Introduction

Litterfall is a key process for nutrient cycling and energy transfer from plants to soils in terrestrial ecosystems (Vitousek 1984). The quality and quantity of plant litter play an important role in structuring the communities of plants and detritivorous invertebrates living in the soil (Facelli and Pickett 1991; Ponge et al. 1993; Xiong and Nilsson 1999), and the aquatic food webs when the litter falls into aquatic ecosystems (Wallace 1997). In addition, litterfall dynamics, i.e., the timing of abscission of different plant organs (e.g., leaves, flowers, branches), determines the period of food availability for detritivores, as well as the microclimatic conditions for litter decomposition (Abelho and GraÇa 1996; Singh et al. 1999; Kirschbaum 2010). The chemical composition of the litter, particularly nutrient content, largely determines the rate of litter decomposition (Simons and Seastedt 1999; Wang et al. 2008), as well as soil properties, such as organic matter (OM), nutrient concentrations and pH (Witkowski 1991; Rice et al. 2004; Yelenik et al. 2004). Given that litter production, litterfall dynamics and chemical composition of plant tissues vary across species (Bray and Gorham 1964; Facelli and Pickett 1991; Abelho 2001; Martin et al. 2009; González 2012), a change in the species composition of the plant community, e.g., by an exotic plant invasion, may alter the ecosystem processes and soil properties affected by litter properties.

Riparian ecosystems are highly vulnerable to exotic plant invasion in part due to their high frequency of disturbance as well as habitat heterogeneity, which provides more opportunities for exotic plants to find appropriate conditions to establish and persist (Hobbs and Huenneke 1992; Hood and Naiman 2000; Chytrý et al. 2008). The invasion by exotic plants implies changes in plant community composition, which in turn may affect the quantity, quality, and dynamics of litterfall and therefore the functioning of both aquatic and terrestrial ecosystems (Bray and Gorham 1964; Facelli and Pickett 1991; Abelho 2001). The magnitude and direction of the effect of exotic invasive trees on litterfall may depend on the invasive species and the invaded community (Witkowski 1991; Ellis et al. 1998; Rice et al. 2004; Martin et al. 2009; Lee et al. 2011). Litter production in riparian forests can depend on soil properties, community structure and distance to the river (González et al. 2010). However, the link between invasion and soil properties through alteration of litterfall patterns remains largely unexplored.

Ailanthus altissima Mill. Swingle (Simaroubaceae) and Robinia pseudoacacia L. (Fabaceae) are two exotic invasive trees natives to China and North America, respectively (Sanz Elorza et al. 2004). They are found invading temperate ecosystems worldwide and are considered among the 100 worst invasive species in Europe (Kowarik and Säumel 2007; DAISIE 2009; Cierjacks et al. 2013). In Spain they were introduced during the 18th century for ornamental purposes, and they are now spreading through riparian areas, among other habitats, being considered among the 20 most harmful species in Spain (Sanz Elorza et al. 2004; GEIB 2006). Both invaders are deciduous and in Central Spain they coexist with other riparian deciduous trees such as Populus alba L. (Salicaceae), Populus nigra L. (Salicaceae), or Fraxinus angustifolia Vahl. (Oleaceae) (De la Cruz and Peinado 1996; Castro-Díez et al. 2009). R. pseudoacacia is a N2-fixing species capable of increasing N pools and speeding up N cycling in nutrient-poor invaded ecosystems (Rice et al. 2004; Von Holle et al. 2013). A. altissima can also increase soil N and pH in a range of different soils (Vilá et al. 2006; Gómez-Aparicio and Canham 2008). In contrast, Robinia pseudoacacia is able to decrease soil pH due to enhanced nitrification (Vítková et al. 2015). In riparian ecosystems of Central Spain, A. altissima produced higher quantities of fast-decomposing leaf litter than coexisting native trees, while R. pseudoacacia had the contrary effect (González-Muñoz et al. 2013; Castro-Díez et al. 2009, 2012). Phenological studies indicated that both invasive species displayed later flowering and fruit set and earlier leaf abscission than most of the coexisting native trees in Central Spain (Castro-Díez et al. 2014). Because other plant organs can be important sources of nutrients to soil (e.g., Lee et al. 2011), studies of litterfall that exclusively consider the leaf fraction provide only a partial understanding of the impact of plant invaders on soil inputs. Moreover, seasonal variation of litterfall may be essential, but it is rarely studied because of the frequent and sustained sampling required. No one has yet estimated the effects of A. altissima and R. pseudoacacia on soil properties through the whole litterfall process, taking into account all plant organs, their nutrient content and the time when they fall.

