EFFECT OF THE REMOVAL OF REGENERATING INDIVIDUALS ON THE SEEDLING BANK DYNAMICS IN PLANT COMMUNITIES

ABSTRACT The use of forest seedling banks is recommended for producing seedlings of species essential for specific forest typologies for which propagation poses challenges. However, given the lack of understanding of how human intervention, such as the removal of regenerating individuals from the forest, influences the composition and structure of the plant community, there is a pressing need for further research. Our primary objective here was to identify the effects of seedling removal on natural regeneration in forest communities and to characterize seedling bank dynamics three years after anthropogenic intervention. A randomized block design was used for the experiment, consisting of five removal intensities (0, 25, 50, 75, and 100% removal), with evaluations conducted every three months for three years. All individuals of the shrub-arboreal component between 5-55 cm heights were measured. Collected data were analyzed to determine the impacts of the removal of individuals and the influence of environmental elements on the seedling community. We found that the shrub-arboreal component of the forest community exhibited marked resilience three years following the removal of regenerating individuals. Removal of up to 50% of regenerating individuals does not appear to interfere with plant community dynamics, suggesting species with a high density of individuals in the seedling bank are resilient to the impacts of human intervention. Although climatic seasonality also affects seedling bank dynamics in plant communities, this effect depends on the level of human intervention in an area and, therefore, has been omitted from the analysis.


INTRODUCTION
Increasing anthropogenic activity has triggered signifi cant changes in environmental conditions globally, threatening the survival of many native species (Kerr and Currie, 1995).Forest and soil degradation are limiting barriers to the sustainable development of terrestrial ecosystems (Liu et al., 2012) and are among the primary causes of biodiversity loss worldwide (Catterall et al., 2012).
Planting native forest species has yielded satisfactory results in restoring biodiversity (Chazdon, 2008;Brancalion et al., 2019).The primary goal of reforesting eff orts using native tree species is to slow or reverse biodiversity loss by counteracting obstacles to natural regeneration (Catterall et al., 2008).As Rodrigues et al. (2009) note, successful restoration of tropical ecosystems requires the use of plant species that are pre-adapted to the regional environmental conditions.However, seedlings of native forest species are often diffi cult to obtain, especially those that commonly occur in the fi nal successional stages (Turchetto et al., 2016), mainly due to diffi culties in obtaining seeds and a lack of technology for producing many native forest species.Such defi ciencies often lead to lowdiversity restoration of uniform plantations, which can compromise the future sustainability of restored forests, as is often the case for areas within fragmented landscapes (Viani and Rodrigues, 2008).
Previous research has led to recommendations that seedlings produced by naturally regenerating individuals in tropical forests be removed and replanted in nurseries to generate additional seedling stock (Viani and Rodrigues, 2008;Viani et al., 2012;Turchetto et al., 2016), allowing seedlings to become established before being transferred to natural sites.
An important advantage of this approach is that large quantities of seedlings of species from forest types that are diffi cult to propagate can be readily produced (Turchetto et al., 2016).
In Brazil, however, seedling removal is not addressed in current environmental legislation due to a scarcity of research on its impacts on forest communities.Because of this defi ciency, at present, this approach can only be used as a mitigation measure for the production of seedlings in cases where formal authorization has been granted for forest containment or replacement.In addition, diff erent regeneration strategies depend on the type of anthropic disturbance (Denslow, 1980).Moreover, strategies in subtropical ecosystems largely depend on seasonal climate, which infl uences plant community composition and structure (Brokaw and Scheiner, 1989).Thus, research on the eff ects of modifi cation or management of natural seedling banks must be conducted to prevent irreversible damage to the diversity, productivity, and connectivity of populations, communities, and the ecosystem as a whole (Viani and Rodrigues, 2008).
In the present study, our primary goal was to identify the impacts of seedling removal on the natural regeneration of a forest community and characterize the seasonal dynamics of the seedling bank three years after anthropic intervention.We sought to address the following two questions: a) Do high intensities of seedling removal compromise the dynamics of natural regeneration?and b) Is removal intensity linked to species density?

