Missing species among Mediterranean non-Siphonophoran Hydrozoa

Hydrozoa of the Mediterranean Sea are well known and a recent monograph covers 457 species. Mediterranean non-Siphonophoran Hydrozoa comprises 398 species, an increasing number due to continuous updates, representing about 10 % of the 3,702 currently valid species reported in a recent world assessment of hydrozoan diversity. Many new records are non indigenous species, previously described species that occurred elsewhere and whose arrival was presumably caused by human activities. However, many species reported in the past are not recorded in recent times. Realistic assessments of species pools require addition of new species, but also subtraction of species not found since a certain period. With the confidence of extinction index, cases of putative extinction can be raised. Out of the 398 known species, only 162 (41 %) have been reported in the last decade, while 53 (13 %) are not recorded in the literature since at least 41 years. According to the confidence of extinction index, 60 % of the 53 missing species are extinct, and 11 % are putatively extinct from the basin. From a biogeographical point of view, the missing species are: 34 % endemic, 19 % boreal, 15 % Mediterranean-Atlantic, 11 % Indo-Pacific, 11 % circumtropical, 4 % cosmopolitan, 2 % tropical-Atlantic, 4 % non-classifiable. Fluctuations in species composition into a certain area cause heavy variability in the expression of both structural and functional biodiversity. As consequence, the regional biodiversity should be analyzed through its temporal evolution, to detect changes and their possible causes. This approach has profound consequences on biodiversity assessments and also on the compilation of red lists.


Introduction
The question ''How many species are there in the oceans?'' provides the key to discover what we know and what we do not know about the life in the seas (Mora et al. 2011). Conservation biologists try to identify the areas in the world where effective conservation actions could protect as many species as possible. The knowledge of species, however, is incomplete since many species are still unknown (Costello et al. 2013a, b) or poorly known. Myers et al. (2000) claim that biodiversity hotspots, areas characterized by high numbers of endemic species as well as high rates of habitat loss, are prioritary, but a question remains: does the cumulative evaluation of biodiversity, in terms of species additions through time, really represent the expression of biodiversity at a given place? Appeltans et al. (2012) compiled WoRMS, the World Register of Marine Species (about 226,000 eukaryotic marine species) and used it as a starting point for estimating how many more species may still be discovered. WoRMS published online information on marine species, but many nomenclatural and classification problems remain (Costello et al. 2013a, b). The introduction rate of synonyms is expected to decline through updated taxonomic revisions (Appeltans et al. 2012).
A further problem with the estimation of biodiversity is that local lists are usually updated by adding new entries, but locally (or even finally) extinct species are seldom, if ever, removed from the lists (Boero and Gravili 2013), a task that only taxonomists can undertake through the critical analysis of species lists and the identification of putatively extinct species.
The sea has been far less studied than the land, and our taxonomic knowledge of many groups remains fragmentary (Hilchey 2003). Attempts to inventory all known species led to cover about two-thirds of all marine species (Appeltans et al. 2012), and half of all species (Bisby et al. 2009). Species lists and their distribution are basic to biodiversity research (Costello et al. 2001). May (1994) and Hammond (1994) reviewed a variety of approaches to predict the number of species that may exist on Earth. Moreover, Costello and Wilson (2011) proposed to predict the number of known and unknown species in European seas using rates of description. Biodiversity research has a long history the Mediterranean Sea, one of the best-known seas globally (Coll et al. 2010;Gravili et al. 2013). In particular, the diversity of Mediterranean Hydrozoa is well known and has been recently updated (Bouillon et al. 2004(Bouillon et al. , 2006Schuchert 2005Schuchert , 2006Schuchert , 2008aSchuchert , b, 2009Schuchert , 2010Galea 2007;De Vito et al. 2008;Gravili et al. 2007Gravili et al. , 2008Gravili et al. , 2010Gravili et al. , 2013Morri et al. 2009;Mastrototaro et al. 2010). The biodiversity of the Mediterranean Sea is high due to ecological, historical, and paleogeographic reasons (Sarà 1985;Bianchi and Morri 2000;Bianchi 2007). The western Mediterranean has strong Atlantic affinities, due to the continued penetration of Atlantic species (Harmelin and d'Hont 1993). Conversely, after the opening of the Suez Canal, the Eastern Mediterranean is receiving species from the Red Sea (Galil 1993). The number of Lessepsian species, now acclimated in the Mediterranean (Golani 1998), is so high that Por (1999) proposed a separate biogeographic province for the Levant Sea.
