Seed bank composition in a secondary forest in the Lower Delta of the Paraná River ( Argentina )

Th e native forests once occupying the coastal levees of the Lower Delta islands of the Paraná River (in Argentina) have been replaced by commercial forests. Many of those forests have been abandoned, resulting in secondary forest formation that is subject to numerous invasive exotic species. A priori observations suggest that successional trends do not lead to recovery of the original forest. Th e aim of the present study was to analyze the role of invasive exotic species in these environments and the likelihood of recovery of the original forest cover. Th erefore, we examined the composition of the seed bank and of the standing vegetation, as well as the population structure of tree species. Secondary forests are described as being in an intermediate successional stage, with few exotic species (mainly trees) but with a high abundance of standing vegetation and seed banks. Th ese exotic species will likely continue to successfully predominate in diff erent forest strata over the next stages. Because of the low density of native tree species, it is diffi cult to predict their future persistence. In conclusion, ecological restoration strategies will be needed in order to increase native tree species richness and biodiversity of the forests in the Lower Delta of the Paraná River.


Introduction
Seed banks contain viable propagules stored above or below the soil (Th ompson & Grime, 1979).Several studies have analyzed the importance of diff erent environmental factors, as well as the size and composition of soil seed banks, in various ecosystems (Bakker 1989;Bertiller 1992;Albrecht & Pilgram 1997;Jalili et al. 2003).Th e role that seed banks play in secondary successions has been studied since the mid-20th century, and it is considered a key factor to ensure the regeneration of plant cover following disturbances (Baker 1989;Chapman & Chapman 1999;Luzuriaga et al. 2005).In the last few decades, there has been growing interest in soil seed bank composition in forest ecosystems (Milber 1995;Holmes & Cowling 1997;Vallés et al. 2005), because of the potential of seed banks to restore original forests (Onaindia & Amezaga 2000;Urretavizcaya & Defoseé 2004), as well as their strong infl uence on the composition and diversity of ecosystems aft er the commercial harvesting of timber (Roberts 1981;Kalesnik 2001;Decocq et al. 2004).
Th e degree of association between the composition of seed banks and that of the standing vegetation can be used to predict the recovery trends of plant communities (Holmes & Cowling 1997;Albrecht & Pilgram 1997;Decocq et al. 2004).In mature forests, the composition of the seed bank usually diff ers from that of the canopy vegetation (Arévalo & Fernández-Palacios 2000), the seed banks exhibiting a higher proportion of early successional or pioneer species (Hanlon et al. 1998;Arévalo & Fernández-Palacios 2000) than of mature forest species (Fenner 1985), because most shade-tolerant species do not develop persistent seed banks (Onaindia & Amezaga 2000).
Ecosystem restoration potential has also been studied solely by analyzing seed bank composition (Middleton 2003;Decocq et al. 2004).Seed banks in areas subjected to anthropic modifi cations can be composed almost entirely of invasive exotic species, decreasing the chance that native species will recover.For instance, the seed banks of mature deciduous forests subjected to management are composed mainly of exotic species, which constitute over 80% of the total seed density (Leck 1989).
In the study area of this work (island levees of the Lower Delta of the Paraná River), the original forest vegetation had been almost entirely replaced by commercial forests of willow and poplar (Burkart 1957;Kalesnik 2001;Vallés et al. 2005).In the last few decades, many of those forests were abandoned because of socioeconomic difficulties aff ecting the region (Kalesnik & Malvárez 2003;Vallés et al. 2005).Despite the fact that certain areas have long been abandoned (for over 30 years), the original forest has not been able to recover.Instead, a mixed secondary forest dominated by invasive exotic species has spontaneously developed.Many of these species have been introduced for ornamental purposes, such as the Asian species Japanese honeysuckle (Lonicera japonica), glossy privet (Ligustrum lucidum) and Chinese privet (Ligustrum sinense), as well as the European elm-leaf blackberry (Rubus ulmifolius).Th ese species oft en coexist with low densities of native tree species seedlings and saplings, such as laurel (Nectandra falcifolia; Kalesnik 2001), "canelón" (Rapanea lorentziana; Kalesnik & Malvárez 2003) and Mexican olive (Blepharocalyx tweediei; Kalesnik et al. 2009).Th is secondary forest can be considered a "neo-ecosystem", a term coined by Morello et al. (2000) to characterize semi-natural open or wooded areas where dominant invasive exotic species are accompanied by native species.
Th e main goal of the present study was to analyze the role of invasive exotic species in levee forests of the Lower Delta and the likelihood of recovery of the original forest, locally referred to as the Monte Blanco forest.Th e seed bank composition was estimated by diff erentiating between invasive exotic species and native species.Regeneration trajectories for the recovery of communities were defi ned by analyzing the similarities between the seed bank and the standing vegetation species.
Our main hypothesis was that the invasive process, occurring in the Lower Delta of the Paraná River at a regional level, is evidenced by the seed bank composition of secondary forests.As a consequence, the dominance of invasive exotic species would continue and the probability of recovery of the original Monte Blanco native species would be low.

