additional nutrients were added. Plants were harvested after 12 weeks and root and shoot biomass determined after drying at 70 °C for at least 48 h.
The difference between the biomass of plants grown in soil inoculated with sterile and non-sterile soil was calculated as the percentage of the biomass of plants inoculated with sterile soil. This value is the proportional reduction in biomass production by the inoculated soil community.
For the conditioning experiment, a general linear model was applied with plant pair, origin of the plant species (native vs. exotic), inoculum origin (native vs. exotic) and nutrients as independent factors, and including all their interactions, and inoculation effect as the dependent variable. Data were then analysed pair-wise with the same model, excluding plant pair as a factor. To test if inoculation effects differed from 0, one-sample t-tests were used. In the feedback experiment a general linear model with origin of species (native vs. exotic) and plant pair was used for both co- and heterospecific treatments. All analyses were carried out using SPSS 12.0.1 (SPSS Inc., Chicago, USA).
In an overall anova (see Appendix S2) we tested the effect of plant status (native or exotic), inoculum, plant pair and nutrient addition on the biomass production in inoculated vs. sterilized soil (= inoculation effect). There were no significant main effects of status, inoculum origin, plant pair or nutrients (P > 0.05). However, there was a significant three-way interaction between status, inoculum origin, and plant pair (F2,96 = 4.659, P = 0.012). Other interactions were not significant (P > 0.05). Therefore, we decided to carry out three-way anova's (see Appendix S2) for the three plant pairs in order to test the effect of status, inoculum origin, and nutrients on the inoculation effect.
For Heracleum the inoculation effect differed between the two species (F1,30 = 4.238, P = 0.048, Table 1). Soil inoculation reduced the native H. sphondylium more than the exotic H. mantegazzianum. Other main factors and their interactions did not influence the inoculation effect (inoculum origin: F1,30 = 0.623, P = 0.436; nutrient level: F1,30 = 0.936, P = 0.341). Therefore, although nutrient addition resulted in an 8.1-fold increase in biomass for H. mantegazzianum and an 11.3-fold increase for H. sphondylium, this did not influence the inoculation effect.
Table 1. Mean effect of inoculation (and standard error), calculated as a reduction of biomass relative to sterile soil, for every combination of plant species, inoculum and nutrient level in the conditioning experiment. Significant reduction is indicated by * (t-test, P < 0.05)
For Tragopogon the inoculation effect was influenced by nutrient addition (F1,32 = 22.178, P < 0.0001), but not by species (F1,32 = 0.738, P = 0.397) or inoculum origin (F1,32 = 0.690, P = 0.412). There was no interaction between any of the treatments. While nutrient addition enhanced biomass 8.1- and 7.8-fold for T. dubius and T. pratensis, respectively, it tended to reduce the inoculation effect.
For E. pilosa and P. annua the inoculation effect did not differ between species (F1,32 = 0.000, P = 0.952), inoculum origin (F1,32 = 0.001, P = 0.927), or nutrients (F1,32 = 0.000, P = 0.981). However, there was a significant interaction between species and inoculum origin (F1,32 = 5.314, P = 0.028). Nutrient addition resulted in a 3.3- and 3.9-fold increase in biomass for E. pilosa and P. annua, respectively, without influencing the inoculation effect.
In a two-way anova (see Appendix S2 and S3), the inoculation effect on conspecifics (= within species) differed between native and exotic species (F1,30 = 15.95, P < 0.001), but not between species pairs (F2,24 = 1.669, P = 0.210). There was no interaction between inoculum origin and species pair (F2,24 = 0.795, P = 0.463). The main effect on natives vs. exotics was due to native species having a stronger negative inoculation effect than the exotic species. On average, inoculation reduced the biomass of the native species by 35% (SE = 4.5%), whereas the biomass of the exotic plant species were reduced by 13% (SE = 3.2%; Fig. 1). In all plant species in all soils the plants on sterilized soil had a significantly higher biomass than the plants in the inoculated soils (one-sample t-test P < 0.05). Therefore, while the biomass of all plant species was reduced by inoculation with soil conditioned by a conspecific, this was more severe in the case of native, as opposed to exotic, species.
The feedback effect of the inoculum conditioned by the other species in the pair (the heterospecific) was not different between the native and exotic species (F1,24 = 2.883, P = 0.102) or species pairs (F2,24 = 1.669, P = 0.210), and there was no interaction between species and species pair (F2,24 = 0.162, P = 0.852) (Fig. 2). On average the biomass of the native species was reduced by 17% (SE = 6.0%) by inoculation, whereas the exotic plant species were reduced by 28% (SE = 2.6%). Therefore, the feedback effects appeared to be species specific within pairs of native and exotic species.
In order to test our second hypothesis, we compared the feedback effect of the inoculum conditioned by the exotic species between the native and the exotic species. Interestingly, there was no difference in feedback from the soil conditioned by the exotic plant to the exotic and the native plant species (F1,24 = 0.239, P = 0.309). This shows that, in their own soil, exotic species did not have a more positive soil feedback effect than a related native species. This result also implies that the exotic species may not exert indirect negative effects, either through the soil community or through allelopathic effects, to (related) native species.
Our results show that exotic plant species that have colonized north-western Europe from southern climate regions have a less net negative plant–soil feedback than similar species which are native in the new range. Whereas plant–soil feedback reduced the average biomass production of native species by 35%, the exotic species demonstrated only a 13% reduction. It appears that the exotic species are less negatively affected by the soil community, either due to reduced exposure to pathogens or due to more effective mutualists, such as arbuscular mycorrhizal fungi or even allelopathic effects. This difference is absent in the reciprocal comparison, indicating that it is an effect of the interaction between plant species and soil community. Our results show that the net pathogenic activity must be specific at the species level, such as has been demonstrated for dune plants that succeed each other (Van der Putten et al. 1993).