The aim of this study was to assess the impacts of A. altissima and R. pseudoacacia on the whole litterfall process and on soil properties of riparian ecosystems of Central Spain by comparing invaded and non-invaded forests. Specifically we tested the effects of the invasive species on: 1) the annual quantity of litterfall from different plant organs and their temporal dynamics 2) nitrogen (N) and phosphorus (P) inputs to the soil from litterfall and their temporal dynamics 3) soil N and P concentration, OM and pH. We expected litter production in invaded forests to be greater than in non-invaded forests, given the higher growth rates and lower N use efficiency reported for these invaders as compared to the coexisting native trees (Kowarik and Säumel 2007; Cierjacks et al. 2013; González-Muñoz et al. 2013; Castro-Díez et al. 2014). Associated with increased litter production, we expected higher inputs of nutrients and higher levels of soil nutrients and OM in invaded forests. We also expected different litterfall dynamics between invaded and native forests due to the contrasting phenologies of these exotic and native trees (Castro-Díez et al. 2014). This study provides valuable data to understand the impacts produced by A. altissima and R. pseudoacacia in riparian Mediterranean forests on the whole litterfall process and thus on soil properties and the detritivore communities that rely on litter.

Materials and methods

Study plots

The study was carried out within the riparian forest (along 80-km stretch) of the Henares River (Tagus Basin, Central Spain). In this area, the climate is continental Mediterranean, with a mean annual precipitation of 341 mm and mean annual temperature of 13 °C during the two study years, and an arid period of at least 2 months (Fig. 1). Native vegetation is dominated by P. alba, P. nigra, F. angustifolia, Salix alba and Ulmus minor. Some areas are invaded by A. altissima and R. pseudoacacia, which may grow in pure stands or mixed with native species. We selected four forest types, i.e., patches of forests dominated by four different tree species: A. altissima, R. pseudoacacia, F. angustifolia and P. alba, occurring under similar environmental conditions (Xerothent + Xerofluvent soils (Fernández et al. 2012, 2013)). For each forest type, we delimited three or four 300 m2 plots where the target species was dominant (see Table 1). Plots were all found within 100 m of the Henares River and within 80 km of each other. Plots had rectangular shape, with the wider side parallel to the river, except one A. altissima plot (AA1) which had triangular shape. In spring 2012 we measured in each plot the diameter at breast height (DBH) of all adult trees (DBH > 7 cm and height > 1.30 m) and calculated the total tree basal area (TTBA) (Ʃ\( \Pi {\left(\frac{\mathrm{DBH}}{2}\right)}^2 \)) and the percentage of TTBA belonging to the dominant species of each plot (DomTBA). In all plots, DomTBA represented more than 50 % of TTBA (Table 1). Average DBH was similar among forest types (Table 1).

Fig. 1
figure 1

Mean monthly temperature and precipitation from the meteorological stations of Jadraque (a) and Guadalajara (b) during the study period (May 2011–May 2013). Data from the Ministry of Agricultura Alimentación y Medio Ambiente of Spain (http://www.aemet.es/es/serviciosclimaticos/datosclimatologicos/valoresclimatologicos and http://eportal.magrama.gob.es/websiar/SeleccionParametrosMap.aspx?dst=1)

Full size image
Table 1 Summary of the community structure of adult trees in the study plots
Full size table

Litterfall collection

Eighteen litterfall traps were distributed within each plot. To collect different plant materials, we used three types of traps: 1) Four traps made of mosquito net (1 mm mesh) forming a square (0.30 m2 surface) and fixed to the ground with spikes; 2) ten “pot” traps consisting of poly-vinyl chloride (PVC) circular pots (0.02 m2 surface) fixed on the ground with spikes and containing a nylon bag (<1 mm mesh); 3) four hanging traps consisting of nylon bags (<1 mm mesh) fixed to a wire circle of 0.18 m2 at 1 m above the ground. Overall, traps covered a total collection surface of 2.12 m2 per plot and were evenly distributed within each plot.

We collected the contents of each trap monthly from 15th June 2011 to 15th May 2013. Once in the lab, the contents of each sample were separated by tree species and plant litter fraction (leaves, bark and branches, flowers, fruits and unidentifiable debris). Then, samples were oven dried at 60 °C for at least 48 h and weighed (Balance Sartorius BP211D, 0.0001 g). We pooled all the data across the different traps for each plot to calculate monthly (g m−2 month−1) and annual (g m−2 year−1) litterfall for each fraction and for all litter.

