Introduction

Invasive species may outcompete natives either due to a higher relative growth rate (Pattison et al. 1998; van Kleunen et al. 2010; Lamarque et al. 2011), higher stress tolerance (Glenn and Nagler 2005; Sher and Marshall 2003), higher capacity to acclimate to a wider range of conditions (Richards et al. 2006; Davidson et al. 2011), or higher resource use efficiency (Funk and Vitousek 2007). However, the final relative advantage of invaders over natives or vice versa would depend on the particular environmental conditions where both groups of species coexist (Daehler 2003).

In the Mediterranean region of Europe, riparian floodplains are among the most vulnerable habitats to plant invasions, because of the mildness of microclimatic conditions and the abundance of resource availability (Chytrý et al. 2008, 2009). In this habitat, irradiance and/or soil moisture are the main factors limiting plant performance, as nutrient availability is usually high in floodplains (González et al. 2010). Many riparian trees of the Mediterranean region are light-demanding species and find their regeneration niche in the gaps opened after flooding episodes (González et al. 2010; González-Muñoz et al. 2011). However, human management of rivers has altered these natural disturbance regimens. River channelization increases the erosion of the river bottom, and therefore increases the depth of the water table, raising the chances for the establishment of water-stress tolerant plants (Glenn and Nagler 2005; Sher and Marshall 2003). Furthermore, river regulation declines the natural rate of gap formation in riparian forests, potentially increasing the chances for the establishment of shade-tolerant species. Contrastingly, human disturbance eliminates riparian forests, creating open habitats that may favor the establishment of new light demanding species. These non-naturally created tree regeneration niches in Mediterranean riparian forests (increased light-decreased water and decreased light-decreased water availabilities) pose the question of whether invasive tree species will be favored by the new environmental conditions over the existing natives.

The performance of tree seedlings may determine the future composition of forest communities (Pacala et al. 1996; Kobe and Coates 1997; Baraloto et al. 2005). Seedlings are more vulnerable than adults to adverse factors, such as low light or soil moisture availabilities (Niinemets and Valladares 2006). If exotic seedlings are able to reach a bigger size at the same time as seedlings of co-occurring natives, they would be able to monopolize below and above ground resources, which may lead to a future domination of forest communities (Blumenthal and Hufbauer 2007; Closset-kopp et al. 2011). Different plant attributes may promote high growth rates in invasive plants such as large specific leaf area (Baruch and Goldstein 1999; Daehler 2003; Porté et al. 2011) or high foliar nutrient concentrations (Ehrenfeld 2003; Leishman et al. 2007; Peñuelas et al. 2010), both being associated to a high photosynthetic rate. Besides, small seed sizes have been associated to both high relative growth rates at early stages and long distance dispersion (Marañón and Grubb 1993; Swanborough and Westoby 1996; Reich et al. 1998; Grotkopp et al. 2002), whereas large seed sizes promote more competitive seedlings, with higher survival rates (Howe and Richter 1982; Howe 1990; Moles and Westoby 2004; Quero et al. 2007). In plant invasions, a small seed size has been previously related to invasion success in some fast-grower species, such as pines (Rejmánek and Richardson 1996; Hamilton et al. 2005), but not in others, such as acacias (Castro-Díez et al. 2011). Indeed, the final advantage of having a certain seed size and its consequences on seedling performance will depend on the environmental conditions in which both the seed emergence and the seedling recruitment occurs (Schupp 1995, 1988). Finally, an early emergence may also be a helpful trait as it contributes to an early space occupation and a consequent competitive advantage over late-emerged seedlings (Jones and Sharitz 1989; Verdú and Traveset 2005; Castro 2006).

Many exotic species occur in the riparian forest of in inner Spain. Among them, Acer negundo L. (Sapindaceae), Ailanthus altissima (Mill.) Swingle (Simaroubaceae), Elaeagnus angustifolia L. (Eleagnaceae) and Robinia pseudoacacia L. (Fabaceae) are ranked as “invasive” in the Atlas of Exotic Invasive Plants in Spain (Sanz Elorza et al. 2004) and in the Inventory of Alien Invasive Species in Europe (DAISIE). These species might be favored by the mentioned environmental changes promoted by human actions to the detriment of the native vegetation, mainly dominated by the tree species Fraxinus angustifolia Vahl. (Oleaceae), Populus alba L. (Salicaceae) and Ulmus minor Mill. (Ulmaceae). However, little information about the environmental preferences of all these species is available so far. To fill in this gap, we assessed native and invasive plant growth under a factorial experiment with four irradiance and two soil moisture treatments, mimicking the wide range of natural and human-mediated environmental conditions existing in Mediterranean riparian forests. We hypothesized that (1) seedlings/saplings of the invasive species would accumulate on average more biomass than those of natives, which in turn would give to the invaders an advantage over co-occurring natives (van Kleunen et al. 2010; Lamarque et al. 2011); (2) both invasive and native species will reach higher biomass under resource-abundant treatments but invaders will perform better than natives, according to the opportunistic strategy attributed to most invasive plant species (Pyšek et al. 1995; Rejmánek and Richardson 1996; Hamilton et al. 2005).

