Community structure of benthic macroinvertebrates inhabiting a highly stratified Mediterranean estuary

The community composition and spatial distribution of benthic macroinvertebrates were studied along the Ebro estuary, a highly stratified estuary located in the NE Iberian Peninsula. During the last decade the oligotrophication process occurring in the lower Ebro River and its estuary has allowed a complex benthic macroinvertebrate community to become established; these results contrast with the poor community found there in the early nineties. A total of 214 taxa were identified, and polychaetes dominated the community both in abundance and species richness. The results showed spatial differences in the structure and composition of macroinvertebrates, which suggests that there are two distinct communities along the estuary. Each community was found in a specific stretch (upper and lower estuary) in function of the presence of the salt wedge. The macrobenthos of the upper estuary was dominated by freshwater taxa, but some euryhaline species were also found. The lower estuary showed a marine community typical of shallow Mediterranean environments. The transition between these two communities fits an ecotone model. The highest abundances, richness and diversities were recorded at the lower estuarine stations, especially those closer to the river mouth, whereas the lowest values corresponded to the stations adjacent to the tip of the salt wedge.

and salinity shows a marked change with a narrow interface between layers called haloclines; and (iv) the isohalines are arranged horizontally.Although this kind of estuary is well represented along microtidal coasts worldwide (e.g. the Mediterranean Sea and the Gulf of Mexico), there is little research on the macroinvertebrate communities that inhabit them.The Ebro estuary has been extensively studied in relation to its hydrology and salt wedge dynamics (e.g.Ibáñez et al., 1997Ibáñez et al., , 1999;;Sierra et al., 2002Sierra et al., , 2004)), and some benthic communities of adjacent areas have also been studied (Capaccioni-Azzati and Martín, 1992;Martín et al., 2000).A few studies have focused on the biota of the estuary (e.g.Rovira et al., 2009), but only one includes a brief description of its macroinvertebrate community (Ibáñez et al., 1995).Furthermore, this study was performed when the lower Ebro River and its estuary were under severe eutrophic conditions, very different from the present situation.Highly fluctuating estuarine systems produce strong environmental gradients, which leads to a patchy distribution of organisms that must cope with a wide variety of stresses (Morrisey et al., 1992;Gray and Elliott, 2009) due to both natural and anthropogenic factors (McLusky, 1999;Dauer et al., 2000;Dauvin, 2007;Elliott and Quintino, 2007).Therefore, the benthic invertebrate communities, often used as indicators of the health of an ecosystem, can be very similar in both impacted and non-disturbed estuarine systems.This therefore increases the difficulty of distinguishing natural from anthropogenic stresses.The Estuarine Quality Paradox concept (Dauvin, 2007;Elliott and Quintino, 2007) refers to the challenge of detecting anthropogenic impacts in naturally stressed systems using biological assessment methods.In Mediterranean regions and particularly in the Iberian Peninsula, besides the spatial fluctuation there is strong temporal environmental variability in the aquatic systems due to limited water availability during part of the year (Caiola et al., 2001;Ferreira et al., 2007a).This variability is exacerbated by a long history of human-induced pressures that have led to serious changes in the natural ecological cycles of estuarine systems from this region (Ferreira et al., 2007b).Therefore, identifying the factors that structure the benthic macroinvertebrate community of the Ebro estuary will provide a clearer understanding of the ecological functioning of the system both at the spatial and temporal scales.Moreover, it will help to interpret the recent changes in the estuarine system observed during the last two decades (Ibáñez et al., 2008).Therefore, this study establishes a robust basis so that macroinvertebrates can be used as indicators of the ecological status of the Ebro estuary.
The purpose of this study was to examine the macroinvertebrate community of the Ebro estuary with regard to species composition, community structure and distribution patterns along spatial and temporal scales and to describe the main abiotic factors affecting benthic communities in this type of estuary.