Study area: the Buenos Aires Province Lower Delta
Th e Lower Delta islands are located along the fi nal stretch of the Paraná River Delta, at the point where the Paraná River divides into two main distributaries, Paraná Guazú and Paraná de las Palmas (Figure 1).Th e area extends from the city of Ibicuy (59°20'W; 33°42'S) to the mouths of the Paraná Guazú (58°24'W; 34°S) and Luján Rivers (58°32'W; 34°26'S) at the Río de la Plata.Th e islands form a growing delta over the Río de La Plata estuary (Iriondo & Scotta 1979;Parker & Marcolini 1992) covering a total area of 2,071.06km 2 (Latinoconsult 1972).
Th e temperate climate is characterized by rainfall throughout the year and an average temperature above 22°C during the warmest month.Th e average annual temperature is approximately 17°C, ranging from 6°C to 30°C (Servicio Meteorológico Nacional 1980).Th e hydrological regime of the area is infl uenced by the Paraná and Uruguay Rivers and the Río de la Plata estuary (Latinoconsult 1972;Mujica 1979).However, the Paraná River regime is determined mainly by tropical and subtropical rainfall in its upper basin, exhibiting seasonal variations in water levels, which begin rising in September and reach peak levels in March (DNCP 1983).
Th e main hydromorphic soil types in the levees are humic and subhumic gley soils, as well as new alluvial soils (Bonfi ls 1962).According to the United States Department of Agriculture soil taxonomy, these soil types correspond to poorly developed mollisols under an aquic regime.
Sampling was performed in a secondary forest located on the banks of the Carabelas River (Figure 1).Th e site is an open Salicaceae forest that has not been logged for 20 years and in which the original vegetation has been replaced by exotic tree species and, to a lesser extent, by native species.Although our data were obtained during 2001, it should be noted that subsequent studies have not shown any variation from these data (Kalesnik et al. 2009).

Seed bank
Four 10 × 10 m plots were randomly set at the site in February of 1999.Forty circular soil core samples (10 cm in diameter and 5 cm in depth) were collected from the topsoil layer, ten samples from each plot (two from each corner and two from the center of the plot).Th e soil core samples were collected by inserting a PVC tube into the ground.Th e samples were immediately stored into plastic bags and labeled.Seedling emergence and direct seed counts were both used to determine seed bank structure and composition.For each plot, one corner sample and one center sample were used for each method, yielding a total sampled surface area of 1570 cm 2 .

Seedling emergence
Twenty soil samples were cold stratifi ed by storing them at 5°C for two weeks, in order to break the typical dormancy of many temperate climate seeds (Schneider & Sharitz 1986;Houle & Phillips 1988).In March, the samples were allowed to germinate in plastic trays inside a greenhouse.Th e trays contained inert sand (2.0 cm under the samples and 0.5 cm over) to maintain humidity conditions.Soil samples were spread over the trays to maximize the likelihood of seedling emergence.We also assembled four control trays containing only inert sand.Th e trays were rotated twice a week, by row and by column, in order to homogenize the environmental variables in the greenhouse.For each tray, the emerging seedlings were counted and then transferred to plastic pots outdoors for identifi cation.