Materials and methods

Study species

We studied native and invasive tree species co-occurring in riparian forests of inner Spain. In these forests, the native vegetation is structured along a double gradient of soil moisture and flooding frequency, stretching from the river edge to outwards. According to the water requirements and flood tolerance of the species, the inner band of woody vegetation (closer to the river) is dominated by several shrub species of Salix, followed by the trees Salix alba L. and Populus nigra L. The middle part of the gradient is dominated by P. alba while the outer part is dominated by F. angustifolia and U. minor (Blanco Castro et al. 2005). This typical structure of central Spain riparian forests has been severely altered by river regulation and channelization, which particularly threaten the inner band. The group of invaders here studied, A. negundo, A. altissima, E. angustifolia and R. pseudoacacia, are spreading along the middle and external bands of the riparian vegetation. These species were introduced between eighteenth and early twentieth centuries in Spain with ornamental purposes (Sanz Elorza et al. 2004). The presence of A. negundo and E. angustifolia is normally associated to river courses, whereas R. pseudoacacia and A. altissima can be also found in disturbed sites, such as road sides or crop borders (Sanz Elorza et al. 2004). Nowadays, these tree species are catalogued as invasive by different authors in Spain and Europe (Sanz Elorza et al. 2004; DAISIE), and their demographic trends are defined as “expansive” in Spain (Sanz Elorza et al. 2004). Moreover, A. altissima and R. pseudoacacia are listed among the 100 worst invasive species in Europe (DAISIE). The invasive group was compared with the natives P. alba, F. angustifolia and U. minor because they are the dominant species in the invaded areas (Sanz Elorza et al. 2004; Lara et al. 1996).

Experimental design

The experiment was set up outdoors at the Botanical Garden of Alcalá University (Madrid, central Spain 40°30′N, 3°20′W, 596 m a.s.l). Climate is continental Mediterranean with hot and dry summers and cold winters. Mean annual maximum and minimum temperatures are 20.5 and 7.8 °C, respectively. Mean annual precipitation is 378 mm (data from Torrejón de Ardoz weather station, Instituto Nacional de Meteorología, 1971–2008).

Four light treatments (L100, L65, L35 and L7, corresponding to 100, 65, 35 and 7 % of full sun irradiance) were established, aiming to mimic the light gradient existing in a typical Mediterranean forest, from the gaps (L100) to the under-canopies of dense forests (L7) (Valladares 2004a, b). Light treatments (except L100) were obtained with green nets of different thickness fixed to metal frames on the top and the four sides.

These light treatments were crossed with two soil moisture levels, corresponding to 61 % (high moisture, HM) and 40 % (low moisture, LM) of soil gravimetric water content. These levels corresponded to soil matric potential of −1.63 and −1.99 MPa, respectively, as estimated by the filter-paper technique (Deka et al. 1995). Soil moisture levels aimed to induce changes in species success without causing massive mortality, based on our previous experience (Castro-Díez et al. 2006, 2007). To determine the amount of water needed to keep the target soil moisture in each treatment during the experimental period, we conducted a pot weight-soil moisture calibration before the beginning of the experiment [see details in González-Muñoz et al. (2011)], which allowed us to calculate the amount of water needed on the basis of the pot weight. Soil moisture treatments were applied from May to October of 2008 and 2009, while in the rest of the experimental period plants just received rainwater. Five trays or 15 pots per treatment were weighed twice a week to estimate their soil gravimetric water content on the basis of the previous calibration. If the weight was lower than that corresponding to the treatment, we added by hand and/or by a watering system the amount of water needed to reach the target soil moisture. Kurskall–Wallis tests revealed significant differences of soil moisture before (H = 112.65, p < 0.001) and after watering (H = 206.48, p < 0.001) between the two treatments.

