E ﬀ ect of Seasonal Rains and Floods on Seedling Recruitment and Compositional Similarity in Two Lowland Tropical Forests

: Research Highlights : Seasonally ﬂooded and terra ﬁrme forests are characteristic ecosystems of the Colombian Orinoco Basin and of great importance in the maintenance of regional biodiversity and ecosystem function. These forests have a unimodal precipitation regime that can cause a temporal e ﬀ ect on the seedling regeneration niche. This could partly explain the high diversity and coexistence of plant species in these forests, as well as the similarity in composition of seedlings and trees. Background and Objectives : Seedlings are a key factor in the assembly of plant communities. We evaluated the e ﬀ ect of ﬂooding and rains on the dissimilarity and compositional a ﬃ nity between trees and seedlings of seasonally ﬂooded and terra ﬁrme forests. Materials and Methods : the tree community of these forests in San Mart í n (Meta, Colombia) was characterized and compared with their respective seedling communities before (June) and after (December) rain and ﬂooding (during the rainy season). We evaluated plant species diversity and abundance (Shannon diversity and Pielou eveness index), as well as the compositional dissimilarities of each tree community with their corresponding seedling community sampled at the beginning and end of rains and ﬂooding (Bray–Curtis dissimilarity). We also compared sampling site composition using a NMDS analysis. Results: We found that the terra ﬁrme forest had higher diversity compared to the ﬂooded forest. Seedling density in the seasonally ﬂooded forest decreased signiﬁcantly after the ﬂood but not in the terra ﬁrme forest at the end of the rainy season. The compositional dissimilarity between trees and seedlings in the seasonally inundated forest also decreased after the ﬂood. However, this pattern was not evident in the terra ﬁrme forest. Conclusions: These results indicate that seasonal ﬂooding generates a strong ecological ﬁlter that a ﬀ ects the realized niche of plants in these forests. Our results can contribute valuable information for the e ﬀ ective development of assisted restoration and conservation programs. TFF are excellent systems to assess the environmental effect of rain seasonality (on the TFF) and the effect of flooding (on the SFF), in the recruitment and seedling composition of each forest. this study we seek to: (i) compare the diversity and composition of SFF and TFF seedlings, before and after the rains and flooding; (ii) evaluate the effect of seasonal flooding on the dissimilarities and compositional affinity between the seedling and tree communities of an SFF; (iii) study the effect of the rainy season on seedlings of a TFF. Our first hypothesis is that the effect of seasonal precipitation is more pronounced on seedling species richness and abundance of SFF than TFF, due to the severe physiological limitations imposed by flooding As a there is less compositional disparity and greater compositional affinity between SFF trees and seedlings after the the abundant seedling that the imposed by the to


