The evolution of the molluscan biota of Sabaudia Lake : a matter of human history

1 Dipartimento di Biologia, Università Roma Tre, Viale Marconi 446, I-00146 Roma, Italy. 2 IBB-Institute for Biotechnology and Bioengineering, Center for Biological and Chemical Engineering, Instituto Superior Técnico (IST), 1049-001, Lisbon, Portugal. 3 Departamento de Ecoloxía e Bioloxía Animal, Universidade de Vigo, Lagoas-Marcosende, Vigo E-36310, Spain. 4 Stazione Zoologica Anton Dohrn, Villa Comunale, I-80121 Napoli, Italy. E-mail: fabio.crocetta@szn.it


INTRODUCTION
Worldwide confined coastal environments have a long history as human-dominated areas (e.g.Mannino andThomas 2002, Davenport andDavenport 2006).They also play a key role within the homogenization of the global biota, being hotspots for the introduction and secondary spreading of alien species (e.g.Occhipinti-Ambrogi 2007).Anthropogenic pressures, land development and eutrophication are the main causes of both the alteration of these ecosystems and the depletion of coastal resources (Beatley 1991, Nixon 1995, Lloret and Riera 2008, Claudet and Fraschetti 2010, Coll et al. 2011).Therefore, periodical annotated check lists would be a useful basis for further studies, assessments and conservation programmes, and a valid tool for characterizing different biotopes inside the study area and making long-term comparisons (Hendrickx and Harvey 1999, Mikkelsen and Cracraft 2001, Costello et al. 2006).
Lack of historical data and the absence of specialists involved in the early studies, however, often make it difficult to reconstruct possible changes occurring within an area.Sabaudia Lake (eastern-central Tyrrhenian Sea, Latium, Italy) is an evident exception to this statement.This basin had very few brackish water periods during the Roman Empire and again around 1700.They were strictly related to the building and the re-opening of the "Roman Channel", a complex sea-lake connection 800 m long, located at the extreme southern tip of the basin and branched in the middle to give two secondary channels.The lack of maintenance, however, led to the closing of the Roman Channel soon after its opening, thus bringing the lake back to its original freshwater status.During the Agro Pontine environmental reclamation (around 1930), another sea-lake connection 500 m long, located at the northern tip of the basin and called "Caterattino", was opened.This connection is now alternatively subjected to severe hypoxia and anoxia events, and sea water flows through it only a few times per year.Sea water renewal and exchange is mainly guaranteed through the definitive re-opening of the Roman Channel.Over the last few decades the area has been subjected to several anthropogenic impacts, including sewages discharges, disturbance by recreational and professional fisheries, seasonal tourism, boat impact and mariculture (including shellfish harvesting).The Sabaudia urban area is located all along the northern tip.
Past datasets covering the local biota are quite satisfactory, and the continuous human commercial activities in the area strongly contributed to the set-up of a conspicuous biological database.Moreover, field research carried out from 1933 to 1995 also offered a quite complete overview of the local molluscan biodiversity (Brunelli and Cannicci 1934, 1940, 1944, Coen 1936, Ferrero 1961, Bini 1983a, 1987, Perdicaro 1984, Mancini et al. 1988, Di Marco et al. 1990, Alia and Russo 2003).A long-term evaluation of faunal changes and human impacts on the local faunal assemblages, however, is still lacking, mainly because of the absence of recent research.The aim of the present work is to fill this gap, through tests on whether temporal variations or qualitative modifications have occurred within the molluscan fauna of Sabaudia Lake during the last century.

Study area
Sabaudia Lake (also known as Paola Lake) (from 41°18'19''N, 013°01'09''E to 41°15'00''N, 013°02'40''E) is the largest of the four coastal lakes integrated in the Circeo National Park (CNP, Latium, Italy).It is located on the extreme southern edge of the Pontine Plain, along the central Tyrrhenian Sea coastline (Fig. 1A), and its main axis stretches along the Tyrrhenian shores for 6.7 km, with six arms crossing it transversely, starting from the mountain side and allowing moderate freshwater to flow through (from north to south: Annunziata, Crapara, Arciglioni, Carnarola, Molella and Bagnara: Fig. 1B).It has an average depth of around 4-4.5 m, a maximum of about 10 m in its southern part and a surface area of approximately 400 ha (Ferrero 1961, Perdicaro 1983, 1984).The Caterattino and the Roman Channel are figured in Figure 1B and 1C, respectively.