Litter nutrients

The litter collected during the 1–3 month period of maximum litterfall in 2012 was reserved for nutrient analyses. We only analyzed the litter of the four target species and the main litter fractions (i.e., the “unidentifiable debris” fraction, which was on average 9 ± 2 % of the total litter, was not analyzed because its composition changes over time). We used only the samples collected by hanging and pot traps because they were not in direct contact with the soil, reducing the risk of fungal colonization. All samples from the same plot, species and plant litter fraction were pooled into a single compound sample and ground with a Cullati mill (0.7 mm particle size). N and P concentration were determined with a segmented flux auto-analyzer (Skalar San++) after digestion with H2SO4 and Cu-KSO4. To calculate N and P inputs per month, we used a single estimate of the N and P concentration of each plant organ for each selected species, then multiplied by the grams of litter of each plant organ supplied by the dominant species, plus any of the additional studied species if present (see Table 1). In each plot, annual N and P inputs were estimated as the sum of N and P inputs in each of the 12 months (June–May).

Soil collection and soil properties

From each plot, in March (spring) and December (autumn) 2012 we collected a mixed soil sample, from around each of the four hanging traps. Each mixed sample consisted of three soil cores (7 × 7 cm × 11 cm deep) collected and pooled together. Soil samples were carried to the lab in polyethylene bags, spread on filter paper and allowed to air dry. Afterwards, soil samples were sieved with a 1 mm mesh, homogenized, and kept at 4 °C until analysis.

To analyze total N and P concentrations, 0.5 g of soils were digested with H2SO4 and Cu–KSO4. The digested solution was analyzed with a segmented flux autoanalyzer (Skalar San++). We determined soil OM following a modification of the method described by Nelson and Sommers (1996). We weighed 5–10 g of soil previously dried at 105 °C for 48 h. Then the soil was burned at 400 °C for 2 h, cooled in a desiccator to room temperature and reweighed. OM (%) was calculated as the proportion of weight loss. We measured pH in a slurry of 20 g of soil with 40 ml of deionized water using a pH meter (micropH 2001, Crison Instruments, Barcelona, Spain) (Allen et al. 1974).

Statistical analysis

Since TTBA (m2) varied across plots, and the litter mass and nutrient inputs at plot level tend to increase with TTBA (Online resource 1), we expressed all litterfall variables per unit of surface area and TTBA (g m−4) to make them comparable across plots. Similarly, we expressed soil N, P and OM per TTBA unit (m2), since they also tended to increase with TTBA (Online resource 2). This approach is consistent with other studies that found significant positive correlations between litterfall and tree size (Negash and Starr 2013) or between litterfall and TTBA (Williams and Wardle 2007; González et al. 2010).

For each plant fraction, differences across species in N and P concentrations were assessed by one-way ANOVA followed by Tukey HSD test. Data were log or square-root transformed when appropriate to attain normality and/or homoscedasticity. To assess differences in the percentage of each litter fraction, the total and per fraction annual litter mass and N and P inputs across forest types and years, we used linear mixed models with forest type and year as fixed factors and plot nested within forest type as a random factor. After 3 months of sampling, one P. alba plot (PA3, Table 1) had to be moved to another location because of a timber harvest in the original plot. This plot therefore is missing data for the first 3 months. The model using the interaction term between year and forest type did not significantly differ (P > 0.05) from the model without the interaction, so we did not include the interaction.

To compare differences in total and per-fraction monthly input of litter mass, N and P produced by the dominant species, we used linear mixed models with dominant species and month as fixed factors and plot nested within dominant species as a random factor. We concluded that the temporal patterns of litterfall, N or P varied across species when the interaction term “dominant species x month” was significant. The model using the interaction term between year and dominant species did not significantly differ (P > 0.05) from the model without the interaction, so we did not include this interaction.

Differences in soil N, P, OM and pH across forest types were analysed using generalized linear mixed models (family = gamma) (Bolker et al. 2009), with forest type and season (March vs. December) as fixed factors and plot nested within forest type as a random factor. Tukey’s all pairwise comparisons were conducted in R package with the glht function in the software package multcomp (Hothorn et al. 2008). We used R package 2.14.0 (R Development Core Team 2011) for all data analyses.