Introduction
Lowland tropical forests are divided into several forest types. These types are characterized by complex ecological dynamics and multiple interactions between biotic and abiotic factors, such as temperature, drought, and flooding [1]. One of these types are the floodplain forests, which are classified according to their seasonality (i.e., season or permanent) and to the river type that floods them (e.g., white-water or várzea, black-water or igapó). The flooding regime and the nature of the river that floods them (nutrient-rich or nutrient-low) impose physiological limitations [1]. Seasonally flooded forests (SFF) perform key ecosystem functions such as water retention, cation exchange, the balance of nutrient cycle, and microclimate; these functions are associated with particular plant communities, which have a narrow regeneration niche [2]. Additionally, organic matter and sediments in the soil largely determine the ecological dynamics of their tree communities [1,3].
Other types of lowland tropical forests, such as terra firme forests (TFF), are also important in maintaining ecological stability and tree diversity in the Amazon and the Colombian Orinoco [4]. Studies have shown the importance of TFF for various primate species in Colombia, as well as the relationship between ecological integrity and the viability of associated primates [5][6][7]. Likewise, TFF of the Colombian Orinoco (especially in the Department of Meta), are considered key in the transition between the Andes and the Amazon [3,8]. Floristically, SFF are less diverse than TFF due to the ecophysiological challenges associated with floods [3,9,10]. Although nutrients and sediments seem to be important in the plant community assembly of TFF, the possible impact of changes in the seasonality of precipitation is not clearly understood [9][10][11][12][13][14].
SFF are characterized by higher monodominance and endemism of most plant species compared to TFF, which makes SFF a key system for ecological, conservation, and management studies [15]. Although the ecological and ecosystem importance of SFF is clearly documented, in Colombia, they have been poorly studied compared to Brazil, Venezuela, and Ecuador [16,17]. The few studies carried out in Colombia have focused on diversity analyzes, richness and abundance of tree species, leaving the ecological regeneration of seedlings, their recruitment and role in the diversity, composition, and functionality of the SFF and TFF barely explored (e.g., [6,18,19]).
Seedlings are an important stage in plant development and are key in the assembly of plant communities [6,[20][21][22][23]. The successful establishment of seedlings recruited to mature individuals reinforces the identity of plant communities, and normally, adults in pristine forests have high compositional similarity or affinity with their seedlings [24][25][26]. For instance, community assembly in a tropical mixed forest of trees and palms is strongly influenced by successful recruitment and establishment of characteristic seedling species [23,24]. However, in other cases, seedlings are unable to establish, as may be the case with pioneer and long-lived pioneer species below closed canopies [27][28][29]. Although seedlings are essential for the assembly and affinity of communities, the ecological processes that allow their recruitment and establishment are not entirely clear and seem to be more complex than processes associated with environmental and geographical distances [30,31].
Thus, the regeneration niche indicates that seedling recruitment and establishment is influenced by heterogeneous microenvironmental conditions [30]. In addition, the temporal division in the regeneration niche allows the interspecific coexistence of seedlings and juveniles of highly diverse communities [6,21,32], more than at any other developmental stage [33][34][35]. In this sense, the temporal differentiation of the regeneration niche can increase the optimization of the available resources based on environmental changes, such as droughts, floods, and fires [21,36]. However, in tropical forests it is not clear when temporal niche differentiation is most effective during plant development [6]; this differentiation can also be altered by ecological factors such as the abundance of pollinators, seed dispersers, and fruiting seasons [5,6,37].
The temporal differentiation of the ecological niche could explain the high richness and coexistence of trees in tropical forests, facilitating the recruitment of seedlings of dominant and rare species, as well as the affinity between adults and seedlings within the community [6,25,32]. Although rare species are often remnants of attacks by natural enemies favored by dominant species [38], this does not fully Forests 2020, 11, 1297 3 of 13 explain the richness and coexistence patterns in various tropical areas [6,38]. Thus, SFF and TFF are excellent systems to assess the environmental effect of rain seasonality (on the TFF) and the effect of flooding (on the SFF), in the recruitment and seedling composition of each forest.
In this study we seek to: (i) compare the diversity and composition of SFF and TFF seedlings, before and after the rains and flooding; (ii) evaluate the effect of seasonal flooding on the dissimilarities and compositional affinity between the seedling and tree communities of an SFF; (iii) study the effect of the rainy season on seedlings of a TFF. Our first hypothesis is that the effect of seasonal precipitation is more pronounced on seedling species richness and abundance of SFF than TFF, due to the severe physiological limitations imposed by flooding [10]. As a second hypothesis, we propose that there is less compositional disparity and greater compositional affinity between SFF trees and seedlings will not differ the flood because the most abundant seedling species that survive the physiological challenges imposed by the flood [3,10,18,39] are more similar to the adult species [25,26,32,36]. Lastly, with respect to TFF, we hypothesize that there are no significant differences in dissimilarity and affinity between trees and seedlings sampled before and after the rains; although rain may be an important environmental factor for TFF, seedling dynamics in TFF are more strongly associated with different types of frugivores than with seasonal environmental conditions [5,34,37].

