Introduction

For decades, variations in biological diversity have been used to evaluate the effect of natural (e.g., flooding in floodplains; tide dynamics in estuaries, rocky shores, and coral reefs; savanna fires) and anthropogenic (e.g., damming of rivers; deforestation; urbanization; soil use; agriculture) disturbances. These variations have primarily been associated the number of species occupying a particular site (alpha diversity) and variations in species composition among communities (beta diversity). Most studies on the relationship between diversity and environmental alterations have focused on taxonomic aspects of diversity. However, some studies have already demonstrated that a functional approach is more sensitive in capturing the effect of such alterations, mainly when changes in taxonomic diversity are not evident (Mouillot et al., 2013; Braghin et al., 2018; Oliveira et al., 2018). In general, taxonomic and functional richness are positively related (Villéger et al., 2008; Abonyi et al., 2018). However, this is not an established pattern, reinforcing the importance of the use of both approaches (taxonomic and functional) in monitoring community changes more efficiently (Bishop et al., 2015; Braghin et al., 2018; Oliveira et al., 2018).

Functional diversity is an aspect of biodiversity that quantifies the value and range of functional traits affecting the performance of individuals in the environment, and thus ecosystem functioning (Díaz & Cabido, 2001). One way to assess it is through functional space (Villéger et al., 2008), a multidimensional space in which the axes are functional traits and species are placed according to their functional trait values (Mouillot et al., 2013). Individuals (even those belonging to the same species) may have differing functional traits (morphological, physiological, and ecological characteristics) (Violle et al., 2007), while species can also differ in resource use, feeding preferences, absorption rates, and nutrient excretion, which are important functional traits in the ecosystem functioning and stability (Hooper et al., 2005).

Environmental disturbances (natural and anthropogenic) promote ecosystem functional losses or gains and, consequently, alterations in the mechanisms maintaining environmental integrity, such as stability, productivity, and biodiversity (Biswas & Mallik, 2010; Clavel et al., 2011). Anthropogenic environmental alterations have triggered the global defaunation process (Dirzo et al., 2014). The range of anthropic disturbances is wide, but among them we can list deforestation, species invasions, river damming, riparian forest removal, domestic effluent runoff, intensive agriculture, and eutrophication. Some impacts produce rapid spatial changes, modifying the environment in heterogeneous patches or homogenizing large areas (Socolar et al., 2016). Other impacts are perceived only temporally when the same site is evaluated before and after an environmental impact (Bovo-Scomparin et al., 2013; Oliveira et al., 2018; Santos et al., 2018a). Environmental impacts filter species from the regional pool in either space or time, eliminating those that are intolerant to the changes and selecting a more resistant set of species (McKinney & Lockwood, 1999; Vinebrooke et al., 2004). The species resulting from such selection processes reflect the selection of the characteristics of individuals (traits) by the environment or responses of individuals to environmental change (niche selection; Hutchinson, 1957). Thus, environmental changes may simplify the regional diversity, driving numeric loss of species, and/or may reduce the performance of species by changing functional traits. The simplification of a regional species pool may be a case of biotic homogenization because it describes the substitution of the regional species pool by a subset of species that are resistant to environmental change (McKinney & Lockwood, 1999). A convergence in the performance of species (for instance, decreasing body size, utilization of the habitat and exploration of resources) in local communities of a region is also a case of biotic homogenization, but relates to species functional traits. This has been defined as functional homogenization, which is a measurement of increases in the similarity of functional traits among sites (Clavel et al., 2011).

Many studies that identified a negative effect of anthropogenic disturbances on alpha diversity also registered changes in patterns of the spatial distribution of species (Uys et al., 2004; Bonecker et al., 2013; Hawkins et al., 2015; Simões et al., 2015), that is, changes in beta diversity. Nevertheless, actual knowledge of the implications of these changes in the spatial distribution of species on the selection of functional traits and, consequently, on ecosystem functioning, is still incipient. The trait-based metrics can be reactive to environmental disturbances even in the absence of species loss, and may, therefore, provide an early warning of ecosystem disturbances (Mouillot et al., 2013). Functional diversity, through trait selection, tends to be more sensitive to environmental disturbance than taxonomic diversity, thus enabling the possibility of conducting remediation of environmental disturbances before species loss is provoked (Bishop et al., 2015; Braghin et al., 2018; Oliveira et al., 2018).