Study area
The study was conducted in the Ebro estuary (40º43'10''N, 0º40'30''E) located in the NE of the Iberian Peninsula (Catalonia, Spain) (Fig. 1).The Ebro is 910 km long and has a drainage area of 85362 km 2 ; it is the Spanish river with the highest mean annual flow and one of the most important tributaries to the Mediterranean Sea.The main land use in the basin is agriculture with more than 10000 km 2 of irrigation, corresponding to approximately 90% of the water usage in the basin (Ibáñez et al., 2008).The whole basin is strongly regulated by nearly 190 dams (Batalla et al., 2004).These affect the mean annual flow, which has decreased greatly since the beginning of the century to the present (Ibáñez et al., 1996).The Ebro estuary is highly stratified (30 km long, 240 m mean width and 6-8 m mean depth) and the microtidal amplitude of the Mediterranean Sea, about 20 cm (Cacchione et al., 1990), promotes the formation of a salt wedge.The river discharge controls the salt wedge dynamic (advance, retreat and permanence): when the flow exceeds 350-400 m 3 s -1 the salt wedge is pushed from the river channel, and the salt wedge reaches its maximum distance upstream (30-32 km from the river mouth) with flows lower than 100 m 3 s -1 (Ibáñez et al., 1997).

Sampling design and laboratory procedures
Nine sampling stations were established in order to cover the whole estuarine stretch of the Ebro River (Fig. 1).Each station was sampled seasonally (summer 2007 to spring 2008).On each sampling occasion, three sediment samples were collected using a Ponar grab (0.046 m 2 ).The samples were washed in situ through a 0.5-mm mesh sieve to separate macroinvertebrates from sediment, and the organisms retained were immediately fixed with buffered 10% formalin.Later in the laboratory, all macroinvertebrates were sorted, counted and identified under a stereomicroscope to the lowest possible taxonomic level.Two sediment aliquots of 30 g and 200 g were taken from each grab and stored at -20ºC to estimate the total organic matter (TOM) with the loss on ignition method following Kristensen and Andersen (1987), and grain-size characterization according to Holme and McIntyre (1984).Bottom water samples were collected at each station with a water pump, preserved on ice in the absence of light, transported to the laboratory and stored at -20ºC until analysis.Posterior processing included estimating the total chlorophyll and pheophytin concentration using the colorimetric method (Jeffrey and Humphrey, 1975), the dissolved and total nutrient concentration (PO 4, P T , NH 4 , NO 2 , NO 3 , N T and SiO 4 ) following Koroleff (1977) and the suspended solid concentration (Total suspended solids (TSS, mg l -1 ) and organic suspended solids (OSS, mg l -1 )) in compliance with the UNE-EN 872 norm (AENOR, 1996).In addi-tion, physicochemical and hydromorphological characteristics were recorded on each sampling occasion.A YSI 556 multi-parameter probe was used to measure water temperature (°C), dissolved oxygen (mg l -1 ), oxygen saturation (%), pH, salinity and conductivity (mS cm -1 ).Water depth (m) was measured using a Speedtech SM-5 depth-meter sounder.Water flow velocity (m s -1 ) was recorded with a Valeport m.001 current-meter, and water transparency was estimated using a Secchi disc.The accumulated permanence time (in days) of the salt wedge was calculated using daily mean flow values measured 40 km upstream from the river mouth (Tortosa) by counting the accumulated days before each sampling occasion with mean flow values lower than 350 m 3 s -1 .This data is available at the Ebro Water Authority (CHE) web site (http://www.chebro.es/).

