The dynamic responses of the species to both environmental fluctuations and interspecific interactions may exert a strong influence on the structural assemblages of communities [1, 2]. Theory suggests that not only the magnitude, but also the temporal frequency of the environmental fluctuations matter in altering the structural and functional composition of ecological communities (e.g. coarse- vs. fine-grained environmental changes sensu Levins [3]). For instance, the frequency of the occurrence of the environmental variations may drive the overall resource allocation within the communities, thereby, influencing the ecological specialization of the species and populations. When the environmental fluctuations are small and temporally spaced, communities should be composed of species locally adapted to the relatively stable environment, even if the environmental conditions are severe [3]. Conversely, marked and frequent environmental fluctuations should promote the interchange of different arrays of species with a high diversity of functional traits undergoing a temporal turn-over according to changing conditions [4].

Monitoring the changes in community structure may help detect early signals of environmental disturbance. In particular, a number of studies highlight a link between environmental fluctuations, including anthropogenic disturbance, and biodiversity loss [5, 6]. Ecologists have been long interested in the loss of biodiversity associated with environmental changes focusing mainly on the taxonomic diversity of communities. However, the functional diversity (i.e., the variation of species functional traits within a community [7, 8]), is a primary aspect of biodiversity known to be an accurate predictor of ecosystem functioning [9–11]. There is also growing evidence that both functional and taxonomic diversity are linked to shifts of ecosystem processes [12].

A crucial issue is the trade-off between specialist and generalist species in explaining the functioning of key ecosystem processes [13, 14], as specialist species should be more affected than generalists by environmental changes because of the strong association with their particular niche [15]. Indeed, the concept of specialization is closely related to the prediction of adaptive responses of species in heterogeneous and/or fluctuating environments [3], and its definition relies on "one of the most confusing, and yet important topics in ecology," the niche concept [16]. The degree of specialization is now considered as an informative component of community structure [17]. Therefore, the use of appropriate metrics able to detect spatio-temporal changes in the specialization of the community components is essential to evaluate the effects of both natural and anthropogenic alterations of the environmental conditions in dynamic systems [18]. In particular, a correct distinction between different facets of ecological specialization is required to understand the effects of habitat changes on the biotic homogenization, which can reshuffle existing species distributions by replacing local-adapted species with more widespread and generalist ones, reducing the spatial diversity of communities [19]. In other words, if the alteration of the environment acts as a non-random filter by selecting the species with a higher fitness in the modified ecosystem [20], then the biotic homogenization influences the replacement of 'losers’ species by 'winners’, which increases the spatial similarity of species’ functional traits over time [21]. As a consequence, impacted communities should have lower levels of specialization, since generalist species may better tolerate the environmental changes associated with disturbance (i.e., loss of habitats, hence niches [22]).

Coastal aquatic ecosystems are extremely dynamic habitats where the environmental variations occur over small temporal and spatial scales [23]. In particular in saline systems, this variability is related to inundation/evaporation cycles, which generate highly fluctuating conditions in terms of both frequency and magnitude of changes in the environmental parameters. These fluctuations produce an enduring state of elevated disturbance on the local macroinvertebrate communities [23], which are then subject to large spatial and temporal variation in abundance and diversity. These habitats are, therefore, particularly suitable to implement the use of functional based metrics, which should reveal the effects of environmental changes on the community structure. Such an approach may help capture effects otherwise masked by aggregate community properties [14, 24–26]. For example, to date, there are many studies linking environmental fluctuations to the abundance and species diversity of macrobenthic communities from coastal lagoons [27, 28], but just a few studies analyzed their impact on the species specialization and functional diversity [22, 29, 30].

In this study, we examined long-term changes on the macroinvertebrate community composition of a hyperhaline habitat represented by halted saltworks. The aim was to test whether and how the level of community specialization could be impacted by both natural fluctuations and human-induced alterations in the environment. About fourteen years of monitoring activities have been divided into three main temporal blocks centered on a three-year period of management activities aimed at restoring the water circulation between the saltwork basins. The variation of aggregate community properties (i.e. average abundance) and species specialization (i.e. temporal variation of abundance) were calculated for each array of environmental conditions to check for the reliability of such metrics in describing patterns of community structure. Also, the trends of these metrics were analyzed in basins characterized by different environmental variability to verify their sensitivity in comparing coarse- vs. fine-grained environmental changes.

We investigated how the temporal variability of the environmental features of a hyperhaline system could influence the responses of macroinvertebrates in their use of habitat resources, i.e. their degree of specialization. In particular, we compared the resolution power of the widely investigated aggregate community properties, such as the average abundance, against the variation of the parameter itself (i.e. the variation of abundance) by considering this latter as a proxy of the degree of specialization [18, 31, 32].

Most of the environmental variability within the study area was due to the spatial and temporal fluctuations in the level of salinity, which is confirmed as the main driving force in the structuring process of both benthic communities and species genetic structuring in the study site [33, 34]. This finding is in agreement with the bulk of literature showing that abiotic processes (e.g. salinization) exert a direct impact on communities’ composition [35] or indirect effects that may influence predatory and/or competitive interactions, which enhances top–down effects on species occurrence, abundance, and biomass [36]. Notably, some of the sampled communities did not show any significant pattern of variation in abundance along the main environmental gradient represented by salinity. Macroinvertebrate communities whose structure did not show any substantial temporal change in abundance inhabited the saltwork basins 2 and 3. These basins are less influenced by marine waters compared to basin 1 and are subjected to environmental conditions that were the less variable across the three periods considered and were always the most extreme. On the contrary, the macroinvertebrate community inhabiting basin 1 showed a significant variation in abundance associated with a high variation of physical-chemical parameters both before and after the restoration action (Figure 2).