Past composition of the molluscan assemblages
An extensive literature survey was carried out.Indexed papers were found and examined, but grey literature was also taken into account.The latter includes historical Italian journals that are no longer published (many of which were never indexed) and malacological papers that still appear in non-indexed journals, thus only allowing for manual searching.Literature record listing was as exhaustive as possible, but for some species multiple records seem to be based on the same samples.Bibliographic data were taxonomically adjusted to allow for some comparison.Although differences in collection methods are unavoidable, and taxonomical knowledge is continuously increasing, most of the typical lagoon species are macrobenthic taxa that have been clearly recognized by biologists and malacologists for centuries, thus suggesting that their presence/absence could not be passed over unnoticed.The evolution of the malacofauna of Sabaudia Lake • 651

Current composition of the molluscan assemblage: sampling design and laboratory work
To describe the current molluscan assemblage of Sabaudia Lake and to evaluate the faunal changes that have occurred within, the taxonomic composition was the main target.Preliminary field observations carried out since April 2007 led us to choose eight sampling sites (Table 1; Fig. 1B, C) on the basis of environmental features (substrate, dominant macrofauna and depth  1).Sediments were sieved through a 1-mm mesh and most of the largest or abundant molluscs were promptly released after field identification, so as to minimize the impact on the local biota.Finally, special emphasis was given to alien species.Upon arrival at the laboratory, samples were first sorted and then transferred to 98% EtOH for subsequent molecular analysis.They are currently preserved at the Dipartimento di Biologia, Università Roma 3, Rome, Italy.Identifications were mostly performed on live material and up to species level.

Updated taxonomy and nomenclature
Updated taxonomy and nomenclature used herein follow the World Register of Marine Species (last accessed 10 July 2013), with few exceptions [taxa listed as a still unsolved complex, if current knowledge does not allow taxa to be correctly identified from the morphology/anatomy only, and new synonymies proposed (see Appendix 1)].The authorities of molluscan species listed below are mostly reported in Tables and Appendix only.

Data analysis
Sampling design was more "faunistic" than "ecological".Nevertheless, the material obtained allowed a first analysis to be made on some ecological traits of the local biota because of the exhaustive census of the molluscan species at the sampling sites.The subsequent analyses were therefore conducted on presenceabsence data only.
Non-metric multidimensional scaling was applied to the data based on the Jaccard index (Legendre and Legendre 1998).Groups of minimum interpretable sites were formed by choosing an arbitrary threshold distance in cluster analysis (Clarke and Warwick 1994), using the complete linkage agglomerative clustering (Borcard et al. 2011).Sites were grouped spatially in accordance with the similarity of species composition (Legendre and Legendre 1998).
Non-parametric permutational multivariate analysis of variance (function Adonis: Oksanen et al. 2006) was applied to test for any changes in the macrobenthic assemblages caused by the different species composition at different sites (Anderson 2001).The analysis was carried out on the dissimilarity matrix calculated using the Jaccard index.Differences in beta diversity were assessed by a test on multivariate dispersion between sites (Anderson 2001, Anderson et al. 2006).Differences in the mean number of species and mean number of shared species between sites were tested by the sign test and Wilcoxon's signed-rank test, respectively (Crawley 2007).
Values are shown as means ± standard deviations.All analyses were carried out using the vegan package (Oksanen et al. 2006), which is run in the free R environment for statistical computing (R Development Core Team 2009).

RESULTS
Literature analysis revealed records of 36 live collected and subfossil taxa from Sabaudia Lake, as published from 1934 to 2011.An updated literature record listing is presented in Table 2 and full details are reported in Appendix 1. Original data from field samplings censused 118 taxa during 2005-2011, all reported in Table 3.The ordination of the sampled sites, showing two dis- tinct groups identified by the cluster analysis, is reported in Figure 3.The distance from the group centroid of the channel and the inner cluster of sites is reported in Figure 4, while the most common species observed in both clusters are reported in Table 4. Finally, faunal data as obtained were divided into four main periods (subfossil records, 1933-1958, 1966-1995, 2005-2011) so as to trace possible evolutionary states of the local molluscan biota.

