Influence of species invasion, seasonality, and connectivity on fish functional and taxonomic beta-diversity in a Neotropical floodplain

Studies that combine functional and taxonomic beta-diversity are essential for explaining some ecological processes, including the process of species invasion. We evaluated whether environmental (such as lake connectivity, subsystem and seasonality) and biological factors (occurrence and richness of non-native and native fish species) affect beta-diversity components (total, richness and replacement) of fish communities living in the Upper Paraná River floodplain in Brazil. For this, a distance-based redundancy analysis was performed for both taxonomic (BDtax) and functional (BDfunc) approaches. In order to see which variables influence the local contribution to BDtax and BDfunc (LCBD), mixed effects regression models were fitted. In our research, environmental factors influenced the replacement component for both taxonomic and functional beta-diversity, while biotic factors (occurrence and richness of non-native species) influenced the richness component. The richness of native and non-native species, the occurrence of non-native species and seasonality showed an influence on LCBD values. Although in our study the occurrence and richness of non-native species are positively related to beta-diversity, some studies have found that, in the long term, these species can cause a decrease in functional and taxonomic beta-diversity, altering the ecological multifunctionality of the environment. Our study emphasizes that both changes in environmental factors and species diversity (such as the introduction of a non-native species) can impact the beta-diversity of Neotropical fish.

models were fitted. In our research, environmental factors influenced the replacement component for both taxonomic and functional beta-diversity, while biotic factors (occurrence and richness of non-native species) influenced the richness component. The richness of native and non-native species, the occurrence of non-native species and seasonality showed an influence on LCBD values. Although in our study the occurrence and richness of non-native species are positively related to beta-diversity, some studies have found that, in the long term, these species can cause a decrease in functional and taxonomic beta-diversity, altering the ecological multifunctionality of the environment. Our study emphasizes that both changes in environmental factors and species diversity (such as the introduction of a non-native species) can impact the beta-diversity of Neotropical fish.

