Regime shifts in coastal lagoons: Evidence from free-living marine nematodes

We test the validity of using the regime shift theory to account for differences in environmental state of coastal lagoons as a response to variation in connectivity with the sea, using free-living nematodes as a surrogate. The study is based on sediment samples from the inner and outer portions of 15 coastal lagoons (5 open to the sea, 5 intermittently open/closed, and 5 permanently closed lakes) along the southern coast of Brazil. Environmental data suggested that there are two contrasting environmental conditions, with coastal lakes being significantly different from open and intermittent lagoons. Marine nematode assemblages corroborate these two mutually exclusive alternative stable states (open vs. closed systems), but assemblages from the intermittently open/closed lagoons showed a gradual change in species composition between both systems independently of the environmental conditions. The gradient in the structural connectivity among lagoons and the sea, due to their regime shifts, changes the movement of resources and consumers and the internal physico-chemical gradients, directly affecting regional species diversity. Whereas openness to the sea increased similarity in nematode assemblage composition among connected lagoons, isolation increased dissimilarity among closed lagoons. Our results from a large-scale sampling program indicated that as lagoons lose connectivity with the sea, shifting the environmental state, local processes within individual intermittently open/closed lagoons and particularly within coastal lakes become increasingly more important in structuring these communities. The main implication of these findings is that depending on the local stable state we may end up with alternative regional patterns of biodiversity.


Introduction
Coastal lagoons are transitional aquatic systems that mediate transfers between the terrestrial environment and the ocean, including potential environmental stressors [1,2]. Lagoons are an evolving coastal landform that may go through a cycle from an open embayment, to a partially back-barrier lagoon with progressive infilling, to a segmentation into small lagoons with unstable inlets and then lakes [3,4] (Fig 1A). The evolution of coastal lagoons is the result of the balance between the processes which act to reduce the size of a lagoon and those, which act to a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Materials and methods
Coastal lagoon sampling and sample processing Coastal lagoons were sampled along the~430 km coast of Santa Catarina State, South Brazil (Fig 2). This coastline can be divided into two major segments:-from the north border, in Itapoá, up to the Cape Santa Marta, at Laguna, the coast is N-S orientated, highly embayed with rocky headlands alternating with small bays;-southwards of Laguna, the straight NE-SW coast is dominated by high energy sandy shores. The most frequent swell wave direction along this coast is from the south, with average heights of 2.5 m; the coast has a micro-tidal regime with higher tides in the north (mean astronomic tide 1 m in Itapoá) than in the south (0.5 in Laguna); the general alongshore littoral drift is from S-SE to ENE-NE, but local reversals take place during strong NE conditions [24]. Coastal lagoons are mainly concentrated in central/ southern portions of the coast (Fig 2).
Fifteen coastal lagoons/lakes (5 closed-Peri, Jaguaruna, Faxinal, Esteves, Cavera; 5 open-Camacho Laguna, Conceição, Barra Velha, São Francisco; 5 intermittent -Lagoinha do Leste, Garopaba, Ibiraquera, Urussanga, Sombrio) were sampled in the austral summer 2012. The lagoons are marginally urbanized, and sampling points have been selected out of the urbanization range to avoid potential influence of anthropogenic impacts. In each of the lagoons, 9 meiofauna samples and 3 of sediment samples (granulometry and total organic content) were taken in the outer and inner portions of the lagoons (a total of 18 samples of meiofauna and 6 samples of sediment per lagoon). The samples of each sample portion were taken dozens of meters apart from each other. For the closed lagoons, inner samples were those taken westward in the most interior region, while the outer samples were taken eastward near the sand barrier (exact location of sampling sites are available in S1 Table). Water salinity was measured in situ with a multiparameter YSI. Meiofaunal shallow subtidal samples (<1m depth) were taken with a plastic syringe of 2 cm in diameter pushed to a depth of 10 cm. They were immediately fixed in 4% formalin, later sieved through a 63 μm mesh and extracted by flotation with Ludox TM (specific gravity: 1.18). Samples were then evaporated with anhydrous glycerol and permanent slides were made [25]. Sediment samples were taken with a PVC corer tube (10 cm Ø and 5 cm height). Granulometry was done by sieving and pipetting analysis and total organic content was determined by loss of ignition (550˚C for 4 hours). Carbonate content of sediments samples was determined by acid digestion [26].
Differences in nematode descriptors and functional attributes among typologies (fixed factor: closed, ICOLL and open), lagoon (random factor: 5 closed, 5 open and 5 ICOLL, nested in typology) and location (random factor: inner and outer, nested in lagoons and nested in typology) were tested with a permutational analysis of variance (PERMANOVA) run on Euclidean distance matrices with 9999 permutations, and the residuals were permuted under a reduced model [31]. To visualize the similarity of the meiofauna composition among different lagoon typologies and location within lagoons, similarity matrices were constructed based on the Bray-Curtis similarity measure. Ordination was done by nMDS, and significance tests for differences in the multivariate structure of nematode assemblages performed using PERMA-NOVA [31]. The variation in species composition of nematode assemblages (beta diversity) was decomposed into replacement and richness difference using abundance data dissimilarities and the Sorensen index [32]. These analyses were performed using the R software [33]. Total β-diversity and decomposed replacement and richness differences were analyzed using PERMANOVA tests with the same design described above. The decomposition of beta diversity can be done by two methods, the "POD" and "BAS" [32]. Although both indices may not show congruent patterns [32,34], in the present study they showed agreement for total dissimilarity and replacement. For richness, the BAS method returned negative sums of squares for the factor typology, while the POD did not (see Results).
Differences in the environmental variables (salinity, mean grain size, sorting, total organic content, sand, silt and clay percentages) were also tested using PERMANOVA using the same design as for the fauna. The relationships between environmental variables and nematode assemblages were explored using distance-based redundancy analysis (dbRDA) that enabled us visualize the percentage of variability in the original data explained by the axis and the relative contributions of each of the predictor variables on the assemblage structure [31].