Study area
The study was conducted in a remnant of the Subtropical Seasonal Forest (29°27'14.71"S and 53°18'17.86"W) in the extreme south of the Atlantic Forest Biome, in the central region of the state of Rio Grande do Sul, Brazil.The soil in the area is classifi ed as Regolithic Neosol (EMBRAPA, 2013) and the climate is subtropical, with an average annual rainfall of 1560 mm (Alvares et al., 2013) and well-defi ned seasons (Figure 1).The forest fragment we focused on is characterized as a secondary forest in an advanced stage of succession, with the last human intervention occurring more than 50 years ago; little sign of this intervention can be seen today.

Experimental design and data collection
A randomized block design was used for the experiment, with treatments consisting of fi ve intensities of removal of individuals from the seedling bank (0, 25, 50, 75, and 100% removal).Treatments were distributed in 15 blocks subdivided into fi ve plots of 1 m × 2.5 m, for a total of 75 plots.Additional details about the experimental design are presented in Turchetto et al. (2018).
Seedlings were considered to be individuals of the shrub-arboreal component ranging in height from 5-55 cm.The initial evaluation (conducted in October 2013) identifi ed these plants based on the APG IV classifi cation system (The Angiosperm Phylogeny, 2016).All individuals meeting this criteria were marked with metallic plates in the initial evaluation, facilitating the removal of individuals via random draw.Determination of basic phytosociological parameters (density, frequency, dominance, and importance) allowed for the identifi cation of species with the highest densities in the seedling bank (Turchetto et al., 2017).
Following the initial evaluation, individuals were removed from each plot in accordance with the assigned removal intensities (0, 25, 50, 75, and 100% removal) in November 2013.Individuals were carefully dug up from the soil using a gardening shovel to minimize possible damage to the remaining seedlings.In all subsequent evaluations (i.e., from 3 to 36 months after removal), information pertaining to the number of individuals and species present in each plot was recorded, and assessments were performed every three months.

Data analysis
To assess the impacts of seedling removal on natural regeneration, an analysis of variance (ANOVA) was performed using generalized mixed linear models (GLM), with negative binomial distribution controlling for overdispersion and the likelihood-ratio test evaluated using the R packages 'MASS' (Venables and Ripley, 2002) and 'car', respectively.Pairwise comparisons of the estimated marginal means were then undertaken using the 'emmeans' package in R (Lenth et al., 2023).
The numbers of individuals and species between the initial evaluation (time zero) and subsequent evaluation periods were compared within each treatment using Dunnett's test to evaluate the resilience of the seedling community.
All analyses and graphics were performed in R Studio 4.0.0 (R Core Team 2020) at a 5% probability.Model fi t was checked by comparing the fi tted plot with a residual value plot, distribution in a QQ plot, and a histogram.

RESULTS
At 36 months, removal intensities below 50% of regenerating individuals presented the best results in terms of the numbers of individuals and regenerating species (Figure 2), whereas removal intensities equal to or greater than 75% had a negative eff ect on the density and richness of species present during natural regeneration.
Diff erences in the number of species observed between evaluation periods were similar to those observed for the number of individuals.Community richness was not restored under removal intensities of 75% (p = 0.007) and 100% (p < 0.0001) (Figure 4), whereas under removal intensities of 25% and 50%, variation was similar to that observed in control plots (0% removal).These fl uctuations corresponded to the emergence, recruitment, and mortality of some of the most abundant species, including Actinostemon concolor (Spreng.)Müll.Arg.and Psychotria leiocarpa Cham.& Schltdl.The reduction in individuals between evaluation periods was also associated with seasonality eff ects, particularly during summer, when temperatures are elevated and precipitation is less frequent (Figure 1).
For the most abundant species in the seedling bank, as determined by the number of individuals counted in each evaluation (Figure 5a) and removal intensity (Figure 5b), diff erent recruitment strategies were observed in response to the removal of regenerating individuals.A. concolor, Nectandra megapotamica (Spreng.)Mez, and P. leiocarpa were more abundant at the initial evaluation (time 0) than at 36 months following seedling removal.Of these, A. concolor was the only species equally abundant after three months of seedling removal as at the start.
Considering the eff ect of seedling removal on species with higher densities, A. concolor and P. leiocarpa were the only species that, 36 months after seedling removal, had higher numbers of individuals in plots where all seedlings were removed (100% removal).Individuals of Eugenia rostrifolia D.Legrand, N. megapotamica, and Trichilia claussenii C.DC. were reduced in number as removal intensity increased; however, at removal intensities of 25% and 50%, the numbers of individuals of these species were similar to those observed in the control plots (0% removal).Notably, the number of individual Sorocea bonplandii (Baill.)W.C.Burger et al. decreased by as much as 60% in treatment plots compared to controls, even under relatively low removal intensities (i.e., 25%) (Figure 5 b).