Results

Annual quantity and dynamics of litterfall

In the study plots, the different litter fractions included flowers (6 % of the total annual litter), fruits (10 %), leaves (69 %), bark and branches (16 %), and unidentifiable debris (9 %), with no significant differences among forest types (F3;21 < 0.83, P > 0.05 in all cases).

The peak of flower abscission in the exotic trees occurred in June, 2–3 months later than in the natives (April for P. alba and March–April for F. angustifolia) (Fig. 2a). P. alba fruits fell in May and June, but the remaining species extended fruit dispersal throughout the year (Fig. 2b). The shedding of bark and branches did not show any clear temporal pattern (Fig. 2c). All species showed their main peak of litterfall during autumn (October–December), coinciding with the period of leaf shedding of deciduous trees (Fig. 2d and e). However, exotic species showed extra litterfall peaks, which were absent in the natives (significant interaction between dominant species and month, F69, 243 = 2.13, P < 0.0001) (Fig. 2e). R. pseudoacacia showed additional peaks in summer (August 2011 and September 2011; Fig. 2e) corresponding to a late-summer leaf abscission (Fig. 2d) and another extra peak in late spring (June 2012) due to the shedding of flowers and fruits (Fig. 2a and b). A. altissima also showed a late-spring peak (June 2011) due to early leaf shedding (Fig. 2d and e). Flowers represented a large proportion of the litter at their time of maximum floral abscission; flowers made up 34 % of the total litter for A. altissima, 43 % for R. pseudoacacia, 56 % for F. angustifolia and 48 % for P. alba during the month of maximum flower fall for each tree species. The proportion of total annual litter fallen in summer 2011 (July to September) was 47 % in forests invaded by R. pseudoacacia but only 24 % and 12 % in forests dominated by F. angustifolia and P. alba, respectively. The late-spring peaks of A. altissima and R. pseudoacacia accounted for 21 and 16 % of total annual litter, respectively. By contrast, this proportion was less than 10 % in non-invaded forests.

Fig. 2
figure 2

Mean flower (a), fruit (b), bark and branch (c), leaf (d) and total litter (e), provided by the dominant species (Ailanthus altissima, Robinia pseudoacacia, Fraxinus angustifolia and Populus alba) in each month (from Jun 2011 to May 2013). Arrows represent litterfall peaks from the invasive species almost absent in the natives. Standard errors were not shown for better clarity

Full size image

There was no significant difference among forest types either in the total annual litter mass or in the annual litter by fractions (flowers, fruits, leaves, bark and branches), but forests invaded by R. pseudoacacia showed a non-significant tendency to produce more litter than the other forests (1.4–1.7 times more; Table 2). The annual leaf litter was significantly greater in the first than in the second year in all forest types (Table 2).

Table 2 Mean (± SE) total and per fraction (flowers, fruits, leaves and bark and branches) annual litter (g m−4 year−1) collected in different forest types (Ailanthus altissima, Robinia pseudoacacia, Fraxinus angustifolia and Populus alba forest) and the F and P values for linear mixed models assessing the effects of forest type and year on annual litterfall variables
Full size table

Litter nutrients

P. alba litter showed low N and P concentrations in most fractions (Tables 3 and 4). The remaining species had similar N concentrations in their different fractions, except R. pseudoacacia fruits, which had the highest N and P concentrations (Tables 3 and 4). F. angustifolia had the highest P concentration in bark and branches of all studied species, and A. altissima had the highest P concentration in flowers (Table 4).

Table 3 Mean (± SE) of nitrogen (N) concentration contained in flowers, fruits, leaves, and bark and branches of each species (Ailanthus altissima, Robinia pseudoacacia, Fraxinus angustifolia and Populus alba) assessed at the time of the maximum abscission in 2012 for each plot
Full size table
Table 4 Mean (± SE) of phosphorus (P) concentration contained in flowers, fruits, leaves, and bark and branches of each species (Ailanthus altissima, Robinia pseudoacacia, Fraxinus angustifolia and Populus alba) assessed at the time of the maximum abscission in 2012 for each plot
Full size table

The temporal pattern of N and P input differed among species (significant interaction between forest type and month; F69,243 = 2.22, F69,243 = 2.04 for N and P inputs, respectively; P < 0.0001). The additional summer leaf peak of R. pseudoacacia resulted in an additional peak of N but not of P (Fig. 3a and b), because R. pseudoacacia leaf litter had low P concentration (Table 4). The additional late-spring peaks of A. altissima and R. pseudoacacia resulted in additional peaks of N and P (Fig. 3a and b).