The 2005-2011 period
Recent field studies covering Sabaudia Lake and first including the whole Roman Channel, recorded by census 119 taxa in 2005-2011, mostly belonging to Gastropoda (62 taxa) and Bivalvia (50 taxa).With the sole exception of Haminoea orbignyana, Heleobia stagnorum and Ecrobia ventrosa, the species recorded in 1974-1995 still continue living in the area (Tables 2,  3, Appendix 1).
The cluster analysis performed on the molluscan distributional pattern identified two groups of sites, Sabaudia Lake inner sites (S1-S4: Inner Cluster) and Roman Channel sites (S5-S8: Channel Cluster).The Inner Cluster had a mean number of 27.7±6.3species per site, significantly different from the 66.5±8.1 species per site recorded in the Channel Cluster (sign test, p<0.001).In turn, the mean number of sites sharing the same species was 3.17±0.74for the Inner Cluster, significantly different from the average of 2.33±1.20 sites sharing the same species in sites of the Channel Cluster (Wilcoxon test, V=930, p<0.001).The ordination of the clusters by non-metric multidimensional scaling is shown in Figure 3.It rendered a good fit of low stress (stress=0.02;Clarke and Warwick 1994).The Adonis multivariate test revealed significant differences in composition between the Inner and Channel Clusters (F 1,6 =8.076, p=0.042) (Table 4), although the results obtained in the multivariate dispersion of the groups were not significant (betadisper, F 1,6 =0.854, p=0.3911) (Fig. 4).For the latter analysis, a higher distance from the group centroid is generally expected in areas with higher beta diversity (Anderson et al. 2006).

Evolutionary states of the local molluscan biota
According to literature data, museum material and field samplings, the Sabaudia Lake molluscan biota has gone through different evolutionary states, confirming with biological assessments the historical traced human driven shift from a freshwater to a strongly marine-influenced lagoon ecosystem.Records of subfossil taxa suggest that previous shifts to environmental features similar to those prevailing today have occurred in the  Callistoctopus macropus (Risso, 1826) Venerupis philippinarum (Adams A. and Reeve, 1850) Octopus vulgaris Cuvier, 1797 The evolution of the malacofauna of Sabaudia Lake • 657 past, according to historical evidence, and represent the only known biological trace of the construction and re-opening of the Roman Channel during the Roman Empire and around 1700.
The opening of the two sea water connections (around 1930) and the continuous freshwater flowthrough into the lake have kept the salinity below 20‰ during the investigated period (Brunelli andCannicci, 1944, Ferrero 1961) and the analysis of the species recorded in 1933-1958, and of their spatial distribution, shows an early colonization of vacant niches by a few ecologically tolerant species.
About 40 years after its definitive geomorphologic conversion to a coastal lagoon, anthropogenic pressures on Sabaudia Lake peaked in the late 1970s.Sewage pollutants from the urban area were mainly discarded within the Annunziata arm, also leading to a high dystrophic crisis, and only since the 1980s has the area reached a steady state.Freshwater flow coming into the lake from drainage canals was, however, strongly reduced to meet agricultural needs, resulting in salinities higher than 30‰ (Perdicaro 1984).A progressive settlement in the area of marine molluscan species was noted (see species recorded during the 1966-1995 period), but the local introduction of several "native" Mediterranean species highlights intriguing questions regarding the strong uniformity in species composition of the molluscan fauna of the Mediterranean lagoon, and calls for the necessity of molecular analysis to elucidate the taxa distributional pattern not only on a global scale but also on a more regional one.
Finally, more than 100 species were recorded during the 2005-2011 period.Indeed, this may have been induced by the strongest searching effort ever achieved in the study area, as well as by the increasing taxonomical knowledge and the planned sampling design (covering all the different habitat types and first including the Roman Channel sea-lake connection).However, the data obtained in the present study confirm a definitive shift to a strongly marine-influenced lagoon ecosystem.