Direct seed count
Th e remaining 20 samples were sift ed twice, aft er which the seeds were counted.Large plant fragments were discarded during the fi rst sift ing.Seeds were then separated by species using a 1-mm mesh sieve.Th is technique was performed mainly to identify L. sinense seeds, the principal invasive species in the area (Kalesnik et al. 2009), which do not germinate using the previous method (Vallés et al. 2005) and whose quantifi cation was assumed to be of importance for this study.
Th e numbers of individuals of each tree species for each pre-determined age class were counted.Age class distribution considered three groups: adults (diameter at breast height > 5 cm), saplings (diameter at breast height < 5 cm, height >1.30 m) and seedlings (height < 1.30 m).Saplings and seedlings were sampled in a smaller 2 × 2 m plot at the center of the larger (10 × 10 m) plot.
Patterns in the population structure and spatial distribution of the tree species-frequent reproducers; recent incorporation or recent invaders; and nomadic or relictual individuals-were identifi ed on the basis of previous data (Knight 1975;Saxena & Singh 1984).Th e future composition of the secondary forest was predicted based on the relative density of seedlings, saplings and adult trees found in the standing vegetation.Relative tree species density in the seed bank was also taken into account for the predictive analysis.

Seed bank
Seed bank species richness was calculated as the sum of species identifi ed on the basis of the seedling emergence and direct seed count techniques.Similarly, in our comparative analysis of seed density in the bank and seed bank species richness for each life-form type and origin, we used both methods.
Standing vegetation -Total species richness, as well as species richness discriminated by life form and origin, was estimated as the sum of the species identifi ed during sampling.Average species cover was estimated by determining the arithmetic mean for all plots.For comparative analysis, the cover for each life form and origin was calculated as the sum of all species belonging to the respective group and normalized to 100%.

Seed bank and standing vegetation comparison
Similarity between seed bank and standing vegetation composition in the secondary forest was estimated by the qualitative Sørensen similarity index (Magurran 1988).Similarity was also analyzed for each life-form type using the same index, although we considered species only according to their life form.For both analyses, calculation of the similarity index took into account the total number of species identifi ed in all four plots.

Tree species density
Relative densities of tree species identifi ed in the seed bank and in the secondary forest structure were analyzed in order to determine the possible future composition of the canopy.Th us, it is possible to study whether dominant species in each forest stratum are likely to sustain their dominance over time or if they will be replaced by other species.For the comparison between native and exotic species, total density was calculated as the sum of standing vegetation, seedling emergence and seed count in the seed bank.

Statistical analysis
Th e data did not meet the normality assumption and therefore required square-root transformation [y = √(x + 0.5)].Because the data distribution was still non-normal aft er this transformation, a non-parametric Mann-Whitney test was performed for all variables to compare native and exotic origins.Th e XLSTAT soft ware, version 7.5.2(Addinsoft , Paris, France) was used for all analyses.

Seed bank
From the 1570 cm 2 of secondary forest soil sampled, a total of 369 seeds were germinated, corresponding to 52 diff erent species and equivalent to 2349 germinated seeds/ m 2 (Table 1).Although L. sinense seeds did not germinate in any of the samples, high densities of that species where found in all samples by direct count, totaling 197 seeds (1254 seeds/m 2 ).If the results of both methods are considered, the species richness in the seed bank was 53 species: fi ve tree species (three of which were exotic); 47 herbaceous species-38 broadleaf (four exotic), seven graminoid (two exotic); and two equisetoid (both native); and one native creeper species.On average, 28.8 species germinated per plot, with an average density of 3603 seeds/m 2 , including L. sinense seeds quantifi ed by direct count (Table 2).
As can be seen in Table 2, the number of native species was signifi cantly higher than was that of exotic species (p=0.019).However, seed density did not diff er by species origin (p>0.05).All types of native herbaceous species outnumbered the exotic species (p<0.05) and their seed density was similarly higher (p<0.05).Among these, Solanum deltaicum showed the highest number of germinated seeds.Tree species and broadleaf herbaceous species accounted for the majority of seeds in the bank.Although species richness did not vary by species origin (p>0.05), the density of exotic tree species was signifi cantly greater than was that of native tree species (p=0.020),particularly due to L. sinense.Creeper species showed the lowest richness and seed density values and did not vary by species origin (p>0.05).No seedling emergence was observed in the four control trays.