To understand how the relationship between taxonomic diversity and functional diversity may provide an early warning of ecosystem disturbances and anthropogenic impacts, we used four datasets of cladoceran communities in distinct localities and time. These microcrustaceans are used as environmental sentinels because they respond rapidly to changes and are, therefore, efficient indicators of various types of environmental change, both abiotic and biotic (Eggermont & Martens, 2011). Cladoceran species have different body sizes, occupy different habitats, and present different feeding types and trophic regimes (Allan, 1976; Pennak, 1989; Elmoor-Loureiro, 1997; Forró et al., 2008).

We evaluated if spatial variation in cladoceran species composition (taxonomic approach) is associated with functional homogenization (functional approach) in environments with different degrees of anthropogenic activity, under spatial and temporal perspectives. To do this, we tested three hypotheses:

  1. (i)

    Areas or situations with a higher degree of environmental impact present greater alteration in species composition than areas with a lower degree of impact. In this case, we predicted that reservoirs and urban lakes would present higher beta diversity when compared with natural and rural lakes (spatial approach) and that the most impacted period (temporal approach) would present higher beta diversity both in natural lakes and in reservoirs.

  2. (ii)

    Areas or situations with a higher degree of environmental impact present lower functional space than areas with a lower degree of environmental impact. Thus, we predicted that reservoirs and urban lakes would present a lower functional richness when compared with natural and rural lakes (spatial approach), respectively, and that more highly impacted situations (temporal approach) would present a lower functional richness both in natural lakes and in reservoirs.

  3. (iii)

    Areas or situations with a higher degree of environmental impact present functional homogenization. We predicted that using the spatial approach, reservoirs and urban lakes would present lower functional beta diversity when compared with natural and rural lakes, respectively. Based on the temporal approach, we predicted that more highly impacted situations would present lower functional beta diversity, both in floodplain lakes and in reservoirs.

Materials and methods

For this study, we used four datasets in two approaches: two datasets were analyzed using a spatial approach, and the other two using a temporal approach. In the spatial approach, taxonomic and functional beta diversities were compared between a reference group and impacted areas, in two different geographic regions. The first dataset included 16 lentic environments (eight rural and eight urban) located in the northeast region of Brazil (16° 17′ 51.89″ S, 39° 36′ 24.18″ W; 16° 23′ 0.15″ S, 39° 22′ 19.39″ W). The urban environments were the most impacted due to a lower frequency of riparian forest, domestic effluents, marginal garbage, and higher nutrient concentrations when compared with rural environments (reference group). The second dataset included 27 natural lakes (reference group) and 30 reservoirs (impacted areas), located in the south regions of Brazil (22° 40′ S, 53° 40′ W; 25° 56′ S, 48° 54′ W). Those lakes are located in the Upper Paraná River floodplain (Brazil), and reservoirs are distributed throughout Paraná State (Brazil) (Julio et al., 2005) in six juxtaposed sub-basins (Iguaçu, Piquiri, Ivaí, Paranapanema, Tibaji, and coastal basin) (more details in Simões et al., 2015). Those reservoirs are artificial systems managed to control the water for energy production or storage (Julio et al., 2005) and showed high variability of abiotic features due to land multiple uses of catchment basins (Pagioro et al., 2005); thus, these are anthropologically controlled environments in which the natural variability of rivers is not retained. In contrast, the natural lakes (reference group) are located within three protected units (the Environmental Protection Area of the Islands and Várzeas of the Paraná River, the Ilha Grande National Park, and Ivinhema State Park), and are therefore afforded a higher degree of conservation than the reservoirs.