Study Area
Our study was conducted in the Private Natural Reserve Rey Zamuro y Matarredonda, in the municipality of San Martín, Department of Meta, Colombia ( Figure 1). The mean temperature in the Reserve is 27.5 • C, ranging from 19 to 35 • C [40][41][42]; seasonal annual rain is between 2600 and 3000 mm [42]. The dry season begins in early December and ends in late May, and the rainy season spans from June to the end of November [41,42]. The flora of the Reserve corresponds to the vegetation cover of the Serranía de Manacacías [44], which mainly consists of natural and artificial savannas, seasonally flooded forests (SFF), and terra firme forests (TFF). The SFF forest is inundated by Caño Cumaral (a small white-water river) during a five-month period (the rainy season), from June to November. Flood depth during the month of June was between 1 and 1.2 m. These forests have not been exposed to recent (at least in the last 50 years) anthropogenic disturbance (P.R. Stevenson and F. Castro-Lima, pers. obs.). and TFF are excellent systems to assess the environmental effect of rain seasonality (on the TFF) and the effect of flooding (on the SFF), in the recruitment and seedling composition of each forest.
In this study we seek to: (i) compare the diversity and composition of SFF and TFF seedlings, before and after the rains and flooding; (ii) evaluate the effect of seasonal flooding on the dissimilarities and compositional affinity between the seedling and tree communities of an SFF; (iii) study the effect of the rainy season on seedlings of a TFF. Our first hypothesis is that the effect of seasonal precipitation is more pronounced on seedling species richness and abundance of SFF than TFF, due to the severe physiological limitations imposed by flooding [10]. As a second hypothesis, we propose that there is less compositional disparity and greater compositional affinity between SFF trees and seedlings after the flood because the most abundant seedling species that survive the physiological challenges imposed by the flood [3,10,18,39] are more similar to the adult species [25,26,32,36]. Lastly, with respect to TFF, we hypothesize that there are no significant differences in dissimilarity and affinity between trees and seedlings sampled before and after the rains; although rain may be an important environmental factor for TFF, seedling dynamics in TFF are more strongly associated with different types of frugivores than with seasonal environmental conditions [5,34,37].

Study Area
Our study was conducted in the Private Natural Reserve Rey Zamuro y Matarredonda, in the municipality of San Martín, Department of Meta, Colombia ( Figure 1). The mean temperature in the Reserve is 27.5 °C, ranging from 19 °C to 35 °C [40][41][42]; seasonal annual rain is between 2600 and 3000 mm [43]. The dry season begins in early December and ends in late May, and the rainy season spans from June to the end of November [41,43]. The flora of the Reserve corresponds to the vegetation cover of the Serranía de Manacacías [44], which mainly consists of natural and artificial savannas, seasonally flooded forests (SFF), and terra firme forests (TFF). The SFF forest is inundated by Caño Cumaral (a small white-water river) during a five-month period (the rainy season), from June to November. Flood depth during June was between 1 and 1.2 m. These forests have not been exposed to recent (at least in the last 50 years) anthropogenic disturbance (P.R. Stevenson and F. Castro-Lima, pers. obs.).

Experimental Design
The sample areas were established in 2018 by Terrasos Habitat Bank (an organization specialized in structuring and operating environmental investments for the conservation of strategic ecosystems in Colombia) and Asociación de Becarios del Casanare, in the TFF and SFF, respectively ( Figure 1). Within each of these areas, we used 10 × 10 m quadrats (TFF: n = 21, total area surveyed = 0.21 ha; SFF: n = 20, 0.20 ha), to sample the tree community. To sample the community of seedlings for woody plant species, we established sub-quadrats of 2 × 2 m at the centre of each of the 10 × 10 m tree quadrats, using the same centroid. The quadrats in SFF were established alongside the river (ca. 10-20 m from the river).
We considered trees as all plants with a diameter at breast height (DBH) ≥ 5 cm, and we considered all individuals < 1 m tall and < 1 cm basal diameter as seedlings [4]. Plants between 1-5 cm DBH were therefore not sampled either as trees or seedlings. We conducted sampling and labelling of all trees per quadrat in September 2018, and we sampled seedlings in both forests at the beginning of the rainy and flooding season in June 2019 and at the end, in December 2019.
Tree species were identified by P.R.S., D.M.C., and F.C., while seedling species were identified by B.S.P. with help of D.M.C. and F.C., and field guides from the Field Museum (fieldguides.fieldmuseum.org), filtered by the categories of "Tropical America", "Colombia or Peru or Brazil", "Plants" and "Seedlings" [44-47].