Ethic statement
No specific permits were required to collect meiofauna as they are microscopic, non-pathogenic and with no special conservation concerns. Field study did not involve endangered species and sampling was carried out in public waters.  Table 1). Most of the nematode genera recorded in open lagoons (70.5%) were those typically found in brackish or marine waters. This proportion was reduced in ICOLLs (49%) with an increasing number of brackish/freshwater or freshwater genera. In closed lagoons, freshwater or brackish/freshwater genera accounted for 95% of the collected fauna. Only 5 genera occurred in all the three types of lagoons, namely Anonchus (Aphanolaimidae), Anoplostoma (Anoplostomatidae), Desmodora (Desmodoridae), Dichromadora and Hypodontolaimus (Chromadoridae). The percentage of exclusive genera (those found exclusively in only one type of lagoon) decreased with increasing connectivity: 13 genera occurred exclusively in the closed lagoons (65%), 26 in the intermittently open/closed (35%) and 21 in open lagoons (30%; Table 1). Trischistoma (Trypilidae), Semitobrilus (Trobilidae) and Ironus (Ironidae) were the most abundant genera in closed lagoons, accounting for 55% of the nematodes collected. At the intermittently open/ closed and open, the genera Microlaimus (Microlaimidae), Spirinia and Desmodora (Desmodoridae) were the most abundant genera in both types of lagoons ( Table 1).

Faunal descriptors and connectivity
The number of genera and diversity of nematodes were significantly higher in the open lagoons, followed by intermittent and were lowest at closed ones ( Table 2 and Fig 3). Density was significantly higher in open lagoons and ICOLLs than in closed ones (Fig 3). Differences in the univariate measures between individual lagoons/lakes occurred mostly within the closed ones (S3 Table). Significant differences between outer and inner portions increased with lagoon connectivity. In the closed lagoons, the descriptors did not show any significant differences between inner and outer portions; in ICOLLs, nematode richness, diversity and density were, in general, higher in outer portion, or did not differ significantly (S4 Table). All Table 1. Summary characteristics of nematode assemblages from coastal lagoons.

Connectivity and assemblage similarities
The non-metric multidimensional scaling (nMDS) analysis revealed substantial differences in nematode assemblages between connected and closed lagoons, but not between locations within the lagoons (Fig 5). The results of the PERMANOVA showed that the greatest variation in the data dissimilarities occurred due to differences in connectivity rather than between locations within lagoons ( Table 3). The statistical tests confirmed that nematode assemblages of closed lagoons differed significantly from more connected ones (Table 3). The PERMANOVA tests also revealed that nematode assemblages of inner and outer portion of closed lagoons did not differ significantly, whilst in the ICOLLS and open they did ( Table 3). The analysis of the average similarity between/within lagoons showed that nematodes assemblages from the open lagoons were more similar to each other than those from the closed ones (Table 3). As lagoons lose connectivity with the sea, nematode composition became more dissimilar. Internal Table 2    similarities, conversely, were higher within closed lagoons, decreasing as lagoons gain connectivity. The results of these analyses were consistent whether analysed by means of presence/ absence or relative abundances. The variability in genera composition (β-diversity) differed significantly according to lagoon typology and location (Table 4). Total β-diversity was significantly higher in closed    Table). Significant variations in β-diversity among lagoons within individual typology were mainly detected in closed ones (S8 Table). At the scale of locations within the lagoon, the genera variability was lower than at the scale of lagoon. Total genera variability between locations varied significantly in open lagoons and ICOLLs, but not in closed ones (S9 Table). The relative contributions of replacement and richness components to the nematode genera variability also differed significantly among lagoon typologies, with and increasing dominance of replacement over richness as lagoon connectivity increased (Fig 6).