In applying the first temporal approach, we compared taxonomic and functional beta diversities in the Barragem de Pedra Reservoir (BA, Brazil) (13° 48′ S, 40° 24′ W; 13° 54′ S, 40° 15′ W), in 2003 and 2013/2014. The reservoir (with its hydroelectric power plant) was constructed in 1967 in the middle-lower reaches of the Contas River, and covers an area of 101 km2, with a storage capacity of 1640 hm3; more details can be found in Santos et al. (2018b) and Simões & Sonoda (2009). There was a greater environmental impact in 2013/14 than 2003 (reference situation) because a railroad construction was initiated in 2011 at the margins of the reservoir, altering physical and chemical water variables (Santos et al., 2018b); Secchi disk, pH, total alkalinity, dissolved oxygen, and electrical conductivity of water were higher in 2003 than 2013/2014 (Supplementary Material—Fig. S1). Cladocerans were sampled at five sites in the reservoir in 2003, in the dry and rainy period, totaling 10 samples. In 2013/2014, cladocerans were sampled quarterly at eight reservoir sites, totaling 32 samples. Four sites sampled in 2003 were also sampled in 2013/2014. For the second temporal approach, we compared taxonomic and functional beta diversities in 12 sites (rivers, channels, and lakes) in which cladocerans were sampled quarterly in 2000 and 2010, in the Upper Paraná River floodplain (22° 40′ S, 53° 40′ W; 23° 00′ S, 53° 00′ W). There were 48 samples collected in each year. Most impacts were present in the year 2000 because the damming of an upstream reservoir affected the flood pulse of the Paraná River in that year: on the Upper Paraná River floodplain, the flood pulse shows a typical seasonal dynamic, with floods that usually occur between December and March. However, atypical variations occurred in the year 2000 with a long dry period. Thus, the studied years (2000 and 2010) showed different flood dynamics (Supplementary Material—Fig. S2). Rocha (2010) reported the historical water level variability and showed that at the beginning of the management reservoir the water level was very unstable (Supplementary Material—Fig. S3). Reservoirs break up the hydric landscape and may operate in different ways depending on natural forces and anthropogenic demand (Tundisi, 1999). Oliveira et al. (2018) also considered damming to be an important intervention in flow regulation by affecting the flood period in the floodplain, by causing hydrometric levels to fall far below the overflow level (3.5 m) and thus preventing the occurrence of a defined flood period.

Cladocerans were sampled in all systems with a plankton net (68 µm) through vertical hauls or a suction pump, with known filtered volume: the first and third datasets were sampled with vertical hauls, while the second and fourth datasets were sampled with a suction pump (Lansac-Tôha et al., 2009; Simões & Sonoda, 2009; Simões et al., 2015; Santos et al., 2018b). Cladocerans were identified to the lowest possible taxonomic level, or to morphotype when the species level could not be reached.

The functional trait matrix was composed of feeding type (Sididae, S-type; Daphinidae, D-type; Chydoridae, C-type; Ilyocriptidae, I-type; Bosminidae, B-type), trophic regime (herbivore, detritivore), length of swimming antennae to body size ratio, eye size to body size ratio, presence of ocellus (with or without), presence of color in the carapace (with or without), development of defensive structures (with or without), predatory escape response (pausing and jumping, rapid swimming, not moving), average egg clutch size, and mean body length (Rizo et al., 2017). Clutch size and mean body length are proxies of the production of new biomass that is available for the next trophic level. Both are used to estimate zooplankton secondary production (Winberg et al., 1965; Edmondson  & Winberg, 1971). Such traits highlight different ecological functions: feeding, growth, reproduction, and survival; they are associated different character types (morphological, physiological, behavioral, and life history) (Litchman et al., 2013; Rizo et al., 2017; Sodré & Bozelli, 2019); and express how these species participate in and determine ecosystem processes such as nutrient cycling and secondary productivity (Barnett et al., 2007).

We analyzed taxonomic beta diversity separately for each group of data (two analyses for the spatial approach and two for the temporal approach), comparing the reference and impacted groups. The beta diversity was measured using the Sørensen dissimilarity index. Then, a permutational analysis of multivariate dispersion (Anderson et al., 2006) was performed to test for differences in beta diversity of the cladoceran communities between environments/situations showing low and high impacts, in both the spatial and temporal approaches. In this method, a centroid is computed for each group, and then the distances between each site and centroid of the same group are considered as a measure of beta diversity. We tested the significance of differences in beta diversity using a permutation test with 999 permutations.

We used functional richness (FRic) to evaluate the functional space of the cladoceran communities. FRic represents the functional space occupied by species in a community, based on the combination of their functional traits. Higher FRic values indicate larger differences in the combination of functional traits for a community (Villéger et al., 2008). The functional space is obtained from a principal coordinates analysis (PCoA) calculated from the functional trait matrix for all species. We used the Gower distance to calculate functional distances between pairs of species, as the functional trait matrix is composed of both categorical and continuous traits (Legendre & Legendre, 1998). We used a Student’s t test to evaluate differences in FRic between the reference and impacted groups (Zar, 2010).