Fig. 3
figure 3

Mean total phosphorus (P) (a) and nitrogen (N) (b) inputs estimated from total litter produced in each community (Ailanthus altissima, Fraxinus angustifolia, Populus alba and Robinia pseudoacacia) per month (from Jun 2011 to May 2013). Arrows represent litterfall peaks from the invasive species almost absent in the natives. Standard errors are not shown for better clarity

Full size image

Total N and P inputs did not significantly differ among forest types because of the high variability, although average N inputs were 1.7–2.2 times higher in invaded than in non-invaded forests (Tables 5 and 6). Annual N inputs from fruits were marginally different among forest types, tending to be greater in forests invaded by R. pseudoacacia than in the other forests (Table 5). However, annual N inputs from other litter fractions did not significantly differ among forest types. Due to greater quantities of leaf litter, both N and P inputs from leaf litter were significantly greater in the first year than in the second year (Tables 5 and 6).

Table 5 Mean (± SE) nitrogen (N) inputs from total and per fraction (flowers, fruits, leaves, and bark and branches) annual litter (g m−4 year−1) collected in different forest types (dominated by Ailanthus altissima, Robinia pseudoacacia, Fraxinus angustifolia and Populus alba) and the F and P values for linear mixed models assessing the effects of forest type and year on annual N input variables
Full size table
Table 6 Mean (± SE) total and per fraction (flowers, fruits, leaves, and bark and branches) annual litter (g m−4 year−1) collected in different forest types (dominated by Ailanthus altissima, Robinia pseudoacacia, Fraxinus angustifolia and Populus alba) and the F and P values for linear mixed models assessing the effects of forest type and year on annual P input variables
Full size table

Soil properties

Forests invaded by R. pseudoacacia had the highest OM and soil N concentrations, while P. alba forests had the lowest soil OM, N and P (Fig. 4). A. altissima and F. angustifolia forests showed intermediate and similar values of soil OM and N (Fig. 4a and b). Soil OM and N were greater in autumn than in spring (Fig. 4a and b). However, soil P tended to decrease from spring to autumn in A. altissima and F. angustifolia forests and increase in P. alba and R. pseudoacacia forests (Fig. 4c). Soil pH did not significantly differ among forest types, but it was higher in autumn than in spring in all forests except in F. angustifolia forests, where it was higher in spring than in autumn (Fig. 4d).

Fig. 4
figure 4

Mean (± SE) soil organic matter (OM) (a), nitrogen (N) (b) and phosphorus (P) concentration (c), per unit of total tree basal area (m2), and pH (d) in two seasons (spring and autumn 2012) in Ailanthus altissima, Robinia pseudoacacia, Fraxinus angustifolia and Populus alba forests. Different letters mean significant differences among forest types (Tukey HSD test). The significance of forest type, season and their interaction is shown in each chart (NS No significant; * P < 0.05; ** P < 0.01: *** P < 0.001)

Full size image

Discussion

In our study, forests invaded by R. pseudoacacia had the greatest soil OM, N and P, while native forest dominated by P. alba showed the lowest values. This result is consistent with patterns of the N and P content of litter, which was greatest in R. pseudoacacia and lowest in P. alba in most litter fractions, even though we failed to find differences in total N or P inputs or in the quantity of litter among forest types. The capacity of R. pseudoacacia to fix N, and its low N use efficiency (González-Muñoz et al. 2013) may explain the higher N concentration of some of its senescent tissues. Additions of N-rich litter from trees have been reported to increase soil N (Johanson 1995). Litter from different plant organs of both invasive species had higher N content than that of P. alba, which likely led to greater soil N in invaded than in P. alba soils. In addition, greater quantities of litter usually increase total soil N and OM (Martin et al. 2009), and we found a (non-significant) trend for R. pseudoacacia to produce more annual litter (up to 44 % more) and N inputs (up to 56 % more) than the remaining species (Tables 2 and 5). This trend likely reflects the high growth rates and poor N use efficiency of R. pseudoacacia (González et al. 2010; Cierjacks et al. 2013; González-Muñoz et al. 2013; Castro-Díez et al. 2014).