Introduction
Studying and analyzing how communities are distributed in space and time is the key to understanding processes underlying community structure (Gaston et al. 2000). Species richness has traditionally been used as a metric to represent community diversity, from which scale-dependent indices can be applied, such as alpha, beta and gamma diversity (Myers et al. 2000). Beta-diversity was initially conceptualized to quantify differences in the taxonomic diversity of species between two or more samples, either on a temporal or spatial scale (Whittaker 1960;Mori et al. 2018). Nevertheless, since species vary in resource use, abundance, and distribution according to certain morphological, physiological and ecological characteristics, referred to as functional traits (Violle et al. 2007), studies that combine functional and taxonomic beta-diversity are essential for explaining how environmental conditions and biotic interactions are related to community structure in space and time with both the species-environment interaction and the trait-environment interaction, which allows for robust pattern identification (Zarabska-Bożejewicz and Kujawa 2018; Chao et al. 2019;Quirino et al. 2021). Furthermore, the use of functional traits to determine beta-diversity allows the measurement of environmental impacts on biotic communities (Carvalho and Tejerina-Garro 2015;Lechêne et al. 2018).
Decomposing taxonomic and functional betadiversity into two components, "replacement" and "richness", as proposed by Podani and Schmera (2011), has become quite common (e.g., Legendre 2014;Lansac-Tôha et al. 2019;López-Delgado et al. 2020). While the replacement component indicates the replacement of species or functional traits from one location to another, the richness component shows the gain or loss of species or traits (Podani and Schmera 2011;Carvalho and Tejerina-Garro 2015). In order to infer the degree of ecological uniqueness of each location where communities occur, a local contribution to beta-diversity analysis (LCBD) can be used. A high LCBD value may indicate that the site has unique ecological conditions that lead to unique species or functional trait combinations (Landeiro et al. 2018;Quirino et al. 2021).
Although difficult to measure immediately, the introduction of non-native species can have major impacts depending on the context of the environment, ranging from direct competition with native species to the extinction of the local biota (Simberloff 2009;Simberloff et al. 2013;Petsch et al. 2022a). The problem is worse for freshwater environments once species introduction has become a major cause of extinction in freshwater environments (Olden et al. 2010;Vitule et al. 2017) just because they are among the most susceptible to invasion (Cohen and Carlton 1998;Petsch 2016), experiencing greater impacts than invasions in terrestrial environments (Thomaz et al. 2015;Petsch 2016), for freshwater environments harbor the highest species richness per surface area on the planet (Dudgeon et al. 2006). Species introduction can lead to great changes in alpha-and beta-diversity (Socolar et al. 2016;Gál et al. 2019). For example, Gavioli et al. (2022) studying fish community data in Italian watercourses found that invasive degree is the most important variable for explaining alpha and beta diversity, having a positive relationship for both taxonomic and functional facets.
River-floodplain systems occur in several freshwater environments and provide many ecosystem services for society (Petsch et al. 2022b). The hydrological regime is the main regulator of river-floodplain systems (Junk et al. 1989), being responsible for the seasonal changes in hydrometric level that produce oscillations in resource availability for fishes, as well as in biotic interactions and limnological conditions Quirino et al. 2017;Petsch et al. 2022b). However, most large rivers of the world are heavily regulated by dams that dampen seasonal pulses and affect the fish community . Consequently, while some lakes in floodplains remain connected to the main river, others become isolated (Lopes et al. 2022). This disconnection from the main channel reduces fish diversity in lakes, since predation and competition are intensified and food availability decreases (Junk et al. 1989;Liu and Wang 2010;Quirino et al. 2017).
Studies using species invasion as a predictor of variation in taxonomic and functional diversity are recent and of great importance (Leão et al. 2020;Gavioli et al. 2022). These studies generally use the degree of invasion as independent of diversity measures (e.g. Milardi et al. 2020;Gavioli et al. 2022), but there is an importance of separately assessing the components of native and non-native species of communities to identify links between invasion dynamics and the loss of diversity of native communities (Gavioli et al. 2022). The Upper Paraná River floodplain in Brazil is composed of distinct ecological zones (subsystems) and, despite being a conservation area (Souza-Filho et al. 2004;Agostinho et al. 2007), presents many non-native fish, being 66 non-native species out of a total of 211 species recorded (Ota et al. 2018). In this study, we evaluated whether environmental factors and attributes of non-native species affect the components of beta-diversity of communities of fish living in lakes in a Neotropical floodplain with a large biodiversity and distinct ecological zones. The objective of this research was to answer three questions: (1) do non-native species richness and native species influence taxonomic and functional diversity similarly? (2) do environmental factors, including lake connectivity, subsystem and seasonality, influence taxonomic and functional beta diversity? (3) does the occurrence and richness of non-native species and environmental factors influence the LCBD?

Study area
The study was conducted in the Upper Paraná River floodplain, which is located in the upper part of the Environmental Protection Area of Islands and Floodplains of the Paraná River (22°45′S; 53°30′W-EPSG: 6933) and constitutes the last undammed stretch of the Paraná River in Brazilian territory, with a length of 230 km (Souza-Filho et al. 2004;Agostinho et al. 2007) (Fig. 1). This floodplain has three flood zones that form three subsystems from the main channels of the Paraná, Baía, and Ivinhema rivers, each of them holding several lakes (Souza-Filho et al. 2004). The habitat diversity of this floodplain includes the alluvial floodplain with numerous secondary channels, connected and isolated lakes, and the main channels of the Paraná, Baia, and Ivinhema rivers Granzotti et al. 2018). Generally, the flood season occurs between November and March, while the dry season occurs between June and September (Agostinho et al. 2004). Although the lakes studied here belong to three different subsystems (Paraná River, Baía River, and Ivinhema River), strong flood pulses manage to homogenize the region completely, since all environments belong to the same hydrological regime ).