Data Analysis
Tree and seedling composition in the two forest types were characterized and compared in terms of abundance and richness for each quadrat. Families and woody species shared between the tree communities of SFF and TFF were determined, as well as the families and species with the highest abundance within each community. We also calculated the Shannon diversity and Pielou evenness index as descriptors of the tree and seedling community of both forests [48] with the Biodiversity R package version 3.5 [49,50]. It is important to clarify that the TFF is not flooded during the rainy season; therefore, for this forest we compared the effect of before and after the rainy season, while in the SFF, we compared the effect of before and after flooding. Changes in the seedling species richness before and after the flooding and rains were statistically corroborated with a repeated-measures ANOVA using R [50].
The Bray-Curtis index of dissimilarity [51] was used for the analysis of beta diversity to compare and determine the compositional dissimilarities of each tree community with their corresponding seedling community, sampled at the beginning and end of the flooding and rains. This was done with the vegdist function of the Vegan R package [52] using species richness and abundance. Differences between trees and seedlings by forest type were statistically corroborated with a repeated-measures ANOVA [50].
Finally, a non-metric multidimensional scaling analysis (NMDS) was used to compare the sampling sites according to their composition [52]. The calculations of the ordinations within each community matrix were run with different random starts until a stable solution was found (maximum random starts trymax = 50) in a two-dimensional system (k = 2); this was done with the metaMDS function of the Vegan R package. Once the ordinations were obtained for each community matrix, we compared trees and seedlings by forest type to assess the compositional affinity at the beginning and end of the rainy season and flooding. This affinity was statistically validated with the one-way ANOSIM nonparametric test (999 iterations) in the Past software (V3.25) [53]. In this test, there are two parameters to assess the comparisons: a p-value (statistical significance) and an R-value, which calculates the degree to which the ordinations are related to each other. R values between 0.75 and 1 indicate that ordinations are completely different, and values between 0 and 0.25 indicate that there is minimal difference [54]. This is a robust test for comparing and validating the affinity between ecological compositional arrangements [55,56].

Forest Diversity and Composition
We sampled 748 individuals representing 44 families, 97 genera, and 143 species in the tree forest communities. Of these species, 36% were shared between SFF and TFF, 20% were exclusive to SFF and 44% to TFF. TFF had higher species, genera, and family diversity values than SFF (Table 1). Species richness in the seedling communities increased at the end of the rainy season in TFF (p = 0.04), but not in SFF after flooding (p = 0.38) ( Table 1). The number of seedling individuals decreased in SFF (p = 0.03) after the flood, compared with the end of the rainy season in TFF (p = 0.61).   . Tree dominance was stronger in SFF than TFF, as the ten most abundant species corresponded to 70% of the abundance in SFF compared with 42% in TFF (Table  S1). In terms of composition, the families Arecaceae, Burseraceae, Combretaceae, Fabaceae, Lauraceae, Moraceae and Salicaceae represented 69% of the abundance of individuals in SFF, 59% of TFF, and 63% of the shared species. The palm family, Arecaceae, had the highest percentage of tree abundance in SFF, TFF and shared species, with 33%, 28% and 30%, respectively. This family was followed by Burseraceae (10%, 9% and 9%, respectively) and Moraceae (10%, 6% and 8%, respectively).
At the beginning of the rainy season, we found that TFF seedlings of Iryanthera laevis Markgr. and Tococa guianensis Aubl. contributed 20% of the relative abundance. At the end of the rains, this 20% relative abundance was contributed by these two along with eight additional species (Table S2) (Table S2). 15.6% of the abundance at the beginning of the flood and 20.1% at the end, corresponded to the palm family Arecaceae (which corresponded to 7.1% and 6.7% of the species richness, respectively.