Comparison between standing vegetation and seed bank
Th e total species richness in standing vegetation was 40 species (Table 1): ten tree species (fi ve of which were exotic); two shrub species (one exotic); 24 herbaceous species-17 broadleaf (one exotic), one native equisetoid, and six graminoid (fi ve native); and four creepers (three native).
On average, 18.5 species were found per plot (Table 3).Although native species outnumbered exotic species (p=0.019), the latter showed a higher mean cover value (p=0.021).Among the broadleaf herbaceous species, number and cover were higher for native species than for exotic species (p=0.018 for both).For the remaining life--form types, the number of species did not diff er by origin.Nevertheless, the mean cover was higher for exotic trees (predominantly L. sinense and Fraxinus sp.) and creepers (predominantly Lonicera japonica) than for their native counterparts (p=0.021 for both).
For all species in the seed bank and standing vegetation, the Sørensen similarity index was 38.7%.When diff erentiating between life forms, we found that similarity was 0% for shrubs and creepers, 29% for broadleaf species, 53.3% for trees, 66.7% for equiseta, to 76.9% for graminoid species.
Seed density was higher for exotic tree species than for native tree species (p=0.020),mainly because 96.6% of the seeds were of L. sinense, compared with only 1.5% for native tree species.Among seedlings and saplings, relative density was also dominated by L. sinense with a signifi cant smaller proportion of native tree species (p=0.046).When considering adults, the density of exotic tree species was signifi cantly higher than for native species (p=0.050),once again because of L. sinense dominating the middle stratum of the forest (Table 4).
By integrating the previous results and analyzing the population structure patterns of the diff erent tree species, Fraxinus sp. can be regarded as a frequent reproducer, seedlings and saplings outnumbering adults (Table 4).Seeds of Fraxinus sp. were also found in the seed bank, which suggests continued dominance of the canopy of the secondary forest by this species.
The remaining tree species composing the canopy, both exotic and native, showed a relictual pattern, with no seedlings or saplings found in the population structure.Th e only exceptions were Morus alba (exotic) and Sapium haematospermum (native), which were identifi ed in the seed bank (Table 4).
With regards to species in the middle stratum of the secondary forest, L. sinense corresponded to a frequent reproducer that has a relatively higher proportion of seedlings     and saplings than of adult plants.Th is short tree species (3-5 m) showed a distinctive high density for all population stages considered and is therefore likely to dominate the middle stratum of the forest in the future.