Local air temperature, % of air humidity and available photosynthetic photon flux density were recorded every 5 min throughout the experiment period with climatic sensors connected to data loggers (HOBO model H08-006-04; Onset, Pocasset, MA, USA). We tested the differences in mean, maximum and minimum temperature and relative air humidity between light treatments throughout the growing seasons (May to September 2008 and 2009) with one-way ANOVAs followed by post hoc Tukey tests. Light treatments did not differ either in mean and minimum temperature or relative air humidity (p > 0.05). However, the shaded treatments showed a lower maximum temperature, due to the effect of hot alleviation produced by shade (F3,40 = 5.99; p = 0.002) (Appendix 1 of supplementary material). On January 2009, during all species dormant period, there was a strong snowfall that equally affected all the light treatments. The potential effects of this snowfall are considered in the discussion.

Sowing procedure

The seed bank of the Spanish Ministry of Environment supplied seeds of the targeted species (MAGRAMA). Before the onset of the experiment, we conducted germination essays at the laboratory to maximize seed emergence (see González-Muñoz et al. 2011 for further details). Average seed mass was assessed for each species by weighting 30 seeds after >72 h in the oven at 60 °C.

On April 2008, we sowed seeds of all species under the eight treatments resulting from crossing the light and soil moisture levels. A. altissima, E. angustifolia, F. angustifolia and P. alba were sown in single trays (38 length × 28 width × 7 depth cm, 40 seeds per tray), as they had small seeds and/or showed low percentages of emergence in the previous germination essays. U. minor, A. negundo and R. pseudoacacia were sown in multipot trays (24 pots of 330 ml per tray, one seed per pot), as they had large seeds and/or high percentages of emergence in the germination essays. The experimental soil was 1:2 volume mixture of washed river sand and commercial substrate 15-10-20 NPK- Kekkilä Iberia S.L., Valencia, Spain.

We calculated the average of days required by a seedling to emerge in a tray (time to emerge or Temerg) as:

$$T_{emerg} = \frac{{\sum\nolimits_{i = 1}^{n} {(i \times No.\, seedlings_{i} )} }}{N}$$

where i = sampling day, n = total number of days sampled, No. seedlings i  = number of seedlings emerged on day i, N = cumulative number of emerged seedlings until day n.

Plants were left in trays until April 2009 with no additional fertilization. Then, 15–20 plants per species and treatment were transplanted to individual 1.5 L pots. The number of transplanted plants depended on the amount of individuals available in each species and treatment, which differed among species and treatments due to differences in emergence and survival (see González-Muñoz et al. 2011). During the second growing season, pots were monthly fertilized with a 100 mg/l solution of NPK 15-10-20 (Peters Professional, The Scott Company, Brantford, Ontario). Trays/pots of all species were randomly arranged in each of the eight treatments and randomly re-arranged within each treatment twice a week to guarantee treatment homogenization.

Data collection

Individuals were harvested at the end of the first growing season (November 2008), in the middle of the second growing season (July 2009) and at the end of the second growing season (November 2009). We collected an average of 4 (3–5) and 7 (4–9) individuals per species and treatment in the two first and in the third harvest, respectively. In July 2009, we did not collect plants of U. minor under L35 and L100 and of A. altissima and R. pseudoacacia under L35 due to the low number of available plants.

Later, plants were separated into roots, stems and leaves and oven-dried at 60 °C for 48 h. Relative root, stem and leaf weight ratios (RWR, SWR and LWR respectively) were estimated in July 2009 harvest, when the seedlings presented full foliage.

Seedling/sapling biomass at the end of the first and second growing seasons (November 2008 and 2009 respectively) were analyzed excluding leaf mass as in this month harvested plants were at different stages of leaf abscission. The average relative growth rate (RGR) between November 2008 and November 2009 harvests was calculated for each species and treatment as:

$${{{\text{RGR}} = \left( {\ln \left( {{\text{Biomass}}_{\text{Nov2009}} } \right) - \ln \left( {{\text{Biomass}}_{\text{Nov2008}} } \right)} \right)} \mathord{\left/ {\vphantom {{{\text{RGR}} = \left( {\ln \left( {{\text{Biomass}}_{\text{Nov2009}} } \right) - \ln \left( {{\text{Biomass}}_{\text{Nov2008}} } \right)} \right)} {\text{days between harvests}}}} \right. \kern-0pt} {\text{days between harvests}}}$$