Compositional Dissimilarity between Trees and Seedlings
The Bray-Curtis dissimilarity analysis ( Figure 2) shows a different compositional pattern between trees and seedlings in each forest type. We found that trees of SFF had a higher compositional dissimilarity (ANOVA: F = 23.03, p < 0.0001, df = 57) with seedlings sampled at the beginning, than at the end of the flood (Figure 2A). The compositional dissimilarity between TFF trees and each of its seedling communities sampled at the beginning and end of the rains presented a marginal difference (F = 0.65, p = 0.05, df = 60) ( Figure 2B). dissimilarity (p < 0.0001) with seedlings sampled at the beginning, than at the end of the flood ( Figure  2A). The compositional dissimilarity between TFF trees and each of its seedling communities sampled at the beginning and end of the rains presented a marginal difference (p = 0.05) ( Figure 2B).

Affinity between Trees and Seedlings
The NMDS analysis ordered the composition of the quadrats sampled in the different communities analyzed. Paired comparisons were reliable according to the range of stress values found (0.18-0.21). Figure 3A shows a clear separation between the compositional ordination of trees and seedlings at the beginning of the flood, which was significant (p < 0.001, R = 0.45). The comparison between SFF trees and seedling arrangements at the end of the flood ( Figure 3B) was also significant (p < 0.001), although the R-value was much lower (R = 0.13). Both comparisons between trees and seedlings in TFF were significant (p < 0.001; Figure 4), but a higher affinity of trees and seedlings was found at the end of the rainy season (R = 0.35; Figure 4B), than at the beginning (R = 0.40; Figure 4A). The affinity between palms (the most abundant and diverse family) and seedlings was higher after the flood (p < 0.001, R = 0.10), compared to the beginning (p < 0.05, R = 0.40) ( Figure S1). In the TFF, a slightly higher affinity was found before the rains (p < 0.05, R = 0.22), than at the end of the season (p < 0.05, R = 0.24) ( Figure S2).

Affinity between Trees and Seedlings
The NMDS analysis ordered the composition of the quadrats sampled in the different communities analyzed. Paired comparisons were reliable according to the range of stress values found (0.18-0.21). There was a clear separation between the compositional ordination of trees and seedlings at the beginning of the flood, which was significant (p < 0.001, R = 0.45). The comparison between SFF trees and seedling arrangements at the end of the flood ( Figure 3B) was also significant (p < 0.001), although the R-value was much lower (R = 0.13). Both comparisons between trees and seedlings in TFF were significant (p < 0.001; Figure 4), but a higher affinity of trees and seedlings was found at the end of the rainy season (R = 0.35; Figure 4B), than at the beginning (R = 0.40; Figure 4A). The affinity between palms (the most abundant and diverse family) and seedlings was higher after the flood (p < 0.001, R = 0.10), compared to the beginning (p < 0.05, R = 0.40) ( Figure S1). In the TFF, a slightly higher affinity was found before the rains (p < 0.05, R = 0.22), than at the end of the season (p < 0.05, R = 0.24) ( Figure S2).