Besides specificity, another prerequisite for enemy release along a climate gradient is that the soil pathogens are less mobile than their host plants. In contrast to agricultural systems (e.g. Levenfors & Fatehi 2004), the specificity and mobility of soil pathogens in natural systems has been poorly studied. Escape from soil pathogens has been shown at local scales (D’Hertefeldt & Van der Putten 1998; Blomqvist et al. 2000; Olff et al. 2000; Van der Stoel & Van der Putten 2006) and at large spatial scales (Beckstead & Parker 2003; Van der Putten et al. 1993; Reinhart et al. 2003; Van der Stoel & Van der Putten 2006). Little is known, however, about the role of dispersal of soil organisms in the population dynamics of their host plants. Soil pathogens are thought to be relatively immobile when compared to most above-ground plant enemies (Van der Putten et al. 2001) and our results suggest that range expansion of plants, for example due to climate change, may, at least temporarily, release the plants from their natural soil-borne enemies. Range expansions are often considered as natural, gradual phenomena and invasion of exotics as abrupt ones, due to anthropogenic introduction. This would imply that different mechanisms underlie these processes. Here, we show that the effects can be comparable. Although biotic interactions are often neglected in climate change studies (Schmitz et al. 2003; Brooker et al. 2007), the need to incorporate their effects in predictions of future distributions is increasingly recognized (Davis et al. 1998), as is also demonstrated in our study.
The plants used in the present study represent three of the largest families of the European flora which themselves are unrelated. The different pairs of plants occur in different habitats and are different functional types. The similarity observed between the pairs therefore can not be attributed to a phylogenetic or ecological bias but indicates that a reduced plant–soil feedback is common amongst expanding plant species.
Experiments with sterilized soil are often confounded by nutrient release due to the sterilization procedure (Troelstra et al. 2001; McNamara et al. 2003). It is difficult to tease apart positive effects of eliminating soil pathogens, negative effects of killing symbionts and positive effects due to enhanced availability of mineral nitrogen and phosphate. In our experiment the inocula were used in a 1 : 6 w/w ratio in a sterilized bulk soil, thus reducing the difference in nutrient content between treatments as the bulk soil has the same nutrient input for all treatments. In the conditioning experiment we would expect that addition of nutrients would decrease the difference between biomass production with sterilized or-non-sterilized inocula if this was due to nutrient release. In two of the three pairs the nutrient level did not influence the effect of inoculation even though the addition of nutrients strongly increased biomass production. It is therefore unlikely that the observed effects were caused by nutrient release due to sterilization of the inocula. In the third pair, Tragopogon, the addition of nutrients did reduce the inoculation effect. The amount of nutrients added was relatively large compared to the soil nutrient content (see Appendix S1) and had a strong effect on biomass production (3–11 fold increase). The differences between native and exotic species in the feedback experiment cannot be explained by differences in nutrient availability as this would result in the same pattern in the reciprocal treatment and not in a species-specific result.
Abiotic plant–soil feedback responses via allelopathic effects have also been reported (Pellissier 1998; Souto et al. 2001) and the structure of some allelochemicals present in the soil can change due to autoclaving, thus causing an apparent pathogenic effect (McPherson & Muller 1969). In particular Heracleum species are known to exhibit allopathic effects on other plant species (Junttila 1975; Myras & Junttila 1981). Allelopathic substances typically have a more negative effect on heterospecifics than on conspecifics (Inderjit et al. 2005) and this would result in a strong feedback effect to the other species in the pair, but not to itself. We did not observe such an effect and therefore it is less likely that allelopathy has played an important role in our study. Moreover, we diluted rhizosphere soil by using a sterilized standard bulk soil and this also would have reduced the direct effects of allelopathy.
The effect of the heterospecific soil community was the same for both native and exotic species. Furthermore, exotic and native plants did not differ in the conditioning phase. This means that when establishing, native and exotic species do not show different responses and that the advantage of exotic species over native species becomes apparent only in the feedback stage. Plants have been suggested to gain an advantage of plant–soil feedback even if the direct effect is negative, for example when the soil community developed in association with a species has a more negative effect on competitors than on the species itself (Eppinga et al. 2006). However, in our study the effect of the soil conditioned by the exotic plants on the native and exotic plants did not differ, so at least in these species this mechanism does not play a role.
Plant species colonizing a new area often exhibit reduced pathogen and/or herbivore load, potentially increasing their competitive ability, growth and reproductive output (Blaney & Kotanen 2001; Mitchell & Power 2003; Callaway et al. 2004; Van der Putten et al. 2005). Recently, soil pathogens have been identified as a potentially important group of organisms that can drive plant colonization and competition processes (Agrawal et al. 2005; Torchin & Mitchell 2004; Van der Putten et al. 2005). Here we show that the negative effect of soil pathogens on plant growth can be quite substantial and differs between native plant species and species that recently colonized an area, even when this is due to a range expansion rather than due to a remote introduction. To study the effects of the soil community as a whole is helpful and sufficient to understand general mechanisms (Colautti et al. 2004). Further studies, however, need to identify key soil organisms and their roles (Brinkman et al. 2005). This might provide insight to the relative importance of the different groups of organisms and underlying mechanisms for the different plant species. Separating the effects of pathogens and mutualists (Richardson et al. 2000) is especially important in order to distinguish the positive and negative components of the net feedback.
We wish to thank Frans Moller and Jan van Walsem for their assistance and Jasper van Ruijven for his useful comments on previous versions of the manuscript. This work was funded by The Wageningen Institute for Environment and Climate Research (WIMEK).