To estimate variability in the combination of functional traits, we calculated the functional dissimilarity (beta functional diversity, BF) between two sites based on the volume of convex hull intersections in a multidimensional functional space, where axes are functional traits or synthetic components summarizing functional traits (Villéger et al., 2008; Villéger et al., 2013). Higher values of functional dissimilarity indicate that two communities have greater differences between their combinations of functional traits, while lower values indicate that communities show more similarity in their trait combination, and are more functionally homogeneous. We also used a Student’s t test to evaluate differences in BF between the reference and impacted groups.

In the Student’s t test, when the sample size was unequal, it was used the Welch’s correction to estimate the degree freedom (Zar, 2010). We performed all the analyses in the FD (dbFD function), vegan (betadisper function), betapart (functional.beta.pair function), and stats packages (t test) in the R environment (R Development Core Team, 2015).

Results

The regional diversity was 23 and 30 species in the spatial approach, while in the temporal approach it was nine and 45 species (Table 1). Species richness decreased in most of the impacted situations.

Table 1 Cladoceran species richness in natural, urban and rural lakes, and artificial reservoirs from northeast and south Brazil
Full size table

Taxonomic beta diversity differed between reference and impacted environments in the spatial approach. Lentic environments in rural zones (reference group) presented a lower taxonomic beta diversity (lower multivariate dispersion) than those located in the urban perimeter (higher multivariate dispersion) (F1;13 = 7.17; P = 0.019) (Fig. 1A); natural lakes (reference group) presented a lower taxonomic beta diversity than artificial reservoirs (Fig. 1B) (F1,55 = 4.65; P = 0.035).

Fig. 1
figure 1

Dispersion diagram of the Principal Coordinate Analysis based on the Sørensen dissimilarity. The red polygon and the triangles represent the impacted group (the most impacted); the blue polygon and the circles represent the reference group (the less impacted). A Reference group = rural environments; impacted group = urban environments; B reference group = natural lakes; impacted group = artificial reservoirs; C reference group = before; impacted group = after an environmental disturbance in the Barragem de Pedra Reservoir; D reference group = after 11 years later disturbance in the Paraná River; impacted group = right after disturbance

Full size image

Taxonomic beta diversity did not differ between reference and high impact situation in the Barragem de Pedra Reservoir, but differed in reference and high impact situation in the Upper Paraná River floodplain, in both cases based on the temporal approach. Spatial turnover in the high impact period in the Barragem de Pedra Reservoir was similar to the low impact period (F1;24 = 1.79; P = 0.195) (Fig. 1C); however, the high impact period in the Upper Paraná River floodplain presented lower spatial turnover than the reference situation (F1;93 = 4.32; P = 0.040) (Fig. 1D).

In the spatial approach, functional richness tended to differ between urban and rural lakes (t = 1.95; df = 13.88; P = 0.071) and differed between natural lakes and reservoirs (t = 5.18; df = 41.73; P < 0.001), with the highest values observed in the reference situations (Fig. 2A, B), i.e., in rural compared with urban lakes, and natural lakes compared with reservoirs. In the temporal approach, functional richness was also lower in both impacted situations (Fig. 2C, D), in the Barragem de Pedra Reservoir, during the railroad construction on its margins (t = 1.98; df = 6.03; P = 0.095); and in the Upper Paraná River floodplain, in the year immediately following the hydroelectric closure (t = 2.88; df = 89.98; P = 0.005).

Fig. 2
figure 2

Mean functional richness values comparing high- and low-impacted environments/situations. A Rural and urban lakes; B natural lakes and artificial reservoirs; C before and after an environmental disturbance in the Barragem de Pedra Reservoir; D right after a disturbance in the Paraná River and 11 years later

Full size image

Although communities in highly impacted situations occupied a smaller functional space, we did not register any functional homogenization in the spatial approach, as functional beta diversity increased from the reference to the impacted situation (Fig. 3), i.e., the impacted group showed a higher functional turnover than the reference group. In the temporal approach, we only registered functional homogenization in Barragem de Pedra Reservoir (Fig. 3C).