Discussion
In our first hypothesis, we proposed that the effect of seasonal rains and floods would be more pronounced on seedling species richness and abundance of SFF than TFF. In fact, the seedling communities of TFF had an increase in species richness at the end of the rainy season, while in SFF there was no significant change. The abundance was significantly lower in SFF after the flood, but not in TFF after the rainy season. The dissimilarity between seedling and tree compositions in SFF decreased significantly and compositional affinity increased significantly post-flooding, as we had hypothesized (Hypothesis 2). For trees and seedlings in TFF, we found that the rainy season marginally affected the dissimilarities, while the affinity increased slightly after the rainy season; this is contrary to our hypothesis (Hypothesis 3).

Forest Diversity and Composition
For the tree community, we found greater diversity, equitability, and specific richness in terra firme (TFF) compared to seasonally flooded forests (SFF) ( Table 1). Our species richness result is similar to the species richness in TFF compared to SFF in the Amazon region of Caquetá [57]. Baez and Garate [14] found a greater number of families, genera, and species in TFF compared to SFF in northern Peru, consistent with our results. Regarding α diversity and equitability, they also found the highest Shannon and Pielou index in TFF [14]. The same pattern of diversity was reported for forests in the Colombian Chocó, where greater species diversity was found in TFF compared to flooded forests [58]. This pattern of higher species richness in TFF versus SFF seems to extend to the Neotropics in general [11,12,57]. The higher diversity values and tree richness of TFF compared to SFF may be mainly due to the hostile environmental conditions that characterize the latter [14,37]. Although Gentry (1988) showed that there

Discussion
In our first hypothesis, we proposed that the effect of seasonal rains and floods would be more pronounced on seedling species richness and abundance of SFF than TFF. In fact, the seedling communities of TFF had an increase in species richness at the end of the rainy season, while in SFF there was no significant change. The abundance was significantly lower in SFF after the flood, but not in TFF after the rainy season. The dissimilarity between seedling and tree compositions in SFF decreased significantly and compositional affinity increased significantly post-flooding, as we had hypothesized (Hypothesis 2). For trees and seedlings in TFF, we found that the rainy season marginally affected the dissimilarities, while the affinity increased slightly after the rainy season; this is contrary to our hypothesis (Hypothesis 3).

Forest Diversity and Composition
For the tree community, we found greater diversity, equitability, and specific richness in terra firme (TFF) compared to seasonally flooded forests (SFF) ( Table 1). Our species richness result is similar to the species richness in TFF compared to SFF in the Amazon region of Caquetá [57]. Baez and Garate [14] found a greater number of families, genera, and species in TFF compared to SFF in northern Peru, consistent with our results. Regarding α diversity and equitability, they also found the highest Shannon and Pielou index in TFF [14]. The same pattern of diversity was reported for forests in the Colombian Chocó, where greater species diversity was found in TFF compared to flooded forests [58]. This pattern of higher species richness in TFF versus SFF seems to extend to the Neotropics in general [11,12,57]. The higher diversity values and tree richness of TFF compared to SFF may be mainly due to the hostile environmental conditions that characterize the latter [14,37]. Although Gentry (1988) showed that there

Discussion
In our first hypothesis we proposed that the effect of seasonal rains and floods would be more pronounced on seedling species richness and abundance of seasonally flooded forest (SFF) than terra firme forest (TFF). In fact, the seedling communities of TFF had an increase in species richness at the end of the rainy season, while in SFF there was no significant change. The abundance was lower in TFF after the rainy season. The dissimilarity between seedling and tree compositions in SFF decreased significantly and compositional affinity increased significantly post-flooding, as we had hypothesized (Hypothesis 2). For trees and seedlings in TFF, we found that the rainy season marginally affected the dissimilarities, while the affinity increased slightly after the rainy season; this is contrary to our hypothesis (Hypothesis 3).