Ecological and biogeographic theories, supported by significant data, predict that half of all present species may be extinct within the next 100-300 years due to climate change, pollution, over-harvesting, habitat fragmentation and loss (Chapin et al. 2000;Jackson 2008;Costello and Wilson 2011). It is often claimed that extinction rates are on the increase both on land and in the oceans (Carlton et al. 1999;Dulvy et al. 2003;Costello and Wilson 2011), and that chances are good that species might go extinct even before a formal description (Costello et al. 2013a, b).  stressed how well documented marine extinctions usually concern conspicuous species (e.g., the Caribbean monk seal Monachus tropicalis, the great auk Pinguinus impennis, and the Steller's sea cow Hydrodamalis gigas), and that the number of proven marine extinctions is very low, if compared with the alarming predictions of most review. This is not due to lack of extinction risks but, instead, to poor knowledge of the conservation status of most species (Roberts and Hawkins 1999;Régnier et al. 2009). , however, claimed that the analysis of the history of the records of each species in space and time might be conducive to roughly assess their state of conservation.
Changes in both the abundance and the distribution of species commonly happen due to the arrival of new species, the rarefaction of common species, or the increase in the abundance of formerly rare species (Boero 1994(Boero , 1996Bonsdorff et al. 1997). These changes are a natural feature of all systems but the rate of change can become alarmingly fast (Boero and Bonsdorff 2007). Biotic assessments are increasingly carried out to detect NIS Katsanevakis et al. 2013), and might be used also for the purpose of testing hypotheses of putative extinctions.
The aim of this paper is to review the knowledge of the diversity of Mediterranean Non-Siphonophoran Hydrozoa (NSH), to detect species that are absent since decades, the ''missing species'', so as to assess current estimate of the species pool-size and raise cases of either regional or local extinction.

Methods
The choice of 41 years as a threshold to consider a species as missing was decided based on the rather intense study of hydrozoan species in the Mediterranean in the last four decades, with the establishment of the Hydrozoan Society in 1985 (Boero 2007) that gathered a rather substantial scientific community focusing on the Mediterranean. Due to intensive sampling, thus, if a previously reported species fails to be recorded chances are good that, at least, it is more rare than before. The knowledge about each species is stored in the scientific literature. Every known species has been described in a taxonomic paper, and the date of its first finding is the beginning of the history of its knowledge. The type locality is the centre of origin of that species, even though it might not be representative of the core of its actual distribution. After the original description, species are usually recorded again in other taxonomic, faunistic, or ecological papers. Analyzing the temporal and spatial distribution of species, as recorded by the scientific literature, we can reconstruct maps of their recorded presence in both space and time. We examined current estimates of the size of the Mediterranean species pool, to detect species that might have gone locally or regionally extinct. Picard (1958a) produced the first modern list of Mediterranean NSH. Since then, the number of species almost doubled due to addition of new records to the new ones. To assess the current state of the Mediterranean species pool with the state of the fifties, we compared Picard's list with the list of the species recorded in the last decade.