Environmental variables and nematode assemblages
Salinity values increased with increasing connectivity and were significantly higher at the outer portions of ICOLLs and open lagoons; salinity did not vary within closed lagoons (S10, S11, S12 and S13 Tables). In general, granulometry was relatively homogeneous among the lagoons, with sediments composed of moderately sorted fine sands (mean grain size and sediment sorting did not vary significantly among nor within lagoons). Grain size, sand, silt + clay percentages did not vary significantly among typologies, nor among lagoons within individual typology (S10 Table). However, total organic content was higher in the inner portion of open lagoons and ICOLLs, while sand percentages were higher in the outer portion of more connected lagoons (S10 and S13 Tables).
The distance-based RDA ordination (Fig 7) indicated that the first two axes explained 30.5% of the variability in the faunal data and 81.1% of the relationship between nematode genera and the environmental variables (Fig 7). The first axis (responsible for 68.4% of the fitted model relating the fauna-environmental variables) was strongly related to salinity, and represented the connectivity gradient from the closed to the permanently open lagoons. The second axis, responsible for 12.7%, was related to sediment sorting, silt and carbonate percentages, and represented the variation within the lagoons.

Discussion
Using particular lagoon status over space as replicates of their evolution over time, we observed that open and closed lagoons are mutually exclusive alternative states of equilibrium, and that ICOLLs are an intermediate or transition phase between them. The gradient in the structural connectivity between lagoons and the sea, due to their regime shifts, changes the movement of resources and consumers, and the internal physico-chemical gradients that directly affected the regional species diversity, abundance and trophic status. Whereas the lack of barriers and the fauna movements through the inlets increased similarity between the more connected lagoons, isolation increased variation in the composition of nematode assemblages with species losses and decrease of trophic diversity between closed lagoons.
Intermittently open/closed lagoons are particularly important in the understanding of biological and physico-chemical shifts between coastal lagoons/lakes. This is because, in the short-term, changes in the connectivity of ICOLLs leads to drastic environmental changes [9,10,11], shifting between the lacustrine or the lagoonar equilibrium state. The sampled ICOLLs in this study were not in the same status of closure (3 were closed and 2 were open) and our expectations were that nematode assemblages of the ICOLLs would be grouped with the closed or the open lagoon samples depending on their inlet state. Instead, we found that ICOLLs univariate and multivariate descriptors of the nematodes assemblages had an intermediate structure between lagoons and coastal lakes. ICOLLs are typically characterized by low freshwater inflow, which leads to sand berm formation across the mouth, preventing mixing with ocean water [37]. Besides, high intra and inter-annual variability in rainfalls and discharges are typical of ICOLLs [38]. A possible mechanistic explanation for the transitional structure of the nematode assemblages is the intermediate pattern of isolation compared to lagoons and coastal lakes. Increasing isolation from open ocean conditions also alters the structure of foraminiferans and macrobenthic assemblages, leading to a decrease in diversity and changes of species [22,39]. The higher diversity of nematode assemblages in brackish water compared to freshwater reflects both the input of marine species and the presence of strong environmental gradients and higher environmental heterogeneity. Overall freshwater nematode communities are impoverished when compared to marine and brackish systems [35]. Regarding nematode abundance, higher values in open lagoons and ICOLLs probably reflects the amount of organic matter. Although just marginally significant, closed lagoons had lower TOC than ICOLL and open ones. Moreover, TOC was significantly higher in the outer portion of open lagoons and ICOLLs, as observed for nematode abundances, giving support to the hypothesis that TOC plays a significant role in nematode abundances [40].
Our results further showed that similarities of the nematode assemblages within and between lagoons also change according to the stable state. While habitat connection and faunal exchange by open inlets increased similarity between more connected lagoons, with variations in the composition controlled by gradients, isolation increased variability of nematode assemblages between closed lagoons. At the same time, internal variability was higher within open lagoons than in closed lagoons, with ICOLL again assuming an intermediate position. This pattern may emerge as a result of the connectivity that modulates the degree to which the inlet state facilitates or impedes the exchange of matter, energy and specimens among landscape elements. Besides, differences in structural connectivity can lead to internal homogeneity or strong physico-chemical gradients that directly affect species composition.