However, an approach simply relying on abundance may fail in describing spatial and temporal variability of communities, as their composition may be altered without detectable changes in aggregate community properties [37, 38]. Measuring the amplitude of the variation in abundance, density, or biomass may capture effects otherwise not detected when measuring the traits themselves [25, 26]. Evaluating the variability of the descriptors of macrobenthic communities can, therefore, provide insights into the structuring processes of the community itself, but this goal requires assessments that are carried out over multiple temporal and/or spatial scales [39, 40]. This multiple approach has been reliable in quantifying the degree of specialization of species by overcoming density-dependent biases and being statistically independent of the average species density [18]. Even in the absence of an apparent decline of abundance or biomass, an increased temporal and spatial variability of such traits may be the signal of anthropogenic disturbances (e.g. fishing overexploitation [39] and legacies of historical agriculture [40, 41]). A conspicuous loss of functional specialization may also be associated with an increase in species richness, as shown by a long-term study on the fish communities of a coastal lagoon undergoing marked environmental degradation [22]. This paradox was generated since the decreasing species were the most specialized and strictly associated with the most degraded habitats, while the newly found species, responsible for the increase in richness, were functionally redundant with those already present in the community [22].

Our findings well fall within this framework even if we recorded steady values of abundance and only a slight increase in richness with 24 taxa sampled during the post-recovery against 21 from the halt period. We found that assessing the variation of the degree of species specialization highlights significant changes within the sampled communities and puts temporal and spatial variability of these traits in relation with environmental impacts. Indeed, our longitudinal study across fourteen years includes abrupt changes (halt, eutrophication, and post-recovery of the saltworks), which clearly distinguish periods characterized by different environmental conditions and disturbance [42]. In particular, the halt period caused a progressive habitat degradation culminating in the eutrophic period and followed by the works of environmental recovery with a strong impact on the saline system. Accordingly, SSI decreased through the halt period, reached its minimum during the eutrophication period, and then significantly recovered after the restoration. The same trend of SSI was recorded in all three communities analyzed (i.e. saltwork basins), thus efficaciously reflecting the effects of environmental changes (Figures 4 and 5B). On the contrary, the average abundances of sampled macroinvertebrates failed to detect this change in two out of three basins (n. 2 and 3) characterized by more steady assets of physical-chemical parameters (Figure 5A).

Temporal changes in the degree of specialization were associated with the environmental fluctuations, which were in turn related to the different states of the habitat. The highest values of SSI have been recorded during the halt period, when the lack of human intervention modified the environmental conditions allowing marine species to enter and colonize the saltworks. Accordingly, basin 1 was the one with the highest values of SSI. This basin is closely connected to the sea and we found marine species like Modiolus adriaticus (Lamarck, 1819), Neanthes irrorata (Malmgren, 1867), and Amphiglena mediterranea (Leydig, 1851) that inhabited the saltworks during the halt period but disappeared during the subsequent eutrophication phase (Table 2). However, when the levels of habitat degradation became so high to cause eutrophication and algal blooms, SSI values dropped to a minimum with similar values in all three of the basins until the recovery work was carried out. The post-recovery monitoring showed quite high values of SSI, indicating a re-establishment of the communities after the perturbation caused by the recovery actions, which, however, did not reach the values observed during the closure period even if the overall number of taxa recorded was the highest (Table 2). This is likely due to the peculiar features of the saltworks habitat during the closure, when the lack of maintenance made the environment suitable for a number of species (as the already mentioned M. adriaticus, N. irrorata, and A. mediterranea) unfit for either the eutrophic or managed (i.e., restored) habitat.

Therefore, our results underline that maintaining a remarkable level of spatial and temporal heterogeneity is crucial for guaranteeing a high biodiversity at the community level. Indeed, the replacement of specialists by more generalist species may have severe consequences on both community and ecosystem functioning (i.e. the so-called functional homogenization of biodiversity) and decrease the synchronization and variability in the responses to disturbance between connected communities [19].

As coastal aquatic environments are among the most dynamic and impacted ecosystems, it is crucial to understand how the spatial and temporal variability of the environmental conditions may alter the biological composition of their animal communities. Here, we showed how changes in the frequency and magnitude of environmental fluctuations relate to the degree of temporal specialization of macroinvertebrate species in spatially defined communities. Our results support recent advances in the biotic homogenization of biodiversity [19], which relies on the measurement of the increase in similarity of species’ functional traits over time. Measurements based on the species’ contribution to the specific characteristics of community specialization (e.g. variation of density, abundance, or biomass) are considered promising tools in detecting impacts of climate or human-induced alterations on communities [17]. Simple properties of community assembly, i.e. the average abundance of sampled species, failed to detect changes within and between communities facing environmental fluctuations, while the species specialization index (SSI) significantly varied throughout the sampling periods within the macroinvertebrate community.

Long term studies linking environmental changes and impacts to the marine community structure are now emerging [43–45], while in hyperhaline habitats data are still quite scanty (but see [46]) making this study the first case assessing the relationships between the level of community specialization and the long term alteration of a hyperhaline environment.

These results have conservation implications, since an increased replacement of specialists by more generalist species may be the signal of a deterioration of environmental conditions, which reduces the spatial and temporal variability of communities facing disturbance. This increased similarity may have effects on species at higher trophic levels, like the avifauna, by reducing trophic niches and by increasing extirpation rates via intensified species–specific interactions (e.g., functionally similar species might utilize the same spatial resources). Coastal wetlands need constant monitoring and management based on a comprehensive ecosystem approach that links the correct management of the physical-chemical characteristics for the maintenance of community specialization up to food web integrity.