Our list of non-Siphonophora Hydrozoan ''Missing'' species (NSHMs) of the Mediterranean Sea is based on a recent monograph (Bouillon et al. 2004), on taxonomic revisions (e.g. Schuchert 2007Schuchert , 2008aSchuchert ,b, 2009Schuchert , 2010, and on an assessment of Mediterranean NSH . To determine historical series and distributions, we consulted 749 faunistic studies published between 1850 and 2014. A database with 8,158 records was organized so as to provide the following information: species, family, author, life-cycle phase, reproductive state, location, date of collection e/o year of publication of the article, water depth, substrate type, synonymy, and cited references. Taxonomic records (i.e. records of each taxon, in any kind of report) are reported on a time scale from the original description to the last citation in the literature. The number of faunistic articles on Mediterranean Hydrozoa since 1850 was organized by decade (Fig. 1). The total number of articles (within the same time range) was then referred to each biogeographic sectors (A-M) identified by Bianchi (2007) (Fig. 2).
We identified NSHMs (not recorded since 41 years or more) examining records from the nineteenth century to 2014, to trace the origin, first and last Mediterranean records, current Mediterranean distribution, and global distribution of each species. With few exceptions, we named taxa according to Bouillon et al. (2006). The date and location of the first observation of each NSHMs in the Mediterranean Sea were extracted from the literature. Whenever possible, the actual date of first record was reported, along with its publication date, since the two dates coincide only in a few cases. Sadler (1987, 1989) introduced the confidence of extinction index in paleobiology (Marshall 1990), as a method to calculate confidence intervals within local stratigraphic ranges.  adapted this method to analyse cases of putative extinction in recent species. The confidence of extinction index was calculated for each species uncited since 41 years by using the following formula on historical taxonomic data: C is the confidence of extinction, G is the number of years since last sighting, R is the number of years between original description and the last sighting, H is the number of individual years in which there is a record, C C 95 % postulates a case of extinction; 80 % B C B 94 % raises a case of putative extinction. , from 20 to 11 years ago (20 years), and from 10 years ago to nowaday (10 years)]. A distance matrix based on Jaccard's distance among sector 9 period centroids was then obtained. A canonical analysis of principal coordinates (CAP) (Anderson and Robinson 2003;Anderson and Willis 2003) based on the distance matrix was then performed for the factor period, in order to portray temporal changes in the whole Mediterranean species pool of NSH. Distinctness among locations was assessed using leave one-out allocation success (Anderson and Robinson 2003). Species most contributing to group differences in the CAP plot were investigated by calculating product-moment correlations (r) of original variables (species) with canonical axes (Anderson and Willis 2003). Only species with correlation values exceeding an arbitrarily chosen value of correlation r C 0.2 were considered.
The 'average taxonomic distinctness' (D ? ) (Clarke and Warwick 1998) and 'variation in taxonomic distinctness' (K ? ) (Clarke and Warwick 2001), complementing D ? , were employed to explore temporal changes in the taxonomic structure of NSH species pool in  Bianchi (2007). See map legend (Fig. 3) for abbreviations the whole Mediterranean basin. D ? represents the average taxonomic path length between two randomly chosen species in the taxonomic tree, whereas K ? reflects the unevenness in the taxonomic tree of a given species' list and represents the variance of these pair-wise path lengths. The indices are independent of the number of species in a sample and thus represent useful tools for analysing historical data (Bevilacqua et al. 2009). A reference list, from species to subclass, was made including all NSH species recorded. The list coupled with the presence/absence data matrix was used to calculate the values of D ? and K ? of Mediterranean NSH species pool in each period. The same step length (equal to 1) was used in weighting all distances between hierarchical taxonomic levels (Clarke and Warwick 1999). For both taxonomic distinctness indices, the 95 % confidence funnel was generated Warwick 1998, 2001) in order to test temporal departures from expectations of D ? and K ? (under the null hypothesis that the species pool in each period was a random subsets of the full NSH species list).