Forest Diversity and Composition
For the tree community we found greater diversity, equitability, and specific richness in TFF compared to SFF (Table 1). Our specific richness result is similar to the specific richness in TFF compared to SFF in the Amazon region of Caquetá [57]. Baez and Garate [14] found a greater number of families, genera and species in TFF compared to SFF in northern Peru, consistent with our results. Regarding α diversity and equitability, they also found the highest Shannon and Pielou evenness index in TFF [14]. The same pattern of diversity was reported for forests in the Colombian Chocó, where greater species diversity was found in TFF compared to flooded forests [58]. This pattern of higher species richness in TFF versus SFF seems to extend to the Neotropics in general [11,12,57]. The higher diversity values and tree richness of TFF compared to SFF may be mainly due to the hostile environmental conditions that characterize the latter [14,37]. Although Gentry (1988) showed that there is a strong positive relationship between periodic precipitation and diversity in lowland tropical forests at large scales, at local scales it appears that periodic flooding exerts more pressure on diversity differences between flooded and terra firme forests [10,17,59].
Several of the tree species with the highest relative abundance for TTF (Table S1) have also been reported as important in La Macarena (Meta), while Combretum laxum was the only species of importance in SFF [5]. Stevenson et al. [5] only reported two tree shared species between TTF and SFF, while we report three (Attalea insignis, Crepidospermum rhoifolium and Oenocarpus mapora). This pattern is in agreement with the niche assembly, given the effect that seasonal floods have on the soil and plants [5,10,60,61]. Furthermore, environmental distances have been the main deterministic factors explaining the compositional differences among lowland forests in Colombia compared to neutral factors such as dispersal limitation [3]. The tree composition of SFF was strongly influenced by the high relative abundance of two palm species Oenocarpus mapora (21.5%) and Attalea insignis (11.2%), while for TFF the contribution of these palms was lower (Table 1). This pattern of higher density of few palm species is characteristic of SFF, as well as their greater richness and lower abundance in TFF [14]. In this sense, TFF and SFF can be considered as mixed forests and these forests are characteristic of La Serranía de Manacacías, which extends throughout San Martín and surrounding municipalities [44].
In the seedling communities, the observed changes in the presence and relative abundance of the ten most important species at the beginning and at the end of the rains and floods (Table S2), may be subjected to environmental changes associated with the temporal differentiation of the species regeneration niche [21,26,30,32,62,63]. Environmental heterogeneity exerts a greater filter in seedling communities of tropical forests than in any subsequent ontogenetic phase [5,10,35]. As a consequence, populations are composed of species with diverse adaptations in order to face the environmental challenges [36]. A fundamental environmental factor that shapes the geographic distribution of forest species and communities, especially in TFF, is drought stress [64]. In the Amazon basin, drought has been associated with large-scale mortality of plants [65][66][67]. Seedlings with high growth rates during the wet season, as well as seedlings adapted to cope with drought, have been shown to have higher probability of survival [67,68]. Although we did not measure the effect of drought, the species richness and Shannon index in the seedling community was lower at the beginning of the rainy season (June 2019, which marks the end of the dry season). This pattern could be related to the effects of the dry season in seedling establishment. In addition to the environmental filter, the changes observed here may be due to temporal divisions in seedling recruitment, decreasing competition and seed bank availability, which could maintain the coexistence of multiple species and high plant diversity [5,32,[69][70][71].
In the SFF of the Central Amazon, richness and abundance patterns of várzea (white-water flooded forest) seedlings differs depending on the successional age of the forest. In early-age forests, the abundance of seedlings is almost ten times higher than that reported in older floodplain forests [72]. Likewise, it was reported that within the same plot (early or late succession) there are changes in abundance and species richness, mainly associated with different levels of flooding or its duration. The greatest changes within-forest were found in early-age várzea. It has also been found that in this forest type seedling richness increases after flooding but the abundance tends to decrease [73]. In contrast, seedling richness in igapó (black-water flood forest) decreases as flooding increases [73]. Our results are congruent with what has been reported for várzea, since Caño Cumaral is a small white-water river flooding the SFF. An additional factor that can influence richness and abundance patterns in SFF is high anthropic intervention, where species dominance is low, compared to areas with low intervention [74].