DISCUSSION
The results of our seedling removal experiment suggest that using seedling banks to produce seedlings from forest tree species for use in forest restoration eff orts may be sustainable.Seedling removal intensities of up to 50% did not compromise the characteristics or structure of the regenerating community, as evidenced by the fact that treatment plots subjected to these intensities remained similar to the control (0% removal).Other studies on diff erent forest typologies also found that low removal intensities (below 50%) do not interfere with plant community dynamics (Viani and Rodrigues, 2008;Turchetto et al., 2018).
However, the removal of 50% or more of seedlings reduced both the number of individuals and species richness, indicating that this level represents a threshold for the sustainability of regenerating individuals.As intensities rose, fewer species constituted the seedling bank, probably because of the need to maintain a "stock" with some species whose dispersion and/or regeneration is more diffi cult.This scenario demonstrates that removing individuals from seedling banks can signifi cantly impact some species, especially those with a low regeneration density.
Therefore, a thorough understanding of species dynamics in each habitat is urgently needed, including the causes, mechanisms, and factors that drive the natural regeneration process.However, at present, little is known about the regenerative capacity and population structure of the species that comprise Brazil's forests, especially in terms of their response to the impacts of anthropogenic activities (Omondi et al., 2016).
Of the most abundant species in our forest fragment, only seedlings of A. concolor were more abundant at the time of the second evaluation (i.e., the fi rst following removal, December/2013) than at the time of the initial evaluation.A. concolor fl owers and develops fruit in October and November (Andreis et al., 2005), suggesting that a higher number of individuals of this species would be expected in subsequent evaluation periods given the dispersal of propagules and/or seed drop during this time; moreover, according to Scipioni et al. (2011), this species thrives in the understory, a characteristic that makes it a prime candidate for seedling banks.
On the other hand, the seedling abundance of the species was considerably lower in the evaluations carried out at 12 months (December/2014) and 24 months (December/2015), most likely attributable to the higher average maximum temperature and reduced rainfall characteristic of this season.Although most individuals were less than 10 cm tall, a functional root system is required if they are to establish, given that, as noted by McLaren and McDonald (2003), high temperatures combined with water defi cits can lead to desiccation and plant death.Thus, because A. concolor is a typical understory species that adapts to mild temperatures, newly emerged individuals are more susceptible to mortality.
Seedlings of E. rostrifolia, N. megapotamica, T. claussenii, and S. bonplandii were reduced in abundance across all treatments, regardless of removal intensity.Sccoti et al. (2011), who examined the mechanisms of natural regeneration in a Deciduous Seasonal Forest remnant, reported that the dispersal of E. rostrifolia propagules was greatly reduced over two consecutive years, demonstrating the reproductive seasonality of this species.Lower rates of regeneration exhibited by N. megapotamica, T. claussenii, and S. bonplandii may also be due to a rapid loss of viability, relatively lower fruit production, and seasonality eff ects.
Diff ering tolerance levels to removing regenerating individuals were also observed among species.For instance, A. concolor and P. leiocarpa exhibited high regeneration capacity, tolerating 100% removal, whereas E. rostrifolia, N. megapotamica, and T. claussenii-despite their populations eventually being restored in the seedling bank-only maintained regeneration capacity at 50% or lower removal intensities.Similarly, S. bonplandii abundance did not show signs of rebounding even at 36 months after removal.As Andrew et al. (2015) point out, even plants that grow in the same community typically diff er in their requirements, and as such, responses to human disturbances will vary among species.
For several species, individual abundances were signifi cantly reduced compared to the initial evaluation, and in some cases had disappeared altogether.Viani and Rodrigues (2008), who analyzed the impact of seedling removal in a forest remnant, observed that both species with lower regeneration densities and locally rare species were extirpated from the experimental area, demonstrating the detrimental eff ects of seedling removal on populations of such species.As such, removing species with low population densities should be avoided when the objective is to produce forest species seedlings from natural regeneration.