This RGR was used to estimate the net assimilation ratio per unit of leaf mass (NARm) (Lambers et al. 2008) as:

$${\text{NAR}}_{\text{m}} = {{\text{RGR}} \mathord{\left/ {\vphantom {{\text{RGR}} {{\text{LWR}}_{\text{Jul2009}} }}} \right. \kern-0pt} {{\text{LWR}}_{\text{Jul2009}} }}$$

Statistical analyses

We conducted four-way nested ANOVAs to test the effect of origin (native or invasive), species (nested in origin), and treatment (light and soil moisture) on the biomass reached in November 2008 and 2009 and on the RWR, SWR and LWR of July 2009. Post hoc Bonferroni tests were conducted to assess differences between species.

Two-way nested ANOVAs were performed to test differences in average RGR, NARm and Temerg between origins and among species (nested in origin). For these variables, we only had the average value for each species and treatment (see above), so we could not assess the effects of treatment. Post hoc Bonferroni tests were also used to assess differences between species.

Finally, in each species, we tested the effect of light and soil moisture on the sapling mass of November 2009 and on the biomass allocation traits by means of two-way ANOVAs. Again, post hoc Bonferroni tests were used to check differences among light treatments. Prior to any analysis, data were checked for homoscedasticity (Bartlett test) and normality assumptions (Shapiro–Wilk test) in any case.

We also conducted Pearson correlations to test the effect of seed size on the biomass reached in November 2008 and 2009, biomass allocation traits, RGR, Temerg and NARm.

Statistical analyses were conducted using R 2.13 package (library “stats”; R Development Core Team 2011).

Results

Differences in biomass between origins

In November 2008, invasive seedlings were slightly larger than native seedlings, due to the high biomass of A. negundo and E. angustifolia, but this difference was not significant (Table 1; Fig. 1a). By contrast, in November 2009, native saplings were significantly larger than invaders, due to the low biomass of A. altissima (Table 1; Fig. 1b). There was a lower effect of the light and soil moisture treatments on the biomass averaged by origins than on the biomass reached by the species (see the significant interactions in Table 1; Figs. 2a–d, 3).

Table 1 Effect of origin (native or invasive, O), species (nested in origin, Sps(O)), light (L), soil moisture (M) and their interactions on the biomass reached at the end of the first (Nov 2008, N = 210) and second growing seasons (Nov 2009, N = 357) and on the root, stem and leaf weight ratios (RWR, SWR and LWR respectively) shown in July 2009 (N = 207), according to four-way ANOVAs. F, degrees of freedom (df) and significances (p) are shown
Full size table
Fig. 1
figure 1

Biomass (mean + SE, in g) without leaves reached by each species at the end of the first (2008) and second (2009) growing seasons (all treatments averaged). Different letters among columns mean significant differences among species according to post hoc Bonferroni tests. Invasive species (white bars): A. altissima (A. alt, N2008 = 33; N2009 = 47); A. negundo (A. neg, N2008 = 32; N2009 = 57), E. angustifolia (E. ang, N2008 = 30; N2009 = 54) and R. pseudoacacia (R. pse, N2008 = 30; N2009 = 51). Native species (black bars): F. angustifolia (F. ang, N2008 = 30; N2009 = 57), P. alba (P. alb, N2008 = 28; N2009 = 48) and U. minor (U. minor, N2008 = 27; N2009 = 43)

Full size image
Fig. 2
figure 2

Biomass (mean ± SE, in g) reached by the invasive (white) and native (black) species in every combination of soil moisture (LM low moisture, HM high moisture) and light treatment (L7—7 %, L35—35 %, L65—65 % and L100—100 % respect to full irradiance) at the end of the first and second growing seasons (2008 and 2009, respectively)

Full size image
Fig. 3
figure 3

Biomass responses to the light and soil moisture treatments showed by each species after two growing seasons. The bubble size is proportional to the mean seedling biomass without leaves achieved in November 2009. Different letters mean significant differences between treatments, according to post hoc Bonferroni tests. L7, L35, L65 and L100 correspond to 7, 35, 65 and 100 % respect to full irradiance, respectively. LM and HM correspond to low moisture and high moisture respectively. Missing values are due to insufficient number of individuals in some species and treatments. See Appendix 2 of supplementary material for further information about the statistical results