Data analysis
The following community descriptive parameters were calculated for each station and season (n=36): total abundance (N), density (D, ind m -2 ), richness (S), Shannon-Wiener's diversity index (H', as log 2 ), Margalef index (d), Simpson dominance index (1-λ') and Pielou's evenness index (J').In addition, species were classified with the constancy index (Dajoz, 1971) into five categories according to the number of stations in which any given taxa was found in relation to the total number of stations: constant (>76%), very common (51-75%), common (26-50%), uncommon (13-25%) and rare (<12%).Each species was classified into feeding guilds based on the available literature.The feeding guilds included deposit feeders (DF), grazers (G), omnivores (O), parasites (Pa), predators (Pr) and suspension feeders (SF).Appendix 1 provides a list of the taxa, together with their feeding guild, that are mentioned in the text.Non-parametric multivari-ate techniques were used as described by Field et al. (1982) to indentify the possible macroinvertebrate communities.A similarity matrix was computed using the Bray-Curtis coefficient (Legendre and Legendre, 1998) after the four root transformation was applied to the abundance data to downweight the contribution of the most abundant taxa to the similarity (Clarke and Warwick, 2001).All the other statistical analyses were performed using the different routines available in the Multivariate Ecological Research software package PRIMER V6 (Clarke and Gorley, 2006).The stations and taxa were ordered using non-metrical multidimensional scaling (MDS) (Clarke and Warwick, 2001).A similarity percentage analysis (SIMPER) that examines the contribution of each variable to the average resemblances between sample groups was performed.This analysis was also used to identify taxa that contributed to dissimilarity among stations and estuary domains that were pre-determined by ordination analysis.Differences in the community composition were identified using the 1-way analysis of similarities test (ANOSIM) that hypothesizes for differences between groups of samples (defined a priori) through randomization methods on a resemblance matrix.Finally, the relationship between the community structure and environmental variables was investigated with the BIOENV routine, which maximizes a rank correlation (Spearman's coefficient) between resemblance matrices derived from biotic and environmental data, iterating for all possible combinations of environmental variables (Clarke and Warwick, 2001).A Spearman's coefficient value close to 0 indicates a weak relation between the community and environmental variables, whereas a value close to 1 indicates that the environmental variables selected explain the community structure.

Water and sediment features
The Ebro estuary has a sand dominated bottom and a relatively low TOM percentage in both the upper (UE) and lower (LE) parts and throughout the entire year (Table 1).During the study period the salt wedge was only found in the lower estuary stations.At these stations, the accumulated permanence time was different in each season: 55, 143, 257 and 344 days respectively for summer, autumn, winter and spring.The null point (the tip of the salt wedge) was located between UE4 and LE5 in all sampling periods.Nutrient concentrations were higher in the upper estuary stretch (Table 1) except for the ammonia, nitrite, phosphate and silicate concentrations in spring and total phosphorous in summer.The chlorophyll concentrations showed marked differences between the upper and lower estuary; the UE stretch had the highest values during winter and spring, whereas the maximum values in the LE stretch were in summer/autumn.Levels of total pheophytin were lower in the UE stretch except for during the two last seasons.The UE stretch always had seasonal mean water flow velocities higher than the LE stretch.The values of TSS and OSS were higher in the LE stretch in summer, autumn and winter, whereas in spring the UE stretch showed the maximum values.

Macroinvertebrate abundance, taxa richness and diversity
During one year of seasonal sampling in the Ebro estuary a total of 21805 individuals were collected belonging to 214 different taxa that comprised 151 species, 115 families, 57 orders, 20 classes and 9 phyla (Supplementary material Appendix 1).Annelida was the dominant phylum and accounted for 71.07% of the total abundance.Polychaeta and Oligochaeta contributed with 49.64% and 21.42% respectively.Spionidae was the most abundant family (28.56%)due to the contribution of the most dominant species Streblospio benedicti (24.10% of the total abundance).Another dominant phylum was Arthropoda, which contributed 15.56% of the total abundance, with Malacostraca accounting for 10.37% of the total abundance.Mollusca was the third most abundant phylum with 12.09% of the total abundance, and Bivalvia contributed 10.61% of the total abundance.In terms of species richness, Polychaeta contributed with 49 different taxa (40 species) and Bivalvia with 37 taxa (32 sp), followed by Gastropoda with 29 taxa (18 sp) and Euentomata with 24 taxa (14 sp).Applying Dajoz's constancy index (considering the 9 stations), 1% of the taxa were found constant, 8% very common, 27% common, 20% uncommon and 44% were rare.Applying the constancy index to UE stations revealed that 9% of the taxa were constant, 14% very common, 19% common, 58% were uncommon and no taxa were rare; whereas in the LE stretch 22% of the taxa were constant, 16% very common, 20% common and 42% were uncommon.
Total density values throughout seasons ranged from 216 to 20022 ind m -2 (Table 2).The highest densities were found at the mouth (station LE9) due to the high abundance of the polychaete S. benedicti.Intermediate densities were found in the uppermost stations UE1 and UE2 with a large contribution of Tubificidae and the introduced bivalve Corbicula fluminea.The lowest densities corresponded to stations UE3, UE4 and LE5 in the middle part of the estuary.Station LE9 had the highest richness values with a maximum of 69 taxa and an annual mean value of 48 taxa; other stations located near the river mouth (LE8 and LE7) also reached high values of richness, whereas stations UE3, UE4 and LE5 showed the lowest richness values (Table 2).Diversity indices showed the same tendency as density and richness, with low values at stations located near the limit of the salt wedge (Table 2).In terms of the trophic structure, the deposit feeders (32%), suspension feeders (29%) and predators (17%) were the dominant feeding guilds in the entire estuary.The contribution of the different feeding guilds in the UE stretch was: deposit feeders (38%), predators (22%), grazers (19%), suspension feeders (14%), omnivores (5%) and parasites (3%).The trophic structure of the LE stretch was dominated by suspension feeders (35%) and deposit feeders (30%).