Current molluscan assemblage: faunal and ecological insights
The two clusters of sites identified in this study (Fig. 3) reflect a distribution pattern in agreement with the "confinement gradient" (Guelorget andPerthuisot 1984, 1992), which predicts a decreasing trend of species richness from less confined to inner lagoon zones.The Inner Cluster was mostly composed of a subset of species also observed in the Channel Cluster, except Paludinella globularis, Hydrobia acuta, Islamia pusilla, Myosotella myosotis and Tectura virginea (Tables 3, 4).Among them, species belonging to the Assimineidae and Hydrobiidae are well known as freshwater and estuarine inhabitants (Hershler andFrest 1996, Fukuda andPonder 2003), and their presence only at S3 and S4 may be justified by a stronger freshwater flow that is absent at S2 and lower at S1 because of marine waters coming from Caterattino.Within the four internal sampling sites, the most represented families among gastropods were the herbivorous Trochidae and the carnivorous-necrofagous Muricidae and Nassaridae, whereas among bivalves they were Mytilidae and Veneridae (Table 3).Indeed, S2 and S4 share a high presence of bivalves (mainly Veneridae and Semelidae) because of the higher availability of a burrowing substratum, while S1 was partially separate from S2-S4 (Fig. 3).The scarce presence of hard substrata available for sessile (Crassostrea sp.) or partially sessile (Mytilidae, Mimachlamys varia) taxa, as well as of some herbivorous species, may be a possible explanation for this finding.In addition, the prevalent NW winds generate a fleeting current to the SE that is supposed to influence S1's ecological conditions, drifting marine water from the northern outfall to this spot: this is presumably at the base of the presence of a Pinna nobilis population in the area.Altogether, the low renewal of the marine element in the inner part influences environmental variables such as salinity, organic matter content, oxygen levels, temperature and grain size, which encompasses a restricted faunal composition (e.g.Gray 1974, Reizopoulou and Nicolaidou 2004, Basset et al. 2008, Magni et al. 2008).
Conversely, the inlet areas of the lagoon are influenced by both lake and sea water flows, and daily water exchanges within the Roman Channel imply that pelagic organisms, including molluscan larvae, constantly replenish the biotope.These environmental conditions favour a high species diversity (Carvalho et al. 2005, Koutsoubas et al. 2000), with molluscs often described as the most diverse faunal group in the sea inlet of marine lagoons (Millet and Guelorget 1994).This finding is in agreement with the high number of species observed within the Channel Cluster (S5-S8), considering that this very tiny spot (0.27 km 2 ) hosted 113 taxa, and 82 were only recorded within it (Table 3).A molecular approach to the sampled species may clarify whether most of them occur as established populations (including some forms of resistance or resilience) or casual, but repetitive, occurrences.The S6 and S8 sites were indeed influenced by saline waters, as confirmed by their dominant macrofauna (Table 1), while S5 is a moderately marine-influenced zone, being a transitional zone between paralic and marine areas of Sabaudia Lake.Although 17 taxa were widely distributed in all four sampling sites (Table 3), differences among sampling sites were appreciable.Sites S6 and S7 shared the exclusive presence of some Rissoidae, Flabellina spp., Diodora graeca and Jorunna tomentosa, while S8 alone accounted for 70 taxa, and 15 were found only within it (Table 3).Among them, there are typical marine species such as Nassarius mutabilis, Acanthocardia tuberculata, Solen marginatus, Ensis minor, Donax semistriatus, Donax trunculus, Solecurtus strigilatus and Callista chione.
Finally, Terlizzi et al. (2009) showed that the use of presence-absence data is a valid approach for assessing beta diversity in molluscan assemblages.Nonetheless, our data show that this is not always possible between sites with clear differences in species richness.Results of the Wilcoxon test suggest a higher degree of homogeneity in species composition between sites at Sabaudia Lake.However, it seems that departures from homogeneity in species composition between sites within Sabaudia Lake created a wider range of multivariate dispersion than in the case of the Roman Channel.As such, sites from the Roman Channel showed a more homogeneous faunistic composition, while a higher range of dissimilarity between sites within Sabaudia Lake resulted in a higher distance from the group centroid (Fig. 4).