Results
The updated list of NSH species, after an accurate systematic revision, sums up to 398 species, representing about 11 % of the 3,702 nominal known species of the superclass Hydrozoa reported by Bouillon et al. (2006). The species recorded from the Mediterranean Sea in the last decade sum up to 162, and 118 of them (73 %) are present in Picard's (1958a) list of 191 species (180 valid species if cleaned up by synonyms); 53 species (13 %) are not recorded in the literature since at least 41 years (Table 1).

Taxa
Class, subclass, species C Confidence of extinction index (C C 95 % to postulate a case of extinction; 80 % B C B 94 % to raise a case of putative extinction) ? Psammocoryne Monniot, 1962 (invalid nomen nudum). Monniot (1962) identified it as Psammocoryne. This name is not a valid genus as it was not associated with a valid nominal species. Furthermore, Monniot's hydroid could easily also be referred to A. ilonae and it is therefore also somewhat unclear whether A. ammisatum also occurs in the Mediterranean (Schuchert, 2006) Sea, based on unclear criteria, Schuchert (2009) suggested that it is a doubtful, unrecognizable species. Finally, Bouillon et al. (2004) listed C. polygonia as doubtful. Uncertain records concern species as Eudendrium arbuscula, whose Mediterranean records are likely misidentifications (Schuchert 2008b); the records of Podocoryna borealis are unreliable according to Schuchert (2008a). The presence of Octotiara russelli in the Mediterranean Sea is uncertain. Goy (1973) published the only European record of this species, as Octotiara violacea, but this reporting should be re-examined due to the state of preservation of the specimen that impedes certain identification (Schuchert 2007). Tiaropsidium mediterraneum was recorded for the first time in Messina (Metschnikoff 1886a) as Tiaropsis mediterranea, whereas its record in the South Adriatic (see Piraino et al. 2013) is doubtful.
Moreover, particular problems are related to records of the micro-meiobenthos NSH species that might be underestimated due to paucity of research in this field, namely: Acauloides ammisatum (whose presence in Mediterranean Sea is unclear; see Schuchert 2006), A. ilonae, Psammohydra nanna (whose taxonomic position is unclear; see Schuchert 2006), Armorhydra janowiczi, and Halammohydra octopodides. A particularly significant example of species that is absent since a very long time is Tricyclusa singularis (Schulze 1876). This species of boreal affinity and, since its original description from Trieste, the sole Mediterranean record, it has never been recorded again from the Mediterranean Sea. Its disappearance represents not only a case of Mediterranean extinction of a species, but also of the whole family Tricyclusidae that comprises only this species and genus (Boero and Bonsdorff 2007).
Studies of the Mediterranean Hydrozoa suffered several temporal gaps during the considered period (Fig. 1). The whole trend, expressed in number of papers per decade, can be divided into five periods, marked by changes in the patterns of scientific production ( Fig. 1): -1850s-1870s, with an average of over 10 papers/decade; -1880s-1910s, with an average of about 30 papers/decade: about 20 papers/decade in the sub-period 1880s-1890s, and about 40 papers/decade in 1990s-1910s, with an increase of scientific production until a peak in the 1910s (51 papers) followed by a sharp decrease due to First World War; -1920s-1940s, (average of over 25 papers/decade) with a marked decrease coinciding with Second World War; -1950s-2000s, with an average of almost 80 papers/decade; -2010s-2014s, with an average of about 55 papers/decade, but monitoring of the entire decade (2010-2020) is still incomplete. Figure 2 shows that since the 19th century many studies were carried out at the Zoological Station of Naples (biogeographic sector C). Messina also attracted high attention (biogeographic sector M), due to the strong currents of its Strait characterized by animals of deep waters. Kölliker (1853), Keferstein and Ehlers (1861), Metschnikoff (1886a, b), worked extensively at Messina contributing to the knowledge of the Hydrozoa. Other Mediterranean places where research on Hydrozoa became prominent were Trieste and Rovinj (biogeographic sector F), Split (biogeographic sector G) and, in France, Villefranche-sur-Mer, Endoume, and Banyuls (biogeographic sector E). In particular, a long series of papers mainly by Picard (1951aPicard ( ,b, 1958b and Goy (1973) gave a great contribution to the knowledge of the Hydrozoa. Moreover, between the years 1960s and 1970 s several researchers (among these, Bouillon, Brinckmann-Voss, Haeckel, Tardent, Uchida, Vannucci, Yamada) worked at the Naples Zoological Station to describe the life cycles of Hydrozoan species.