Discussion
Th e present work represents the fi rst description of the type of seed banks that species of this secondary forest can develop, allowing a better understanding of the role of exotic species in such environments.Particularly, L. sinense can develop transient or type II seed banks during the winter months (Th ompson and Grime, 1979), with germination occurring only in the spring.Our sample collection period (February 1999, which is fall in the southern hemisphere) and the fact that seeds of this species lose viability relatively quickly is in agreement with the aforementioned assumption.
Contrarily, Fraxinus sp.forms seed banks classifi ed as type III (sensu Th ompson & Grime 1979), also known as type B (sensu Nakagoshi 1985), in which seeds germinate immediately aft er being released during the fall.A small fraction of seeds permanently incorporate into the seed bank, which would account for the results of our seed emergence analysis.Fraxinus sp. also show traits of the Fagus type, as defi ned by Nakagoshi (1985), which is a shade--tolerant, deciduous tree species with large seeds and type A or B seed banks (types II or III according to Th ompson & Grime 1979).
Th e occurrence of R. ulmifolius in the seed bank was noteworthy, despite being absent from the standing vegetation.Th is case is compatible with persistent seed bank types (type IV according to Th ompson & Grime 1979), in which a low proportion of seeds germinate upon being dispersed and the remaining vast majority are incorporated into the seed bank.Th is strategy is typical of dominant species in disturbed environments (Th ompson 1998).In analyzing successional trends, Nakagoshi (1985) described the Rubus type as a functional group composed of shade-intolerant shrubs with a short life-span and seed dispersal by birds.In the event of disturbances (e.g., weeding), R. ulmifolius can germinate and invade forests.
In the forest studied, the species similarity of the seed bank and standing vegetation (38.7%) was similar to that reported by Hopfensperger (2007) for temperate and tropical forests (31% on average for both).Th is degree of similarity represents an intermediate successional state, because mature forests usually consist of small seed banks composed of species absent from the standing vegetation (Nakagoshi 1985;Th ompson 1998).
Th e exotic tree species, acting as invasive species in the secondary forest, showed a population structure that is compatible with a frequent reproducer pattern and a prominent presence in the seed bank.Th e results of our seed bank analysis underscore the notion that exotic species will continue to dominate all strata of the secondary forest in future generations.Th e colonization success of Ligustrum species is most likely attributable to seed dispersal by birds, as reported by Montaldo (1993) for similar regions.
When we compared our results with those obtained by other authors (Gunther et al. 1984;Schneider & Sharitz 1986;McGraw 1987;Titus 1988;Goodson et al. 2002;Nicol et al. 2003;Leck & Schütz 2005;Richter & Stromberg 2005), we found that the species richness of the seed bank in the secondary forest we studied here was high (53 species), particularly in broadleaf herbaceous life forms.Our seed bank also showed a high estimated seed density (3603 seeds/ m 2 ), not only due to the herbaceous species but also due to trees.Although tree species richness in the seed bank was low, we observed considerable tree seed density (1299 seeds/ m 2 ), mainly due to three exotic tree species.
Th e invasive nature of exotic species was not refl ected in the total seed bank richness, because native species outnumbered exotic species, mainly because of a high proportion of native herbaceous species.Th is notwithstanding, the high abundance of exotic tree seeds compensated for the total seed densities when we discriminated between species by their origin.
In summary, the secondary forests in the study area showed a low number of exotic species, mainly trees, but there was high species abundance, in standing vegetation as well as in soil seed banks.One explanation for this scenario is the ubiquity of L. sinense.Th is species can be responsible for severely reducing species richness and tree regeneration (Merrian & Feil 2002), because of shading and allelopathic eff ects, a high degree of phenotypic plasticity and its strong tendency toward asexual reproduction (Grove & Clarkson 2005;Pokwinski 2008).Th e signifi cantly low density of tree species native to the original forests, both in standing vegetation and seed banks, will likely lead to their disappearance from those forests.Even the few native species composing the canopy exhibited a relictual pattern and are therefore not expected to be components of the canopy in the future.
Th e secondary succession observed in these abandoned commercial forests can be considered a natural and spontaneous passive restoration process (Rey Benayas & Camacho Cruz 2004) that will not result in the regeneration of the original Monte Blanco forest.Th erefore, strategies focused on active forest restoration measures (Maestre et al. 2001;Rey Benayas 2005) are needed in order to increase biological diversity and native tree species richness in the secondary forests of the Lower Delta.

Figure 1 .
Figure 1.Study area: Lower Delta of the Paraná River (Argentina).Sampling site is indicated by a dotted circle.

Table 1 .
Seed density in the seed bank and standing vegetation cover in four plots within a secondary forest of the Lower Delta of the Paraná River (Argentina).

Table 2 .
Mean species richness of native and exotic species density in the seed bank of a secondary forest of the Lower Delta of the Paraná River (Argentina), by life form.*

Table 4 .
Population structure (seed density in the seed bank, as well as the spatial distribution of seedlings, saplings and adults) of tree species in a secondary forest of the Lower Delta of the Paraná River (Argentina).*

Table 3 .
Mean species richness of native and exotic species density in standing vegetation of a secondary forest of the Lower Delta of the Paraná River (Argentina), by life form.*