Sampling
The fish studied here were sampled by the Long-Term Ecological Research Program of the Paraná River Floodplain (PELD-PIAP 441356/2020-6), funded by the National Council for Scientific and Technological Development (Conselho Nacional de Desenvolvimento Científico e Tecnológico-CNPq). Twenty-two lakes of different sizes were sampled, 11 connected to the main channel of the Paraná, Baía, and Ivinhema rivers and 11 isolated from them ( Fig. 1; Table S1). Fish were sampled quarterly (March, June, September, and December) from 2010 to 2011, using a seine net (10 m long, 0.5 cm mesh size), in the littoral area of the 22 lakes. As some lakes were inaccessible for a few months, a total of 157 samples were taken. All sampled fish were anesthetized with benzocaine, euthanized, fixed in 10% formaldehyde and subsequently identified and classified as native and non-native, according to Ota et al. (2018). The entire process followed a protocol approved by the Ethics Committee on Animal Use of the University (CEUA/ UEM nº 1420221018-ID 001974). Species sampled as part of the PELD-PIAP were deposited in the Ichthyological Collection of the Limnology, Ichthyology and Aquaculture Research Center of the State University of Maringá (NUPÉLIA-UEM).

Functional characterization
Functional traits of the collected fish species were characterized according to Winemiller et al. (2015) as modified by Quirino et al. (2021) for species from the same floodplain. Categorical, binary and continuous traits were used (Table 1), the traits used were related to trophic niche (characterized by the trophic guild of the species), life history (evaluated by the presence or absence of parental care and type of spawning and fertilization) and ecomorphology [considering body shape, mouth position and maximum standard length (sl)]. All functional data were extracted from the literature (more details in Table S2).

Statistical analysis
All statistical analyses and graphs were performed using R software (R Core Team 2020). Taxonomic beta-diversity (hereafter referred to as BDtax) was calculated from a taxonomic dissimilarity matrix of pairs of sampling points by applying the Sørensen Index (Legendre 2014), suitable for presence/ absence data, through the beta function (BAT package; Cardoso et al. 2015). A distance matrix was calculated using the functional traits of species and Gower distance through the "gowdis" function (FD package; Laliberté et al. 2014), which is suitable for mixed data (a matrix with numerical and categorical data) (Gower 1966). A cluster analysis was performed with the distance matrix using the "hclust" function (STATS package; R Core Team 2020). As proposed by Podani and Schmera (2011) and Carvalho et al. (2012), and later expanded to the functional approach by Cardoso et al. (2015), functional beta diversity (hereafter referred to as BDfunc) was then calculated using the taxonomic dissimilarity matrix and the results of the cluster analysis through the "beta" function (BAT package;Cardoso et al. 2015).
Following Podani and Schmera (2011), total betadiversity (referred to as BDtotal) was partitioned into species richness difference (referred to as rich) and species replacement (referred to as repl), as described by Legendre (2014). To determine whether the occurrence and richness of non-native species, subsystem, seasonality (dry and rainy seasons) and lake connectivity (whether they are isolated lakes or connected to the main channel) influence the BDtax and BDfunc of fish, distance-based redundancy analysis (dbRDA; Legendre and Anderson 1999) were performed using the "dbrda" function (vegan package; Oksanen et al. 2019).
The local contribution to BDtax and BDfunc (LCBD; Legendre and De Cáceres 2013), as well as to their three components (total, repl and rich), was calculated using the "LCBD.comp" function (adespatial package; Dray et al. 2021). The relationships of taxonomic and functional LCBD, as well as each of their components, with environmental (connectivity, subsystem and seasonality) and biotic [nonnative species occurrence (NNOccur) and richness (NNRich) and native species richness (NATRich)] factors were assessed by adjusted mixed effects regression models. Models with and without the sampling campaign as a random effect (categorical variable indicating which of the eight sampling campaigns