Fig. 3
figure 3

Mean and standard deviation of functional beta diversity values comparing high- and low-impacted environments/situations. A Rural and urban lakes; B natural lakes and reservoirs; C before and after an environmental disturbance in the Barragem de Pedra Reservoir; D right after a disturbance in the Paraná River and 11 years later

Full size image

Discussion

As predicted, in the spatial approach, highly impacted environments presented higher taxonomic beta diversity than the reference situations and lower functional richness. However, they did not present functional homogenization. This demonstrated that a reduction in functional space (functional simplification) was independent of changes in species composition (species turnover) and changes in the ecological roles of species (turnover of traits). Functional space simplification in more highly impacted sites represents a local loss of functional traits (Oliveira et al., 2018) and an increased risk of losing environmental functions. The higher taxonomic (Fig. 1A, B, and D) and functional (Fig. 3A, B, and D) beta diversities in the more impacted regions showed that different communities (spatial turnover) presented different functions distributed through the studied landscape. Environmental changes may not only alter total species richness, but can also cause a shift in functional space occupation by removing species with traits that are poorly adapted to the new environment and, sometimes, allowing colonization by better-adapted species (McKinney & Lockwood, 1999; Mouillot et al., 2013). We expected to see functional homogenization in impacted sites because of changes to water quality, such as the cultural eutrophication leads to functional homogenization in urban wetlands, i.e., species with similar ecological requirements perform similar ecosystem roles (Dunck et al., 2019). Nevertheless, changes in species composition were followed by increased functional beta diversity. Although many studies have sought to understand if environmental disturbances promote functional homogenization as a form of biotic homogenization and loss of ecosystem functions, recent studies have found that anthropic disturbances do not always cause functional homogenization (Iserhard et al., 2019).

In the temporal approach, when we analyzed the same region in different periods (more than 10 years apart), species spatial turnover was similar between periods, independently of the disturbance caused by railroad construction at the margins of the Barragem de Pedra Reservoir. However, functional richness decreased in impacted situations, and the Barragem de Pedra Reservoir suffered functional homogenization. This demonstrated that, even when there is no difference in taxonomic turnover between different periods, the functional approach may capture the presence of disturbance through decreased functional space, which simplifies the ecological functions of species within ecosystems and, in some cases, leads to biotic homogeneity. Communities that are more similar, with a reduced functional space, may result either in the loss of different species-supported ecosystem functions or decreased efficiency of the supported functions (O’Connor & Crowe, 2005). This combination of results demonstrated the importance of evaluating communities from both a taxonomic and functional perspective to obtain complementary and more precise information on the effect of environmental disturbances on ecosystem health. Furthermore, it also demonstrated the sensitivity of the functional approach to disturbances when traits are excluded from the functional space. Urbanization and environmental changes shape community functional structure by filtering out species according to traits that make them more or less tolerant to the new environmental conditions (Mouillot et al., 2013). The disturbance in the Barragem de Pedra Reservoir combined taxonomic and functional simplification and functional homogenization suggesting more intense disturbance for cladoceran biodiversity.

The disturbance (damming of the river upstream) in the Upper Paraná River floodplain (temporal approach) combined low taxonomic beta diversity (Fig. 1D), functional richness (Fig. 2D), and high functional beta diversity (Fig. 3D). This disturbance, therefore, promoted taxonomic homogenization and functional simplification (reduction in functional space) but was not at a sufficient scale to promote functional homogenization. Floodplains respond more positively to disturbances because are ecosystems with high natural variability (Junk et al., 1989; Neiff, 1990) and, therefore, their biodiversity affords some mechanisms that can buffer environmental disturbances (Simões et al., 2013), such as resilience, which is time of return after a disturbance (Pimm, 1984; Fischer et al., 2001).

