A central theme in community ecology is the relative importance of local and regional processes on the structuring of local biological communities (Ricklefs, 1987). Local processes, such as competition, predation, resource limitation, disturbance, and stochastic events, are known to influence local diversity. Regional processes, such as lifetime of the habitat, differences in speciation and immigration rates, and differences in extinction history (Ricklefs and Schluter, 1993), are more difficult to be incorporated into ecological studies. These processes act on a larger spatial and temporal scale. The metacommunity concept has been applied to understand the direct influence of local processes and indirect influence of regional processes in community structure (Ricklefs, 2008; Leibold et al., 2004; Holyoak et al., 2005). A metacommunity is considered a set of local communities that are linked by dispersal and that potentially interact with one another (Hanski and Gilpin, 1991; Wilson, 1992; Holyoak et al., 2005).

The elements of metacommunity structure (EMS) (Leibold and Mikkelson, 2002) constitute a pattern-based approach to examine the metacommunity. EMS identify which idealized metacommunity structure best reflects the species distribution along the latent gradient. The idealized patterns of species distribution include checkerboard, nestedness, evenly spaced, Gleasonian, Clementsian, and random distributions (Leibold and Mikkelson, 2002; Presley et al., 2010).

The checkerboard pattern occurs when pairs of species co-occurrence is less than expected by chance, demonstrating the existence of competitive exclusion (Stone and Roberts, 1990). The nestedness pattern is found if poor communities that occupy a smaller portion of the environmental gradient are subsets of those that occupy a larger portion of the gradient (Patterson and Atmar, 1986). If species distributions exhibits turnover along the environmental gradient, three patterns can be identified. Evenly spaced distributions occur if the boundaries are hyperdispersed, indicating maximal differences in environmental tolerances among species (Tilman, 1982). If the boundaries of the species distributions are idiosyncratic, the pattern is Gleasonian, with individualistic responses to the underlying environmental gradients and overlapping ranges (Gleason, 1926). The Clementsian pattern occurs if the boundaries are clumped in reference of distinctive “communities” in which the boundaries of species ranges are coincident (Clements, 1916).

Stream systems are characterized by branched and hierarchically organized dendritic networks (Grant et al., 2007) in which the upstream water course is smaller and tend to become larger in the downstream due to the connections between them. The mobility pattern of species is influenced by this structure (Brown and Swan, 2010). In dendritic networks, it is expected that communities of macroinvertebrates and fish assume distinct metacommunity patterns because these organisms respond differently to local and regional effects along the longitudinal gradient (Grant et al., 2007; Brown et al., 2011; Henriques-Silva et al., 2013; Fernandes et al., 2014). The environmental heterogeneity along the upstream-downstream gradient influences the community composition, creating species zonation and addition (Balon and Stewart, 1983; Rahel and Hubert, 1991; Petry and Schulz, 2006; Ferreira and Petrere, 2009).

The longitudinal gradient is central to stream hydraulics, depth, and substrate type and act as a filter on functional species traits (Lamouroux et al., 2002). The upstream-downstream gradient may affect the extinction-colonization events, shaping ecological community and facilitating the identification of the metacommunity pattern (Datry et al., 2016). Motivated by these ideas, we expect that stream fish community will exhibit turnover along the environmental gradient (Clementisian or Gleasonian metacommunity pattern), because both the environmental harshness in the downstream-upstream gradient and local environmental conditions determine the species occurrence in a transition of high to intermediate dispersal and within a instream to across streams gradient (Heino et al., 2015). Our objective was to verify whether a fish metacommunity in the Brazilian Atlantic Forest has a nonrandom structure along the longitudinal stream gradient. To do so, we applied the elements of metacommunity structure (EMS) to examine fish composition patterns at the micro-basin extent for 20 stream fish assemblages.

In the 20 sampled streams stretches, 1189 individuals were collected and classified into 27 fish species (Table S2). The observed number of embedded absences (81) was statistically significant lesser than the mean produced by the null model (137.36, sd=11.36), indicating a nonrandom structure with positive coherence, and species distribution determined by a common latent environmental gradient. The observed number of replacements (3562) was statistically significant and greater than average (1216.04, sd=409.23), which is consistent with a nonnested structure. Morisita's Index (I=1.03) was positive but non-statistically significant (p=0.41), indicating that boundary clumping was stochastic and the metacommunity structure is Gleasonian (Fig. 2).