Soil OM depends not only on the quantity and quality of litter inputs, but also on the decomposability of the litter, the root exudates of organic carbon, the persistence of the different OM compounds, and the soil microbial biomass and activity (Schmidt et al. 2011). In the case of R. pseudoacacia, previous studies in Central Spain have pointed to the potential of R. pseudoacacia to accumulate recalcitrant OM and N in the soil due to its high lignin content and the consequent slower leaf litter decomposition than coexisting native species (Castro-Díez et al. 2009, 2012). Moreover, flowers from the exotic species fell in June, at the beginning of the dry season (2–3 months later than those from the natives), when microbial and plant activity declines due to water shortage. These arid conditions may slow down decomposition and the uptake of the derived nutrients, enlarging the period when OM and nutrients remains in the soil.

The summer and late-spring inputs from the exotic species were mostly due to leaf abscission and probably reflect the adjustment of these species to reduce water loss during the arid period. Neither A. altissima nor R. pseudoacacia evolved in regions with strong summer water stress, and they may have fewer adaptations than native species to tolerate water shortage. In fact, the time of the additional leaf litter peaks displayed by exotics match the times of lowest precipitation and highest temperature (Figs. 1 and 2d). The leaf litter that falls in summer could decay more slowly, due to water shortage during the summer. This would enlarge the time when organic nutrients remain in the soil and contribute to the increment in soil nutrients and OM in invaded forests (Kirschbaum 2010).

Given that N and P concentration may vary in plant tissues across the year, measuring nutrients only at the peak of each fraction fall may introduce errors into the calculation of the annual input of nutrients in the soil, particularly in the case of R. pseudoacacia and A. altissima, which showed early peaks of leaf abscission (Fig. 2). Given that pre-senescent leaf litter can have greater nutrient content due to a more limited nutrient resorption (Killingbeck 1996), our calculation may have underestimated the amount of N and P input to the soil in the sites invaded by the two exotic species. In the remaining species, we believe that the potential error was lower because 1) the fraction of flowers, fruits and leaves that fell out of the period of maximum fall was lower; 2) P. alba leaves have been found to have similar N concentrations whether they fall in summer or autumn (González-Muñoz et al. 2013), and 3) annual changes in N and P of branches of Mediterranean phanaerophytes in general have been found to be quite low (Milla et al. 2005).

The additional litterfall peaks produced in late spring by the invasive species may generate additional soil litter layers, which may act as physical barriers for germination, establishment and growth of species that shed their seeds in spring and do not form soil seed banks (Facelli and Pickett 1991). This is the case for P. alba, whose seeds disperse in May–June and are viable for only a few weeks (Prada and Arizpe 2008). Germination of small seeds, such as those of P. alba, may be more hindered by litter layers than that of bigger seeds, such as those of A. altissima and R. pseudoacacia (Facelli and Pickett 1991; Seiwa and Kikuzawa 1996). Therefore, the re-colonization of invaded forests by the native P. alba could be hindered by the accumulation of spring litter layers from exotic trees. At the same time, seed germination of these exotics species could be less affected by their own summer and spring litter since they shed their fruits throughout the year (Castro-Díez et al. 2014).

Additional litter inputs with different quality from invasive species may affect abundance and composition of detritivore communities, which are in synchrony with litterfall processes (Takeda 1987; Ponge et al. 1993). Some detritivores may not be adapted to feed on recalcitrant leaf litter (Bastow et al. 2008), such as that of R. pseudoacacia, which may lead to a longer persistence of R. pseudoacacia litter layers. However, the abundance of some arthropod species may also increase with additions of high quality litter (Cesarz et al. 2007), such as A. altissima litter with its high nutrient content and rapid decomposition (Castro-Díez et al. 2009, 2012). In fact, soils invaded by A. altissima host altered soil arthropod communities compared to non-invaded soils (Gutiérrez-López et al. 2014; Motard et al. 2015).

In conclusion, our study highlighted the potential of the invasive trees, A. altissima and R. pseudoacacia, to alter soil properties even when we did not find significant differences in the annual quantities of litterfall and nutrient inputs between invaded and native forests. R. pseudoacacia tended to produce more annual litter and N inputs than the remaining species, likely due to its high growth rate and poor N use efficiency. Forests invaded by R. pseudoacacia had higher concentrations of soil nutrients and OM than native riparian forests. These soil variables were also elevated in A. altissima forests, but only when compared with P. alba forests. Finally, invasive trees altered litterfall dynamics by producing additional litterfall peaks and N and P inputs in summer and late-spring, which may modify soil nutrients, the activity of detritivores, as well as seed germination success.