Compositional Difference and Affinity between Trees and Seedlings
The Bray-Curtis analysis showed that the dissimilarity between the SFF seedling community and trees decreased at the end of the flood (Figure 2), with respect to the dissimilarity at the beginning of the flood. This indicates that after the seasonal flooding, the seedling community is more similar to the Forests 2020, 11, 1297 9 of 13 trees of the sampling sites. Likewise, the greater proportion of shared tree and seedling species at the end of the flood corroborates these results (Table S2). The NMDS analysis ( Figure 3B), also showed that after the flooding the affinity between seedlings and trees is significantly stronger. Seasonal flooding is considered to have a relevant deterministic effect on plant communities [11,12]. This could be due to the fact that seedlings that physiologically tolerate flooding, oxygen deficiency and differential nutrient availability [3,10,32,57], are more likely to establish their regeneration niche and advance to adult stages, maintaining the compositional identity of the community [23,30]. In the particular case of palms of the Amazon floodplain, they have been shown to accumulate carbohydrates during the flooding [75], to the extent that reserves in the roots are not exhausted even after 300 days of submersion [76]. Given that many palm species have diverse adaptations to flooding (e.g., [77]), this could explain the higher affinity between seedlings and trees after this period ( Figure S1). The niche temporal division [34] may also explain the changes in compositional dissimilarity and affinity maintaining the assembly of the plant community throughout the different stages of plant development [25,26].
Regarding the compositional affinity between the tree and seedling communities in TFF, marginal differences were found between the dissimilarities of trees with the seedling community ( Figure 2B). The NMDS analysis showed a slight increase in compositional affinity (R = 0.35) just after the rainy season, but this affinity was not very different to what was observed at the beginning of the rains (R = 0.40). Even though TFF is not under the effect of seasonal floods, it can be susceptible to intense rains for a period of almost six months, which can decrease the nutrient availability of the soil [3,10]. However, drought and its duration may be the decisive factor affecting the plant composition of TFF [64][65][66]. This could explain the higher dissimilarity ( Figure 2B) and lower affinity ( Figure 4A) at the end of the dry season (May) and beginning of the rainy season (June), since seedlings of the TFF may be more sensitive to the effects of drying [67,68].

Conclusions
In this study we found that seasonal flooding is an important filter on the temporal regeneration niche of SFF seedlings. Likewise, we found that at the end of the flood, the compositional affinity between seedlings and trees of the SFF increased with respect to the beginning of the flood. This indicates that before the flood many seedlings are established, but at the end of the flood, only seedlings able to withstand the extreme conditions survive. Regarding the TFF, we found that the compositional affinity between seedlings and trees increased slightly at the end of the seasonal rains, possibly reflecting the effects of the dry season on the seedling community. Our results provide valuable information on the role of a regular environmental event (unimodal precipitation and flooding) on the assembly of seedlings and their temporal affinity with trees in two types of tropical forests of great importance for the Colombian Amazon and Orinoco. Additionally, our findings could be useful for restoration and conservation programs aimed at maintaining community identity in these forest types, especially in SFF, which is one of the less studied ecosystems in Colombia and one of the most threatened by anthropogenic factors [3,9].
Supplementary Materials: The following are available online at http://www.mdpi.com/1999-4907/11/12/1297/s1: Table S1: Ten most important species according to their relative abundance (%) in the tree community per sampled forest; Table S2: Ten most important seedling species sampled according to their relative abundance (%) per forest. Figure S1: Comparison between ordinations (A) SFF palm trees-seedlings at the beginning of the flood (R = 0.40; p < 0.05; stress = 0.12); (B) SFF trees-seedlings at the end of the flood (R = 0.10; p < 0.001; stress = 0.17) by means of Non-Metric Dimensional Scaling (NMDS) analysis. The empty squares correspond to the composition of seedlings per sampling quadrat and the solid circles to trees. SFF = seasonal flooded forest. Figure S2