While the low variability of nematode assemblages among lagoons is likely to be a result of faunal transport due to their physical link, the high dissimilarities of the assemblages between coastal lakes might be consequence of their spatially disconnection and exposure to different environmental conditions as a result of the discrete and variable surroundings. The coastal lakes could be colonized by different adjacent freshwater sources, by flooding events [41] or phoresy [42]. Moreover, some common taxa of freshwater and brackish habitats, such as enoplids and chromadorids, could be dispersed from the nearby coastal areas by wind, salt spray or sea foam [41]. Although nematode composition and abundance are known to be closed related to the lakes trophic state and related sediment characteristics [29,43,44], in the present study, all coastal lakes can be classified as oligotrophic, with bottoms composed of clean sandy sediments and very low total organic content (<0.5%). Our results indicated that the nematode assemblages of coastal lakes are primarily structured by the intrinsic properties within each lake and to a possible limited dispersion ability of nematodes between lakes.
Natural and gradual shifts from lagoons to lakes are long-term processes that result from large-scale (e.g. sea-level and climate changes) and local processes (e.g. sediment supply, alongshore drift, coastal morphology) [5,45]. The impoverishment of the nematode assemblages and the substitution of brackish water species by freshwater species also promotes a change in the trophic status of the benthic system and a significant decrease of trophic diversity. The dominant genera of closed lagoons Semitobrilus and Trischistoma are predators, while the genera Desmodora and Theristus are, respectively, epigrowth and non-selective deposit feeders [35,46]. These findings indicate that the availability of trophic resources is strongly affected by shifts from lagoons to lakes, resulting in loss of functional diversity. Similar result was also observed for meiofauna in hard-bottom macroalgal meadow/barren regime shift [47]. Besides, natural shifts may also interact with human interventions, increasing the speed of the shift and changing the dynamics of coastal lagoon evolution [48]. Our results from the large-scale sampling program showed that as lagoons lose connectivity, gradually shifting the state, local processes within individual ICOLLs and particularly within lakes become increasingly more important factors in the structuring of these communities than differences in large scale process (such as geomorphology or biogeography). The main implication of these findings is that depending on the local stable state we may end up with alternative regional pattern of biodiversity.
These findings also have direct implications for management and conservation plans of lagoon environments. As an intermediate state, ICOLLs would play a key role in the management of regime shifts and, based on our results, the most suitable approach for management purposes would be to consider each ICOLL as a unique situation requiring a localized approach, slowing environmental change towards the tipping point (e.g., sediment infill). In the particular case of subtropical coastal lagoons, this imposes additional difficulties as they are mostly distributed among unplanned populated areas. As ICOLLs typically have small river catchments, it makes them sensitive to changing inflow conditions [49]. Poor occupation practices within lagoon floodplains can result in pressure for intervention-dredging and bulldozing to artificially breach or close the lagoon inlet, potentially reducing resilience. Monitoring, establishment of local estuarine management plan and permanent policy review would ensure that the most ecologically appropriate and cost effective options are being implemented at any given location [50].

Conclusions
We conducted an extensive sampling program, using specific lagoon status over space as replicates of their evolution over time, and observed that open and closed lagoons are mutually exclusive alternative states of equilibrium, and ICOLLs are an intermediate or transition phase between them. The gradual regime shift of coastal lagoons, as they lose connectivity with the sea, changes the movement of resources and consumers, and the internal physico-chemical gradients that directly affected regional diversity, abundance and trophic status. Absence of barriers increased the diversity of nematode assemblages and the similarity between the fauna of more connected lagoons. Isolation increased the variation in species composition between lagoons and similarities within lagoons. As local processes within individual lagoons become increasingly more important as they lose connectivity, depending on the local stable state an alternative regional pattern of biodiversity may emerge.
Supporting information S1    Table. Results from pair-wise PERMANOVA tests total beta diversity, and decomposed replacement and richness differences for location (inner vs outer) nested in lagoon and typology. (DOCX) S10