Continuous
Average length (mm) Influence on dispersal ability, secondary productivity, and energy transfer the collected community represents) were adjusted to determine the existence of possible dependence of the sampling campaign in the variables. The models were subsequently compared according to Akaike Information Criterion (AIC; Akaike 1974). As LCBD is an index with values from 0 to 1, a beta distribution was assumed for LCBD of BDtotal and BDrich, with a logit link in the "glmmTMB" function (glm-mTMB package; Brooks et al. 2022). A normal distribution was assumed for LCBD of BDrepl, in the lme function (nlme package; Pinheiro et al. 2022), since these values met the assumption of normality and this model better fitted the data. During data exploration, we noticed nonlinear relationships between NNRich and NATRich and some LCBD components, for which second and third-order polynomial models were adjusted. The best selected models had homogeneous residuals and the lowest AIC values (Table S3). Graphical construction was accomplished using the "ggplot" function from the ggplot2 package (Wickham 2016).

Characterization of the fish community
We collected a total of 74 species belonging to 28 families and 7 orders, with species richness ranging between 1 and 24 among samples (Table S2). There were a total of 33 non-native species (44.6% of total richness), with non-native sample richness ranging from 0 to 12 species and native sample richness ranging from 0 to 13 species. Of the 157 samplings, in two samplings there were only non-native species and in 130 (82.8% of samplings) had at least one nonnative species.

Distance-based redundancy analysis
For the taxonomic approach, total dbRDA showed that BDtax is influenced mainly by NNRich, followed by NNOccur, connectivity, NATRich, subsystem (Paraná subsystem showed the opposite influence from that of the other two subsystems, as shown by the direction of the arrow in the graph, Fig. 3A), and seasonality (Fig. 3A, Table S4). Repl is influenced by subsystem, connectivity and seasonality (Table S4), while rich is influenced by NNRich, NNOccur and NATRich (Table S4). For the functional approach, total dbRDA showed that BDfunc is mainly influenced by NNRich, followed by NNOccur, connectivity, NATRich and subsystem (Fig. 3B, Table S4). Repl is influenced by connectivity, subsystem and seasonality (Table S4), while rich is influenced by NNRich, NNOccur and NATRich (Table S4).

Taxonomic and functional LCBD
The models without the temporal random parameter (sampling campaign) were the best, having lower AIC values. Therefore, it was assumed that the data do not have a temporal correlation between the sampling campaigns. The best model for each facet and their components, chosen according to lowest AIC, had some selected parameters in common (Table S5). NATRich, NNRich, and NNOccur were chosen in all models. Overall, LCBD and its components showed a similar pattern for taxonomic and functional approaches in response to NATRich and NNRich (Fig. 4, Table S5).
As for the influence of NATRich on the taxonomic approach, a quadratic relationship was verified for all LCBD components. For the LCBD of BDtotal and LCBD of BDrich this relationship presented a U-shape (Fig. 4, Table S5), indicating lower values of total and rich taxonomic LCBD for intermediate richness of native species. For the BDrepl LCBD, on the other hand, this relationship Fig. 3 Distance-based redundancy analysis for taxonomic (A) and functional (B) beta-diversity. NATRich = native species richness; NNRich = nonnative species richness; NNOccur = non-native species occurrence (□ = Absence; ○ = Presence); Sub = subsystem presented a hump-shape, indicating higher values for intermediate richness of native species (Fig. 4, Table S5).
For the functional approach, a negative linear relationship was observed between NATRich and LCBD of BDtotal (indicating a trend of decreasing total LCBD with increasing native species richness) (Fig. 4, Table S4). Besides that, a quadratic relationship between NATRich and BDrepl LCBD with a hump-shape was observed (indicating higher LCBD values for intermediate richness of native species) (Fig. 4, Table S4). Also observed was a third-order relationship NATRich between BDrich LCBD, since LCBD decays up to a certain native species richness (five species), from where it tends to increase slightly (Fig. 4, Table S5).
For the taxonomic approach, NNRich demonstrated a negative linear relationship with total LCBD (indicating a trend of decreasing total LCBD with increasing non-native species richness) (Fig. 4, Table S5). For BDrepl LCBD, NNRich demonstrated a quadratic relationship with a hump-shape, indicating higher repl LCBD for intermediate non-native species richness (Fig. 4, Table S5). For BDrich LCBD, NNRich had a U-shaped quadratic relationship, indicating lower rich LCBD values for intermediate richness of non-native species (Fig. 4, Table S5).
For the functional approach, NNRich showed a U-shaped quadratic relationship with BDtotal LCBD and BDrepl LCBD, indicating lower values of total LCBD and replacement for intermediate richness of non-native species (Fig. 4, Table S5). For BDrich LCBD, NNRich demonstrated a negative linear relationship, indicating a decreasing trend of rich LCBD with increasing non-native species richness (Fig. 4, Table S5).
NNOccur did not show a relationship with taxonomic or functional total LCBD (Fig. 5), but partitioned LCBD had a relationship with it, both for taxonomic and functional diversity. In both approaches, BDrepl had a positive relationship with NNOccur, while BDrich had a negative relationship (Fig. 5, Table S5).
Seasonality was selected in five of the six best-fitting models, and it had a relationship with taxonomic LCBD of BDtotal and LCBD of BDrepl, and with functional LCBD of BDtotal (Fig. 6, Table S5). The rainy season had higher LCBD values than the dry season in all models. That is, the rainy season flood in this floodplain positively influences LCBD compared to the dry season. Connectivity was significant both Fig. 4 Predicted values extracted from the models, representing the significant relationships between the richness of native and non-native species and the functional and taxonomic LCBD values and their partitions; BDtot = Beta diversity total; BDrepl = Beta diversity replacement; BDrich = Beta diversity richness Influence of species invasion, seasonality, and connectivity on fish functional and taxonomic… for taxonomic and functional LCDB of BDrich, with isolated lakes having higher LCBD values than connected lakes (Fig. 7, Table S5).