Numbers of Non-Siphonophoran Hydrozoa Missing species (NSHMs) in each biogeographic sector within the Mediterranean Sea is shown in Fig. 3  Results form analyses on taxonomic distinctness highlighted a decrease of both D ? and K ? of the Mediterranean NSH species pool through time (Fig. 5). The species pool of [40y showed significantly higher values (P \ 0.05) of D ? and K ? , indicating a higher breadth and heterogeneity of taxonomic structure. In contrast, the species pool in the last decade (10 years) exhibited values of D ? and K ? significantly below random expectation, indicating that Mediterranean NSH species were more closely related than expected by chance, with a significant reduction of taxonomic distinctness (Fig. 5).

Discussion
The scarcity of well-documented cases of extinction in the marine environment shows how difficult it is to deal with the conservation status of marine invertebrates ).  Bianchi (2007). For each sector, NSHMs percentage and NSHMs number/non-Siphonophoran Hydrozoa (NSH) total number are shown Boero and Bonsdorff (2007) wondered if this is the consequence of low global risks of extinction in the sea or, rather, if we fail to notice that species become extinct. According to Roberts and Hawkins (1999), there might have been numerous extinctions in recent times that we failed to realise. Fontaine et al. (2007) addressed the problem of the current indicators that do not cover the species at risk of extinction, as most of rare species are not considered in the European Union's 2020 target. Alternative indicators about rare species must be developed, in addition to the existing ones that provide information on biodiversity trends (Butchart et al. 2005;De Heer et al. 2005). Indeed, the choice of indicator species should be expanded through a rigorous assessment based on various parameters which take into account also rarity (Fontaine et al. 2007). Moreover, the fundamental question is how soon such changes will occur (Hughes 2000), as well as the particular time ranges chosen for the data sets can greatly affect apparent trends (Hughes 2000). Carlton et al. (1999) observed that the processes of species extinction run at different paces, involving several mechanisms working at different spatial scales. In general, the three main changes in response to environmental stress of the marine communities consist in regression to dominance by opportunist species, reduction of the dominating species resulting in lower diversity (Pearson and Rosemberg 1978;Gray 1989). The features of species that have gone extinct or are nearly extinct (population turnover, reproduction, capacity for recovery, range and distribution, commonness and/or rarity, trophic level) often contribute to their disappearance (Dayton et al. 1995;Roberts and Hawkins 1999).
The Mediterranean Sea is predisposed to local extinction because it is almost closed and much smaller than the open ocean, responding more quickly to environmental change and, furthermore, has a high rate of endemism (Boero and Gravili 2013). This sea is characterized by a particular biota made of highly seasonal species, tropical and boreal contingents being present respectively in the summer, and in the winter (Bavestrello et al. 2006). The Mediterranean marine ecosystem, being subjected to a period of temperature increase that is tropicalizing its biota, represents a model basin for oceans and other seas (Bianchi 2007;Boero and Bonsdorff 2007;Lejeusne et al. 2010).
It is very difficult to confirm the disappearance of a species in the marine environment, mostly due to lack of taxonomist and the existence of synonyms in the species lists (different names attributed to the same species). Therefore, simple lack of suitable sampling or of expertise in recognizing synonyms in previous samplings, might determine their absence from subsequent records. It is debatable whether lack of records is due either to changing of abiotic or biotic factors, or to low sampling efforts or, eventually, to the combination of these causes. Surely, the Mediterranean Sea is going through a radical change that is almost unparalled in respect to any other part of the world (Boero 2014;Templado 2014).