Full size image

Differences in biomass and related traits among species

In November 2008, A. negundo, E. angustifolia, F. angustifolia and U. minor showed similar biomass and higher than that of A. altissima, R. pseudoacacia and P. alba (Fig. 1a). In November 2009, R. pseudoacacia and P. alba caught up the group of the larger species and only A. altissima remained smaller (Fig. 1b). R. pseudoacacia and P. alba showed the highest RGR and, together with A. altissima, the highest LWR (Table 2). F. angustifolia and E. angustifolia required longer time to emerge and exhibited the highest investment in roots and stems respectively (Table 2). There were no significant differences among species in NARm (Table 2). Seed size did not significantly account for the differences found in the final biomass, biomass allocation traits, RGR, NARm and Temerg (Appendix 3 of supplementary material).

Table 2 Mean values (±SE) of the relative growth rate calculated from the end of the first to the end of the second growing seasons (RGR), root, stem and leaf weight ratios in middle second growing season (RWR, SWR and LWR respectively), net assimilation ratio per unit of leaf mass (NARm), time needed to emerge (Temerg) and seed mass in each origin (invasive or native) and species
Full size table

Effects of light and soil moisture on the species biomass

After two growing seasons, all species tended to grow more with an increase in light availability, this response being steeper in P. alba, F. angustifolia, R. pseudoacacia and E. angustifolia (Fig. 3). Soil moisture positively affected the biomass of all species but A. negundo and U. minor under most light treatments (Fig. 3).

Effects of light and soil moisture on the species biomass allocation

Biomass allocation patterns were affected by light and soil moisture treatments (Table 1; Fig. 4). Among the invaders, A. negundo exhibited a quite balanced allocation between roots, stems and leaves across the light gradient, but increasing moisture enhanced the investment to leaves to the detriment of roots (Fig. 4, Appendix 2 of supplementary material). E. angustifolia invested more biomass in leaves under the extreme light treatments (L100-L7) but increased RWR under the intermediates (L35–L65) (Fig. 4, Appendix 2 of supplementary material). A. altissima and R. pseudoacacia increased SWR with decreasing light (Fig. 4, Appendix 2 of supplementary material). Moreover, R. pseudoacacia increased RWR with decreasing soil moisture (Fig. 4, Appendix 2 of supplementary material). The three natives increased aboveground biomass under L7 (LWR or SWR, Fig. 4, Appendix 2 of supplementary material). F. angustifolia decreased LWR and P. alba decreased LWR and increased RWR with increasing soil moisture (Fig. 4, Appendix 2 of supplementary material).

Fig. 4
figure 4

Root weight ratio (RWR, black), stem weight ratio (SRW, grey) and leave weight ratio (LWR, white) values (mean + SE) showed by each species under each combination of soil moisture (LM low moisture, HM high moisture) and light (L7—7 %, L35—35 %, L65—65 % and L100—100 % respect to full irradiance) in July 2009. Missing values are due to insufficient number of individuals in some species and treatments

Full size image

Discussion

Differences in biomass between origins

Invasive species as a group attained the same biomass as natives in the first growing season, or even less in the second one, in contrast to our initial hypothesis (Fig. 1a, b). This suggests that other species traits, different to those evaluated in this work, must explain their establishment success in inner Spain riparian forests. A high interspecific competitive ability together with a high tolerance to intraspecific competition have been suggested to promote both dense stands of invasive species and the competitive exclusion of native species (Baker 1965; Roy 1990). This was corroborated by the extensive meta-analysis of Vilà and Weiner (2004), which showed that the competitive effect of invasive species on natives is usually stronger than vice versa. Other features have been suggested to confer high competitive ability to invasive plants, such as high fecundity, high propagule pressure, the ability of profusely resprout, the production of allelopathic compounds, and a reduced herbivory damage (Rejmánek and Richardson 1996; Callaway and Aschehoug 2000; Keane and Crawley 2002; Richardson and Pyšek 2006). For instance, adults of A. altissima produce allelopathic compounds that may inhibit the seed germination and seedling growth of other species (Heisey 1990, 1996; Heisey and Heisey 2003; de Feo et al. 2005). Besides, all our studied invasive species are able to profusely resprout and to produce a high amount of seeds (Weber 2003; Katz and Shafroth 2003; Sanz Elorza et al. 2004; Kowarik and Säumel 2007; Masaka and Yamada 2009). Finally, the studied invaders may have benefited from the spatial or temporal empty niches that floods generates in river sides, where they would not need a superior performance over natives to succeed (Godoy et al. 2009).