Discussion
The results presented here indicate that connectivity, subsystem and seasonality, as well as non-native species occurrence (NNOccur), non-native species richness (NNRich), and native species richness (NATRich) have an influence on taxonomic and functional beta-diversity patterns of fish between lakes. Both for taxonomic and functional diversity, environmental factors influenced the replacement component, while biotic factors (NNOccur and NNRich) influenced the richness component. For sites that contributed the most to beta-diversity (LCBD), both native and non-native species richness were related to the three diversity components (total, richness and replacement) for taxonomic and functional approaches. In contrast, NNOccur was related to the LCBD of BDrepl and LCBD of BDrich of both approaches. Seasonality influenced both taxonomic and functional LCBD, while connectivity did not influence any component of LCBD. Overall, our results showed similar results both in taxonomic and functional beta diversity. This result is potentially worrying since any change in taxonomic diversity would likely have a direct impact on functional diversity.
The influence of non-native species on beta-diversity Almost half of the fish species collected in the present study are not native to the region and are present in Fig. 5 Graphs, made with the predicted values extracted from the models, representing the relationships between the occurrence of non-native species and the functional (Func) and taxonomic (Tax) LCBD values, as well as their partitions, pre-dicted by the regressive models. LCBDtot = LCBD for betadiversity total; LCBDrepl = LCBD for beta diversity replacement; LCBDRich = LCBD for beta-diversity richness; Gray: non-significant relationship; Blue: Significant relationship most sampled lakes. At least 66 non-native fish species have been recorded for the Upper Paraná River floodplain (Ota et al. 2018), which may be potential competitors of native species for resources. These introductions are mainly associated with the construction of the Itaipu reservoir in 1982, which flooded the geographic barrier Salto de Sete Quedas that effectively separated two ichthyofaunistic regions, the Upper and Lower Paraná River basins (Langeani et al. 2007;Júlio Júnior et al. 2009). The subsequent construction of a quasi-natural canal for fish transposition in 2002 became a big problem (Vitule et al. 2012). Introducing non-native species into a new environment can directly or indirectly change ecological multifunctionality (Constán-Nava et al. 2015;Moi et al. 2021;Petsch et al. 2022a). The large number of nonnative species in the study area can become a critical problem since environments dominated by non-native species tend to be more susceptible to stochastic extinction of native species (Erős et al. 2020).
The present results show that both the richness of native species and the occurrence and richness of non-native species influenced beta-diversity similarly, mainly the rich component. A biotic acceptance mechanism, common in Neotropical environments, can explain this influence (Ortega et al. 2018;Muniz et al. 2021), since, in these environments, native and non-native fish species respond similarly to increases in available resources (Muniz et al. 2021). According to the biotic acceptance hypothesis, natural ecosystems favorable for the survival of native species, such as those with high resource availability, tend to accommodate the establishment and coexistence of non-native species (Stohlgren et al. 2006;Fridley et al. 2007;Santos et al. 2018). Thus, a local rich in native species tends to become richer with the introduction of non-native species since if the environment is favorable for native species, it tends to be also good for similar non-native species (Stohlgren et al. 2006). Fig. 6 Graphs representing the significant relationships between seasonality and the functional and taxonomic LCBD values and their partitions, predicted by the regressive models. LCBD DBtotal = LCBD for beta-diversity total; LCBD BDrepl = LCBD for beta-diversity replacement Fig. 7 Graphs representing the