Analysis of benthic macroinvertebrate communities
Two different communities were determined according to the ordination of stations and taxa of the MDS analysis based on macroinvertebrate abundance.The ordination showed two definite groups of sampling stations: those corresponding to the upper estuary (UE) and lower estuary (LE) respectively (Fig. 2).The UE group (UE1-UE4) included stations located in the upper estuary stretch and corresponded to a freshwater community, whereas the second group comprised the lower estuary stations (LE5-LE9) and had a community with a large marine influence.In addition, we also applied the MDS analysis considering lower taxonomic categories e.g.genus and family; the results obtained showed the same grouping of stations regardless of the taxonomic level employed in the ordinations.Significant differences in community composition were found between these two groups (ANOSIM r: 0.891, p<0.001).Significant differences were also found among stations (ANOSIM global r: 0.694, p<0.001) except for the following pairs: UE1-UE3, UE3-UE4, UE4-LE5, LE5-LE6, LE6-LE7, LE6-LE8, LE7-LE8, LE7-LE9 and LE8-LE9, p>0.05 (Table 3).
The SIMPER analysis showed that the mean community similarity within the UE group was 32.30%.The taxa that most contributed to the high similarity among stations were C. fluminea (27.26%),Tubificidae (18.34%),Naididae (12.02%) and Chironomidae (17.00%).The mean similarity of the LE group was 29.67% with a high contribution from S. benedicti (10.44%),Corophium orientale (8.56%) and Caulleriella zetlandica (6.01%).The similarity contribution of taxa within this group was more balanced than in the UE group, since a total of 35 taxa was necessary to accumulate 90% of the similarity.The mean dissimilarity between these two groups was 96.58% with C. fluminea, S. benedicti, Tubificidae, C. orientale, Naididae, C. zetlandica, Pseudopolydora antennata and Armandia cirrhosa as the taxa with the highest contributions to dissimilarity.
The BIOENV analysis showed that the combination of salinity, dissolved phosphate, total phosphorous, ammonia and the distance from the mouth have a large influence on the structure of the macroinvertebrate communities (r=0.741).The combination of salinity, dissolved phosphate, ammonia and nitrate explained the differences in taxa abundance in the upper estuary (r=0.308).However, within the community of the lower estuary, the combination of ammonia, total chlorophyll, sand percentage, Table 2. -Community descriptive parameters for each sampling station and season.N, total abundance per 0.14 m 2 ; D, density (ind m -2 ); S, richness; H'(log 2 ), Shannon-Wiener diversity index; d, Margalef index; 1-λ', Simpson's index; J', Pielou's evenness; DF (%), deposit feeders; G (%), grazers; O (%), omnivores; Pa (%), parasites; Pr (%), predators; SF (%), suspension feeders.See Figure 1  TOM and the permanence time of the salt wedge showed the highest correlation and explained the main differences in the macroinvertebrate abundance data (r=0.681)