Until recently, it was assumed that better-preserved environments would present a higher species turnover than more impacted environments (Socolar et al., 2016). However, recent studies have recorded higher taxonomic beta diversity in sets of sites that are subject to impacts and disturbances (Bonecker et al., 2013; Hawkins et al., 2015; Aspin et al., 2018). This outcome depends on the balance of processes that cause increased differentiation in species composition (biotic heterogenization) or decrease of differentiation (biotic homogenization) between sites (Socolar et al., 2016). The unequal effect of anthropic disturbances on species may either inhibit or favor their occurrence and establishment in a distinct manner, depending on the tolerance of each species, by working as a local environmental filter and contributing to this differentiation (Hawkins et al., 2015; Mykrä et al., 2017). Our results showed that a similar discussion should be addressed to functional beta diversity, as environmental disturbances showed no pattern of association with functional beta diversity among datasets. It is probable that environmental disturbances could select for specific functional traits throughout a landscape, increasing functional heterogeneity among communities, which could, in turn, lead to some ecological functions become rarer and therefore increasing their chances of becoming lost regionally.

Currently, many studies cite the complementary importance of measuring both taxonomic and functional diversity to understand the effect of disturbances on ecosystems. Nevertheless, to our knowledge, a clear relationship between taxonomic diversity, functional diversity, and environmental disturbances has not yet been proposed. Assuming that (i) a local community is a subset of a species pool with a variety of functional traits selected by physical and chemical conditions and biological interactions; and that (ii) an environmental disturbance will select specific functional traits and, consequently, contribute to defining which species will be present in a community, we propose five hypotheses that relate disturbance with taxonomic and functional alterations (functional richness and taxonomic and functional beta diversity), which could lead to loss of ecosystem functions and environmental risk (Fig. 4). These are (1) low environmental risk—the disturbance is not sufficient to produce significant taxonomic and functional changes in the community; (2) moderate environmental risk—the disturbance causes species and functional traits to increase in rarity in the ecosystem; (3) medium environmental risk—the disturbance decreases significantly functional space occupation (FRic), but maintains functional and taxonomic heterogeneity; (4) high environmental risk—the disturbance decreases functional space occupation (FRic) and homogenizes functional traits (BF), but retains taxonomic heterogeneity, and (5) critical environmental risk—the disturbance decreases functional space occupation (FRic) and promotes biotic homogenization (homogenization of both species and functional traits).

Fig. 4
figure 4

Scenarios of environmental risk in function of ecosystem loss due to taxonomic and functional changes in communities. A local community is a subset of species (each letter is a species) of the regional pool with a variety of functional traits (boxes of the same color represent the same functional traits) selected by physical and chemical conditions and biological interactions. Environmental disturbances select specific functional traits and, consequently, define the presence of species in the communities; or they can favor traits and species from species pool once the disturbances can create new environmental conditions

Full size image

There are three limitations of our study: (i) although we consider four datasets (environments under higher and lower influence from anthropic impacts), we obtained only two datasets for each situation (space: urban and rural lakes, and natural lakes and reservoirs; and time: reservoir sites and floodplain environments after disturbance). (ii) Although we selected functional traits that do not directly reflect ecosystem functions related to nutrient cycling (i.e., clearance rate, nutrient absorption), we have used integrative traits as proxies for such functions, e.g., the body size increases respiration and excretion rate (Sodré and Bozelli, 2019), and the content of nutrients (C, N, and P), which is relatively constant in zooplankton groups (Litchman et al., 2013); and the trophic regime and feeding type determine what the organism will eat, and therefore, it is also associated with content of nutrients (C, N, and P). (iii) The definition of higher and lower influence of anthropic impacts is relative and does not guarantee a similar degree of environmental disturbance among the evaluated situations. For example, urban lakes and reservoirs were both considered to be more highly impacted sites, although they are subjected to very different types of disturbance.

Our results demonstrated that the higher alteration of species observed in more impacted environments may result in the impoverishment of local communities as a consequence of disturbance and decreased environmental quality. This decrease in “quality” has caused a reduction in the number of functionally distinct species because of the pressure suffered by local environmental filters. A landscape with higher variation in species composition does not always indicate increased ecosystem functionality. Here, the increase in beta diversity, promoted by anthropogenic action, has decreased the functional space occupied by local communities, and thus the ecological functions of the cladoceran community in the studied ecosystems. Nonetheless, compositional and functional heterogeneity (high taxonomic and functional beta diversity) indicate that species and functional traits are rarer in the landscape and, consequently, more susceptible to local and regional extinction, through either deterministic or stochastic events. In this way, it is important to evaluate the response of species and traits of species to environmental impacts, to that the combination of taxonomic and functional diversity metrics works as a health indicator of an ecosystem.