significant relationships between connectivity and the functional and taxonomic LCBD values and their partitions, predicted by the regressive models. LCBD DBRich = LCBD for beta-diversity richness The influence of environmental factors on beta-diversity Some environmental factors were found to be related to both taxonomic and functional beta-diversity, such as subsystem and lake connectivity. In contrast, seasonality was only related to the taxonomic beta-diversity. Environmental factors were found to be important mainly for the replacement component, showing that these factors are essential in exchanging species between environments. Indeed, environmental factors directly influence fish community diversity (e.g., Agostinho et al. 2004;Petsch 2016;López-Delgado et al. 2020;Lopes et al. 2022).
The hydrological regime is the main component of floodplain seasonality (Junk et al. 1989;Agostinho et al. 2004). The water flow of the Upper Paraná River floodplain is currently regulated by upstream dams, however, the hydrological regime is still the main force influencing the structure and functioning of communities (Agostinho et al. 2004). The alternation between dry and flood seasons in the Upper Paraná River floodplain leads to both limnological and resource availability oscillations Quirino et al. 2017;Petsch et al. 2022b). Some studies in this same floodplain found a greater homogenization of physical and biological characteristics during the flood season (Agostinho et al. 2004;Granzotti et al. 2018), while in the dry season, the sites have less connection with the main channels, becoming more susceptible to tributary inputs, leading to greater dissimilarity (Agostinho et al. 2004;Granzotti et al. 2018). These limnological distinctions can act as environmental filters by selecting fish species that tolerate the conditions of each environment (Leibold et al. 2004;Fernandes et al. 2014), which may contribute to increasing beta-diversity.
Considering that seasonality is important for maintaining the beta diversity of this floodplain, changes in the seasonal variation of the hydrometric level can cause great impacts on the fish community. Recent climate changes in the study area have been responsible for a decrease in rainfall through a decline in extreme rainfall events in the region (Zandonadi et al. 2016), making the seasons more similar. Langer et al. (2018) found variation in annual rainfall to have an influence on fish beta-diversity and suggested that climate change and anthropogenic water level stabilization can impact this diversity. In addition, the insertion of dams upstream of the study site has dramatically modified variation in hydrometric level over the years, reducing the frequency and intensity of floods, which has directly influenced fish diversity (Gubiani et al. 2007;Alves et al. 2021).
For both taxonomic and functional beta-diversity, the Paraná subsystem showed an opposite influence on beta-diversity compared to the other two subsystems. This difference may be associated with two intrinsic environmental characteristics of this subsystem. The first is the presence of a cascade of reservoirs upstream of the Upper Paraná River floodplain, which directly influences the habitats of the Paraná subsystem, while having a substantially smaller influence over the Baía and Ivinhema subsystems (Agostinho et al. 2004;Roberto et al. 2009). The second concerns the fluctuation of the hydrometric level, which is independent in each of the three subsystems (Souza-Filho et al. 2004), leading to environmental changes and high fish diversity in these locations (Agostinho et al. 2004;Lopes et al. 2022).
Habitat connectivity was another factor explaining the taxonomic and functional beta-diversity. This was expected as the disconnection of lakes from a main channel generally reduces fish diversity (Liu and Wang 2010). Although these subsystems are connected, the presence of different types of habitats and tributaries within them increases environmental heterogeneity and, consequently, regional biodiversity ), especially during dry periods when several environments are disconnected.