DISCUSSION
The whole Ebro estuary is dominated by sand; however, the percentage of fine deposits such as clay or mud was higher in the lower stretch due to flocculation and settling processes and low velocities recorded at the salt wedge (Sierra et. al., 2002).During the study period the bottom water layer of the estuary showed important differences in physicochemical features between the lower and upper estuary stretches.We found freshwater stations (UE1-UE4) that were not exposed to marine intrusions, and saltwater stations (LE5-LE9) that were permanently exposed to marine intrusions and had a well stratified water column.At LE stations, salinity in the salt wedge decreased upstream with small fluctuations but with values always higher than 30, which evidences the weak mixing between water layers.In highly stratified estuaries the salt wedge dynamics are complex and can be explained by a combination of hydromorphological factors, such as the tide amplitude, river channel cross section and flow, and the freshwater runoff is one of the main factors determining the salt wedge regime (Ibáñez et al., 1997).Nevertheless, in the lower estuary the salt wedge was present on all sampling occasions and the permanence time almost reached a complete year.Although other long periods of marine intrusion in the Ebro estuary have been recorded before (Ibáñez et al., 1995), under natural conditions this period should be approximately 6 months per year (Ibáñez et al., 1997).These conditions of the quasi permanent presence of the salt wedge in the lower estuary stretch are exacerbated by the strong flow regulation and the almost total absence of peak flows, which leads to reduced turbulence and therefore to highly stable densitythermal stratification (Ibáñez et al., 1995(Ibáñez et al., , 1996)).
The present conditions of nutrient loading of the Ebro estuary are quite different from the past situation of eutrophication (Ibáñez et al., 1995).Under eutrophic conditions, and with long periods of permanence of the salt wedge in the lower estuary at the same time, the water quality was worse below the wedge than above it due to organic matter deposition and low water renewal.This organic enrichment caused oxygen depletion through microbial consumption (Ibáñez et al., 1995;Casamayor et al., 2001).Recent changes in the nutrient content of the river, especially the reduction of phosphates, have reduced the primary production in the upper layer, whereas in the lower layer it has increased due to higher light penetration (Falco et al., 2010); thus, the hypoxic conditions in the lower layer have decreased (Casamayor et al., 2001;Ibáñez et al., 2008).
Under the present oligotrophication process, the long periods of salt wedge permanence ensure the stability of the water column, which allows the complexity of the benthic communities to increase, as suggested by Sousa et al. (2006a).The present situation is very different to that of the early nineties, when a survey conducted in October 1992 showed an impoverished macroinvertebrate community (only seven different taxa were found) due to eutrophication, which caused severe anoxic episodes below the halocline (Ibáñez et al., 1995).
The benthic macroinvertebrate community in the Ebro estuary shows considerable spatial and temporal differences, with a complex structure and composition.The multivariate analysis defined two different communities: one from the lower and one from the upper  estuary stretch.In contrast, the pattern described in more mixed estuaries (Rundle et al., 1998;Ysebaert et al., 1998;Sousa et al., 2008) supports the idea that these systems work as a continuum of overlapping communities along the salinity gradient, which fits with the ecocline boundary model suggested by Attrill and Rundle (2002).However, the weak longitudinal salinity gradient and the narrow transition zone between fresh and marine water suggest that the Ebro estuary fits much better into an ecotone model, when ecotone is defined as an area of relatively rapid change that produces a narrow ecological zone between two different and homogeneous community types (Van der Maarel, 1990).
The upper stretch of the Ebro estuary was characterized by an impoverished macroinvertebrate community dominated by the non-indigenous bivalve C. fluminea, which tends to acquire an invasive pattern (Sousa et al., 2006b), together with tolerant taxa such as Tubificidae, Naididae (Oligochaeta) and abundant Chiromidae.The amphipod C. orientale was well-represented in number of individuals but its presence was restricted to stations UE3 and UE4 located close to the salt wedge tip due to its euryhaline nature.The salt wedge community was dominated in terms of abundance by the Polychaeta and Malacostraca classes, followed by the phylum Mollusca.Nevertheless, in terms of richness it was dominated by molluscs, polychaetes and crustaceans in