Differences in biomass and related traits among species

We found strong differences among species in all studied variables. Thus, the invader E. angustifolia reached the highest biomass after two growing seasons (although this difference was not significant) whereas A. altissima attained the lowest (Fig. 1b, Table 2). This result disagrees with previous studies that describe A. altissima as fast grower (Knapp and Canham 2000; Sanz Elorza et al. 2004; Kowarik and Säumel 2007). The strong snowfall of January 2009, an uncommon event in the study region, may explain this low growth, as A. altissima seedlings have been described as not frost resistant (Sanz Elorza et al. 2004; Kowarik and Säumel 2007). The large biomass reached by E. angustifolia may be consequence of its large seed size, nearly four times larger than that of the next species in the seed mass ranking (see Table 2).

The invader R. pseudoacacia and the native P. alba attained the highest RGR among studied species whereas the invader A. negundo and the native F. angustifolia attained the lowest (Table 2). The high investment in leaves of the former two species may explain their high RGR, as this trait contributes both to a better light interception (Pearcy et al. 2004) and to high CO2 uptake at the whole plant level (Chmura et al. 2007) (Fig. 4, Table 2). The low RGR of F. angustifolia may be attributed to its large investment in roots (the highest among studied species), what may adversely affect carbon gains, through decreasing leaf mass allocation and increasing root respiratory loss (Weiner 2004); (Fig. 4, Table 2). Finally, the low RGR of A. negundo may be due to its low net assimilation rate per unit of leaf mass (Table 2).

Recent frameworks aiming to predict invasiveness have considered the seed size as an important trait (Moles et al. 2008). However, according to our results, the differences in biomass and related traits found here were not explained by the species seed size. This can be due to the existing similarity between the studied species life strategies (all species were deciduous trees), as the role of seed size can be more relevant between contrasted habitats and between life forms with different productivities (Moles et al. 2008).

The high heterogeneity of growth patterns found in both natives and invaders has also been described in other studies and highlights the independence of the species strategies to their native or invasive origin (Bellingham et al. 2004; Feng et al. 2007; Feng and Fu 2008; Gurevitch et al. 2008) and explains why we did not find differences of growth among invasive and native seedlings.

Effect of light and soil moisture on the biomass and biomass allocation of native and invasive species

All species achieved their highest biomass under high irradiance (L65 or L100) and, in most cases, high soil moisture, as we hypothesized. In low resource conditions, only the native U. minor and the invader A. negundo showed a relative good performance, in accordance to previous studies that described them as shade tolerant species (Fig. 3); (DeWine and Cooper 2007, 2008; González et al. 2010; González-Muñoz et al. 2011; Porté et al. 2011). Indeed, U. minor naturally occurs in the most external vegetation band of riparian forests, where water availability is lower and gap-opening by natural disturbances is less frequent (Blanco Castro et al. 2005).

Regarding biomass allocation patterns, most of the studied species responded to the light and soil moisture treatments following the optimal allocation theory (Weiner 2004). This theory suggests that the species tend to allocate more biomass to the tissues that may favour the capture of the most limiting resource. Indeed, all species except A. negundo increased aboveground biomass (LWR and/or SWR) with decreasing light and A. negundo and R. pseudoacacia increased their investment in roots with decreasing moisture (Fig. 4). Interestingly, the three natives and A. altissima not only increased aboveground biomass when light was low, but also increased RWR with increasing light, then favoring water and nutrients acquisition when light is not limiting.

Conclusions

The lack of superior performance of invasive seedlings/saplings over natives in terms of biomass gain suggests that other mechanisms, such as reproduction strategies or the production of allelopathic compounds may explain their invasion success. Under low resource conditions, as it happens in the floodplains of regulated rivers, either in gaps (high light/low moisture) or under dense canopies (low light/low moisture), only the establishment of the native U. minor and the invader A. negundo would not be hampered (Fig. 3). As U. minor seed recruitment is reduced by the Dutch elm disease in our study area (Brasier et al. 2004; Martín et al. 2006), A. negundo young plants would potentially dominate these scenarios. Therefore, special care should be paid to monitor the presence of A. negundo populations in the floodplains of regulated rivers. Due to the difficulties to restore river channelization, an early detection and eradication of emergent populations is probably the best option to avoid the spread of A. negundo throughout floodplains.