Special emphasis on alien species
Italian hotspots of introduction and secondary spreading of alien species were recently highlighted by the SIBM Allochthonous Species Group and Anna Occhipinti-Ambrogi (2010), who did not include any location in Latium.Our samplings showed that Venerupis philippinarum and Crassostrea specimens were among the most widespread species in the study area, the former being recorded at seven sites and the latter colonizing most of the available natural and artificial banks (Table 3).A wider study for the occurrence of Crassostrea oysters in the CNP marine area revealed its presence in the Circeo harbour (Latium, Tyrrhenian Sea) too, where it can be commonly found attached to the dock walls (Macali, unpublished data).Recent records of Haminoea japonica from Sabaudia Lake have considerably expanded its known Mediterranean range to the Tyrrhenian Sea (Crocetta et al. 2013) and, according to our field observations, this species has completely replaced populations of other native Haminoea species.This finding is paralleled by a similar observation by Hanson et al. (2013), who reported the complete replacement of the once common Haminoea vesicula (Gould, 1855) in Boundary Bay (North America).The records of Arcuatula senhousia confirm the presence of this species in the area (see Cossignani and Ardovini 2011).Its presence at S8 only (Fig. 1 and Table 3) would suggest a colonization of the area from the Roman Channel and not an introduction by mariculture.We are currently not aware of any further record from the Pontine Lagoons, but monitoring its local spread may be a future target.The presence of Godiva quadricolor and Bursatella leachii was first reported from Latium.Godiva quadricolor has only been reported in the Mediterranean Sea from the Gibraltar Strait area and Fusaro Lake (Campania, Tyrrhenian Sea) (Cervera et al. 2010), although its further presence in the Adriatic Sea (Piallassa Baiona, Ravenna) is also first reported herein (Rinaldi 2012 as Facelina auriculata).This would suggest that a correct assessment of its distribution may be mostly concealed by taxonomic impediments and the absence of field research.On the other hand, Bursatella leachii has been only recorded in the central Tyrrhenian Sea from Tus-cany, Sardinia and Campania (Crocetta et al. 2009).Finally, four individuals of the slipper limpet Crepidula fornicata were found attached to Crassostrea: this is the second record of this species in the CNP wetlands after its first record from Caprolace Lake (Bini 1983b).However, no established populations have been found in the area, so the species is conservatively considered as casual (Crocetta et al. 2013): these specimens may be the result of periodical re-introduction of Crassostrea oysters in the lake, presumably from the Atlantic shores.The same can be said for Rapana venosa: despite its recent collection in 2005 (Crocetta and Soppelsa 2006), the absence of further records suggests that its ephemeral introduction may be connected with mussel imports from the northern Adriatic Sea.synonym of Rissoa membranacea (Adams J., 1800) on the basis of the analyses of the type material (Coen 9652 -HUJ 53487) and of specimens stored in the private collection of the authors and belonging to this taxon.The taxon listed as "Rissoa sp." by Ferrero (1961)  Nassarius nitidus (Jeffreys, 1867) Literature records.Hinia reticulata (Linnaeus, 1758) -Bini 1987: 116-117, Fig. 4B (+).

Fig. 1 .
Fig. 1.Study area. A. Location of Sabaudia Lake (Italy, Tyrrhenian Sea) in the Mediterranean Sea.B-C.Sabaudia Lake.Map of the sampling sites: numbers as inTable 1.

Fig. 3 .
Fig. 3. Non-metric multidimensional scaling plot showing the ordination of the sites as well as the groups identified by cluster analysis: Inner and Channel Clusters.

Fig. 4 .
Fig. 4. Distance from the group centroid of the Channel and Inner Clusters of sites, using the Jaccard index as a dissimilarity measurement.The horizontal line within each box shows the median, the bottom and the tops of the box shows the 25th and 75th percentiles, respectively.The vertical dashed lines show either the maximum and minimum value or 1.5 times the interquartile range.

Table 2 .
(Crocetta 2012, Crocetta et al. 2013he Sabaudia Lake (from the literature only).A record is marked with an asterisk (*) if a local introduction or translocation has been traced in the literature: see Appendix 1 for further details.Bold: alien species in the Italian territorial seawaters(Crocetta 2012, Crocetta et al. 2013, see Appendix 1 for discussion on C. angulata).
not.With the sole exception of Crassostrea angulata, locally introduced in 1966, the records hereby listed come from the 1974-1995 period.Rissoa membranacea disappeared from the area, though it had been widely recorded during the previous research, and Haminoea navicula was replaced by Haminoea orbignyana (Table

Table 4 .
(Crocetta 2012, Crocetta et al. 2013served in both clusters (MCSBC), the most frequently observed species exclusive to the Channel Cluster (MFSCC) and exclusive species of the Inner Cluster (ESIC), listed in systematic order within each cluster.Bold: alien species in the Italian territorial seawaters(Crocetta 2012, Crocetta et al. 2013).