this order.This pattern was slightly different from those found in other temperate intertidal areas, where polychaetes are the most diverse group, followed by molluscs and crustaceans (Ysebaert et al., 1998;Rodrigues et al., 2006).Comparing our results with those from other European estuarine ecosystems we found that the Ebro estuary was colonized in its mouth area by typical marine species associated with the Abra alba-Lagis koreni community (colonizing fine sediments rich in organic matter) and with the Nephtys spp.community (colonizing sandy sediments).These two communities are widely distributed throughout European estuarine and coastal areas (Dauvin, 2000(Dauvin, , 2007;;Martín et al., 2000;Van Hoey et al., 2004;Puente et al., 2008).In addition to these communities, we also found tolerant groups dominated by Capitellidae and Spionidae (Polychaeta), together with Corbula gibba, which usually colonizes disturbed areas; whereas in the upper stations close to the null point the community was dominated by eurybiontic taxa like Hediste diversicolor, Perinereis cultrifera, Heteromastus filiformis, C. orientale and Cyathura carinata.These species are also very common in other European estuaries and coastal areas (Marques et al., 1993;Ysebaert et al., 1998Ysebaert et al., , 2003;;Martín et al., 2000;Chainho et al., 2006;Rodrigues et al., 2006;Sousa et al., 2006aSousa et al., , 2008)).
Currently, the Ebro estuary shows high levels of richness compared with other European estuaries (Rodrigues et al., 2006).The trophic structure is well represented with six different trophic guilds.Deposit feeders, suspension feeders and predators are dominant, which suggests that different resources are available (Brown et al., 2000).In the upper stretch the diversity and richness decreased seawards, with minimum values found close to the null point because the salinity fluctuation is a physiological barrier for stenohaline freshwater and marine taxa (Mannino and Montagna, 1997).However, diversity and richness at the salt wedge stations declined with increasing distance from the sea, which is a recurring tendency in mixed estuaries (Remane and Schleiper, 1971;Schlacher and Woolbridge, 1996).In the Ebro estuary this impoverishment tendency could be explained by the increase in organic matter, ammonia and total phosphorous towards the tip of the salt wedge in combination with the salinity fluctuations in the same area.
The present study provides baseline data that can be used in future ecological studies on this important Mediterranean estuarine ecosystem, as well as in comparisons with other highly stratified estuaries.Complementary studies are necessary to improve our understanding of the spatial and temporal variability of the macrozoobenthic estuarine community.This knowledge could be an important tool for conserving the biodiversity in the Ebro estuary and could be used to develop biological indices for assessing its ecological status according to the Water Framework Directive of the European Union.
Appendix 1. -List of the identified taxa that were found at all the stations over the entire study period.The stations where each taxon was found are also listed.See Figure 1 for sampling station codes; FG, feeding guild (see Table 2  ).-List of the identified taxa that were found at all the stations over the entire study period.The stations where each taxon was found are also listed.See Figure 1 for sampling station codes; FG, feeding guild (see ). -List of the identified taxa that were found at all the stations over the entire study period.The stations where each taxon was found are also listed.See Figure 1 for sampling station codes; FG, feeding guild (see ). -List of the identified taxa that were found at all the stations over the entire study period.The stations where each taxon was found are also listed.See Figure 1 for sampling station codes; FG, feeding guild (see

Fig. 1 .
Fig. 1. -Location of the Ebro estuary and its deltaic plain showing the nine sampling stations.UE, upper estuary stations; LE, lower estuary stations; SW, position of the salt wedge tip.

Fig. 2 .
Fig. 2. -Two dimensional MDS plots based on Bray-Curtis similarities of fourth-root transformed macroinvertebrate abundance data: (a) ordination using inter-species resemblance matrix of nine stations; (b) ordination of the nine stations sampled in the Ebro estuary.The dashed line and the solid line encircle the freshwater and marine communities respectively.See Figure 1 for sampling station codes.
for sampling station codes.

Table 3 .
-One-way ANOSIM test to compare the macroinvertebrate communities at different sampling stations.The test results are shown in the lower diagonal of the table.Significant differences between stations (P<0.05) are indicated (*).The R values are shown in bold letters in the upper diagonal of the table.See Figure1for sampling station codes.