The absence of a species, furthermore, might be only apparent, due to the existence of resting stages that can remain dormant for long periods and that, when activated, are responsible for the so-called ''Lazarus effect'' (Jablonsky 1986).
As expected by Boero et al. (2008), global warming is favouring the tropical contingent, whereas the boreal one is in distress. If global warming can damage species, the potential sufferers are Mediterranean endemic species (34 % of the NSHMs), those of cold water affinity (19 %) or Mediterranean Atlantic ones (15 %). Indo-Pacific and circumtropical contingents represent each 11 % of the total NSHMs, extinctions in the Mediterranean being probably linked to lack of establishment of species that recently reached the basin. The results of this study confirm the trend characterized at a first time by the abiotic change, induced by increasing temperatures, and followed, at a later time, by biotic change, since the arrival of aliens of tropical affinity (Zenetos et al. 2012;Gravili et al. 2013;Ç inar et al. 2014), or the prevalence of the summer contingent that competes against the species of cold water affinity ).
The results of this study, as well as the data analyzed over the long term by Puce et al. (2009), suggest that the regional species pools tend to remain stable in terms of species numbers but not in terms of species identity. In fact, the number of present-day Mediterranean NSH species (162) matches closely the number of species (180 valid species) that Picard (1958a) recorded in his first assessment, cumulating all previous knowledge on the group. However, our finding showed that the composition of the species pool at basin scale changed through time, with changes heavily driven by NSHMs and NIS. Moreover, we detected a progressive contraction of the taxonomic width of NSH, imputable to the loss of taxa poor in species or monotypic, which raises concerns about potential ensuing depletion of taxonomic and functional diversity.
The tropical NIS, colonising the Mediterranean Sea, are probably filling the ecological spaces of species that are becoming rare or are locally extinct. Bianchi (2007) predicted that the northern areas of the Mediterranean Sea will be invaded by warm-water native species, while the southern areas of the basin will be occupied by tropical exotic species. Furthermore, the warming of the Mediterranean Sea might probably cause a decrement of native cold-water species, or even their disappearance (Bianchi 2007). Boero et al. (2008) proposed the so-called 'cold engines' (the northernmost part of the Western Mediterranean, the Northern Adriatic, and the Northern Aegean) as the areas with greater probability of presence of putatively extinct species in a period of global warming. These places, the drivers of the vertical remixing of Mediterranean waters, are significantly colder than the rest of the basin. They are inhabited by many species of cold-water affinity. The compilation of lists of species for all significant taxonomic groups that live only in these areas might provide a tool for creating lists of putatively extinct species, and allow the programming of surveys to ascertain their conservation state (Boero and Gravili 2013). The results shown here suggest that species lists are dynamic, requiring continual updating (introduced species) and putative subtractions of missing species. Without these subtractions, biodiversity is always on the rise due to the arrival of NIS and the species lists will never show possible biodiversity crises at the level of species pools. Mendelson et al. (2006) required an unprecedented conservation response to stop the loss of species and populations. The rates of marine species description, driven by the increasing ability to explore previously unknown geographic areas, have never been higher, as well as the challenge to estimate the diversity of cryptic species through molecular studies (Appeltans et al. 2012). The rapid influx of NIS and the disappearance of the species of cold-water affinity are heavily influencing the rich but vulnerable Mediterranean ecosystem, heavily affected already by a host of multiple impacts (Claudet and Fraschetti 2010;Boero 2014).
The application of the present analysis to all other taxa will allow for a better assessment of the state of biodiversity in all seas and oceans.  ([40, 40, 30, 20, 10 years) plotted against the corresponding total number of species characterizing each period. For both indices, the expected mean (dotted line) and the 95 % confidence limits (solid lines) were also plotted from 1,000 independent simulations drawn randomly from the full list of Mediterranean NSH species