The results of the present study indicate that Picrasma crenata (Vell.)Engl.and Diospyros inconstans Jacq., both rare late secondary species with discontinuous distributions (Carvalho, 2006), vanished from the experimental area following the removal of regenerating individuals, indicating that seedling removal reduced the capacity of these populations to regenerate sustainably.
Populations of Prunus myrtifolia (L.) urb.and Cordia americana (L.) Gottschling JSMill., which require high levels of light during development (Leonhardt et al., 2008;Vaccaro et al., 1999), were signifi cantly lower in the seedling bank following the removal of regenerating individuals.Developing and maintaining soil seed banks is an essential adaptive strategy for early-successional species (Dalling et al., 1998).Regeneration occurs after signifi cant disturbances lead to increased light exposure in the understory and cause seeds to translocate to the upper strata of the forest soil.Because of this, excessive removal of these species from the plant community can hamper forest successional processes following disturbances (e.g., clearings that form after trees fall).
During this experiment, it was possible to identify seasonality in the recruitment of seedlings from the shrub-arboreal component.Previous research on seedling banks in Seasonal Tropical Forests have also indicated species density and richness variations over time during regeneration (McLaren and Mcdonald, 2003;Venturoli et al., 2011).Both richness and density are infl uenced by climatic factors.
We observed marked seasonal eff ects on species density and richness of species during natural regeneration, accounting for the reduction in both individuals and species in the evaluations conducted at 12 months (December/2014) and 24 months (December/2015) following removal.These assessments were performed in summer when temperatures were considerably higher (Figure 1).Previous studies have shown that natural regeneration rates of seasonal forests are greatly aff ected by ambient temperatures (McLaren and McDonald, 2003), and that environmental conditions infl uence a wide range of plant physiological processes, including seed production patterns, seedling germination, survival, and development (Khurana and Singh, 2001).For instance, Gerhardt et al. (1992) reported a reduction in the number of individuals during warmer periods of the year, while Metz et al. (2008) noted that seedling communities are more susceptible to higher temperatures and seasonal water defi cits because of their shallower root systems restricted to the surface layer of the soil.Moreover, periods of infrequent precipitation can result in water defi cits, exacerbated by higher temperatures and more intense solar radiation, leading to plant desiccation and mortality (McLaren and McDonald, 2003).
The results of our analysis suggest that using seedlings taken from natural areas as a source of propagules for subsequent use in forest restoration projects through the transplantation of regenerating individuals may be sustainable if undertaken at moderate removal intensities (up to 50% of individuals).However, restoration eff orts for which this technique might be used should rely only on species with abundant natural regenerative capacity and that develop seedling banks as an adaptive strategy.As Souza et al. (2018) observed, of the thousands of seedlings in a forest, only a handful of individuals complete their life cycles, while the majority succumb to abiotic and biotic fi lters and stochastic events.

CONCLUSION
Our experiments show that removing regenerating individuals at high intensities (>50% of seedlings) compromises the recruitment of new individuals and seriously threatens successional continuity.
Late secondary species generally have a high density of individuals in the seedling bank, and are thus generally resilient to anthropogenic intervention.However, the potential for recovery varies greatly among plant species.Our analysis indicates that removing up to 50% of individuals in the seedling bank is tolerable for A. concolor, E. rostrifolia, T. claussenii, N. megapotamica, and P. leiocarpa.
In addition, any level of seedling removal from natural banks can greatly impact or even lead to the local extirpation of rare or low-density species, and therefore, removal of seedlings is recommended only for species with high densities of individuals in the seedling bank.

Figure 1 -
Figure 1 -Meteorological data for the study region (total precipitation, average maximum and minimum temperatures).Source -Meteorological Station of the Department of Plant Science of the Federal University of Santa Maria (2018).Figura 1 -Dados meteorológicos da região de estudo (precipitação total, temperatura máxima média e temperatura mínima média).Fonte -Estação Meteorológica do Departamento de Fitotecnia da Universidade Federal de Santa Maria (2018).