Local contribution to beta-diversity
In general, this study found that environmental and biological characteristics influence LCBD. High LCBD values may point to degraded, species-poor places in need of ecological restoration or sites with uncommon species combinations (e.g. combination of native and non-native species) and high conservation importance (Legendre and De Cárceres 2013). Some studies show that extremes conditions are important for increasing LCBD, thus sites with low species richness (regardless of whether they are native or non-native) have higher LCBD (Gavioli et al. 2022), as well as a high environmental variability (Leão et al. 2020).
Both BDtotal LCBD and BDrich LCDB were, in general, negatively related to native and non-native species richness, at least to some extent. High LCBD values are generally negatively related to species richness (Heino and Grönroos 2017;Landeiro et al. 2018). Therefore, sites with low species richness can generally contribute to beta-diversity as they harbor unique compositions (Legendre and De Cáceres 2013). Therefore, they may be ecologically unique and more susceptible to impacts. However, despite the negative response, after a certain number of species, the increase in species richness (native or non-native) generated an increase in LCBD values for taxonomic BDtotal and BDrich and functional BDtotal. This positive relationship between LCBD and richness was also observed by Kong et al. (2017), who suggested that it reflects the introduction of novel species (e.g., migratory species) in communities. Thus, sites of both low and high-richness require special attention in terms of environmental protection (Legendre and De Cáceres 2013), as both may represent unique species and traits.
On the other hand, the results of BDrepl LCBD showed a greater species substitution in an intermediate species richness, indicating that in a high richness there is only an increment of species, while in a low richness, there is only the presence of common species. The results of the present study did not show a relationship between the occurrence of nonnative species and total taxonomic and functional LCBD. Nonetheless, they did show a relationship between the LCBD of BDrepl and LCBD of BDrich, both functional and taxonomic with opposite trends. These results show the importance of decomposing beta-diversity into two components to fully understand the predominant processes occurring in a community (Legendre 2014). Thus, sites with the occurrence of non-native species proved important for the replacement of species. In contrast, places without non-native species are more important for the richness component. Since non-native species comprised about half of the species and were present in 76.5% of the sampled points studied, sampled points lacking them generally had lower species richness, leading to a peak of LCBD for the richness component.
The results presented here also showed an influence of environmental variables (seasonality and connectivity) on LCBD. There is less connectivity between environments and a decrease in food availability during the dry season, compared to the flood season (Junk et al. 1989;Quirino et al. 2017). In addition, there is an intensification of some biological interactions during the dry season, such as predation and competition for resources (Thomaz et al. 2008;Fernandes et al. 2014), making each environment more unique, a crucial factor for LCBD.

Conclusion
Although the present study found the occurrence and richness of non-native species to have a relationship with beta-diversity, these species can cause a decrease in functional and taxonomic beta-diversity in the long term (Angulo-Valencia et al. 2022). In fact, this invasion process changes the ecological multifunctionality of the environment (Constán-Nava et al. 2015;Moi et al. 2021;Petsch et al. 2022a). The results showed that the taxonomic and functional beta-diversity of fish in lakes are influenced by environmental factors, such as lake connectivity to the river, seasonality and the different subsystems of the floodplain, which create an environmental gradient, and by biotic factors, such as the occurrence and richness of non-native species. In addition, these parameters influenced the local contribution to beta-diversity. Therefore, this study emphasizes that changes in both environmental factors and species diversity (such as the introduction of a non-native species) can impact the beta-diversity of Neotropical fish.