Fig. 2.

Metacommunity Gleasonian pattern of stream fish assemblage (species code in Table S2).

(0.23MB).

The DCA axes 1 and 2 explained about 50% of data variation (Table 2). On DCA1, the longitudinal gradient ordering the streams stretches from the headwaters (C1, C2, C3, CL1, and P1 with negative scores) to the downstream stretches (C6, C7, C8, A2, A3, P4, P5, and P6) (Fig. 3). The species associated with headwater stretches were Characidium schubarti (Csch), Trichomycterus sp. (Tsp), Astyanax paranae (Apar), and Phalloceros reisi (Prei). The range sample score is about 4.5 along the DCA1 axis, indicating a high beta diversity and that the samples at the low confluence distance share few species with those at the high confluence distance. The upstream stretches have higher confluence distance and declivity, and slower temperature, conductivity, depth, and width (Table S3).

Fig. 3.

Ordination of stream stretches from the detrended correspondence analysis and significant environmental variables (species code in Table S2 and environmental abbreviation in Table S3).

(0.36MB).

We found that fish communities respond to the same latent environmental gradient, showing positive species turnover, but with individualistic responses, characterizing a Gleasonian structure, consistent with metacommunities molded by habitat preferences or specializations. Gleasonian pattern suggested that the communities varied continuously over space, potentially reflecting the degree to which species tolerances overlap. The metacommunity structure may have resulted from the environmental gradient, mainly from regional variables on a geomorphological scale, i.e., confluence distance and marginally with declivity, and local scale, i.e., temperature, conductivity, depth, and width. This environmental gradient can cause dispersion limitations, and for species to colonize a given site in the upstream, it must first have the ability to disperse from its source pool to arrive in the new environment, and then be tolerant to the new biotic and environmental conditions (Brown et al., 2011).

The Gleasonian pattern was molded by biotic and abiotic characteristics that change gradually. In streams systems, which are more continuous, the communities may show less ordered variations, and individualistic responses of species to multiple environmental gradients are expected (Heino, 2013; Heino et al., 2015). This gradual change in the physical, chemical, or structural condition did not allow discrete boundaries or the Clementsian pattern (Henriques-Silva et al., 2013; Willig et al., 2011). The nestedness pattern can be generated by low environmental gradient or dispersal limitation with low colonization capacity (Kodric-Brown and Brown, 1993; Cook et al., 2004). Our streams have gradual but high environmental gradient.

The sites ordination was slightly different from EMS to DCA. Each ordination addresses a different question that is contingent on the basis of ordination, and that the answers to each question may be ecologically valid (Presley et al., 2009). In the EMS ordination (CA), we defined the gradient that is important to metacommunity structure and ordinates communities and species along that latent gradient (Willig et al., 2011). The range sample score along the DCA1 axis indicated a high beta diversity.

Nevertheless, the two ordinations have A. paranae, C. schubarti, Trichomycterus sp., and P. reisi in their extremes. These species dominated most sampling points of upstream assemblages. These streams are characterized by spatial limitations, high food availability, and strong currents with selected species, which have adaptations for maintaining position in rapid current and foraging on substrates (Winemiller et al., 2008). Such species represent high functional diversity and belong to different feeding groups (omnivorous, invertivorous, herbivorous/detritivorous) and habitat use (nektonic, benthic, and benthic) (Casatti and Castro, 2006).

Our results indicated that streams fish communities can change in composition gradually along the environmental gradient, which may be an important driver of the metacommunity structure. Knowing the type of structure and the drivers that shape metacommunities is also relevant for their conservation, and could be useful to understand and predict the impacts of eventual streams disturbance. For example, increases in Small Hydropower Plants are expected as the result of anthropogenic activities (ANEEL, 2015). Our results indicated that fish stream metacommunity follow gradual species replacement structures that are highly influenced by confluence distance. Consequently, communities found at low streams have high share with those found at high confluence distance. Ensuring the connectivity of streams is a good conservation strategy as the species move from one to another, being very dependent on the colonization source. This environmental management can affect biodiversity at local and regional scales, thus we would require devoting local conservation efforts to a large number of different streams reaches and in a micro-basin regional scale.

The authors declare no conflicts of interest.