The effect of salinity on larval development of Uca tangeri (Eydoux, 1835) (Brachyura: Ocypodidae) and new findings of the zoeal morphology

in order to evaluate their tolerance to low salinities, zoeae of the fiddler crab Uca tangeri from the rio San pedro population (southwestern Spain) were reared in the laboratory at 20oC and at three salinities (16, 24 and 32). the zoeal development was completed at 24 and 32 but the crabs died as zoea i or zoea ii, and very rarely as zoea iii, at 16; tolerance to low salinities varied among clutches produced by different females. the duration of the first zoeal stage and of the complete zoeal development was shorter at 32. our observations showed that the zoeae of U. tangeri could not tolerate retention in the mesohaline water of estuaries, and that export to oceanic waters would be optimal for their successful development. Survival at 24 suggests that larvae could also develop in polyhaline conditions if they were retained in the ocean-estuary interface. the presence of an additional zoeal stage (zoea vi) was observed in some individuals and associated with unfavourable combinations of temperature and salinity. in addition, some previously omitted aspects of zoeal morphology were re-described and illustrated, providing new evidence which supports the basal position of this species in a proto-atlantic origin of the genus Uca. it is proposed that the differences between Uca tangeri and the rest of the Uca species should be highlighted, and that it should be placed in its own genus as Afruca tangeri.

Palabras clave: desarrollo larvario, salinidad, mortalidad, duración del desarrollo, ocypodidae, Uca, cangrejo violonista, morfología de zoea.iNtroduCtioN the ontogeny of most estuarine crabs includes a free larval phase.two developmental strategies were described in larvae of species whose adults live in brackish waters: retention in the parental habitat, and exportation to the more stable conditions of coastal or oceanic waters (anger, 2001).When the tolerance range of larval stages to salinity is narrower than that of adults, the exportation strategy allows larvae to develop in a less stressful environment.in addition, once in the ocean, exported larvae may face a reduced predation pressure and play an important role in the dispersal of a species.the effects of salinity on the larval biology of decapods were exhaustively reviewed by anger (2003).However, the tolerance to reduced salinities is not an "all or not" trait: it has been demonstrated that genetic variability at the clutch level, individual variability in the provisioning of eggs with yolk, and salinity prevailing during embryonic development could affect salinity tolerance during crustacean development (e.g. Lee and petersen, 2002;Giménez and anger, 2003;devreker et al., 2004).
Fiddler crabs (genus Uca; Brachyura: ocypodidae) consist of 94-97 species (depending on authors) of semi-terrestrial intertidal crabs that excavate burrows into muddy or sandy substrates and are characterised by extreme cheliped asymmetry in males (rosenberg, 2001).they occur in marine and brackish waters of temperate and tropical regions in the western atlantic (21 species), the eastern pacific (36 species), the indo-West pacific (39 species), and the eastern atlantic (1 species: Uca tangeri).However, adults of at least one species (the western atlantic Uca minax) can live in freshwater habitats as far as 50 km from the sea (Brodie et al. 2007).
the geographical distribution of U. tangeri ranges from southern Spain and portugal (~37ºN) through the west coast of africa, down to angola (~16°S); it is the only species of fiddler crab to be found in Europe and its ecology, physiology, and behaviour have been intensely studied (e.g.Burford et al., 2001;Jordão and oliveira, 2005;Wolfrath, 1993;krippeit-drews et al., 1989;Medina and rodríguez, 1992;rodríguez et al., 1997;von Hagen, 1987 and references therein).the complete larval development of U. tangeri was first described by rodríguez and Jones (1993) based on material from rio San pedro, an inlet in Cádiz Bay, SW Spain, and it consisted of five zoeae and one megalopa stage.the first zoeal stage had been previously described by paula (1985), who reported data missed in the description by rodríguez and Jones (1993( ). Later, rodríguez et al. (1997) ) studied the larval abundance patterns in the same site and suggested that larvae released in the inlet were exported to the bay.However, the tolerance of U. tangeri larvae to reduced salinities has not been studied.
the genus Uca was divided into nine subgenera by Crane (1975): Afruca (with a single species: U. tangeri), Amphiuca, Australuca, Boboruca, Celuca, Deltuca, Minuca, Thalassuca, and Uca. a phylogeny based on a 16S rrNa regrouped the species into 3 clades and suggested that the ancestral clade, which includes Uca and Afruca, is basal to the indo-West pacific clade and the derived american clade (Levinton et al., 1996;Sturmbauer et al., 1996).Later, rosenberg (2001) reduced the number of genera to seven, unifying Afruca + Uca and Boboruca + Minuca, but Beinlich and von Hagen (2006) established eight subgenera: Australuca, Cranuca subgen.nov., Gelasimus, Leptuca, Minuca, Paraleptuca, Tubuca, and Uca s. str.(which includes former Uca and Afruca).this schema was followed by Ng et al. (2008) in a recent revision of the systematics of crabs.Beinlich and von Hagen (2006) recognised that U. tangeri "clearly represents the earliest branch of Uca s. str.and is aberrant in its front width and geography".recently, molecular studies confirmed the validity of the monotypic subgenus Afruca, clearly separated from the other representatives of the subgenus Uca (Landstorfer, Felder and Schubart, unpublished).
the original objective of the present study was to evaluate the tolerance of Uca tangeri zoeae to low salinities, and to determine whether this tolerance varies among clutches produced by different females.While rearing larvae in the laboratory, we had the opportunity to carefully observe the zoeal morphology; on these grounds, we decided to re-describe and illustrate some previously omitted aspects that may be useful contributions to the knowledge of the phylogeny of Uca.Finally, by comparing the results of our larval cultures with those of rodríguez and Jones (1993), we proposed that the absence of populations on northern atlantic coasts may be related to a very low tolerance of zoeae to low temperatures.

Collection
in July 2007, ovigerous females of Uca tangeri were collected by hand at low tides in rio San pedro, a shallow inlet of the salt marsh zone in the eastern shore of the Bay of Cadiz, Spain (36º23'-37'N, 6º8'-15'W).Water temperature and salinity of rio San pedro in summer vary between 22.2 and 29.6ºC and 36.5 and 43.5, respectively (González Gordillo et al., 2003).Four ovigerous females with near-tohatch embryos (named F1-F4) were carried to the laboratory at the instituto de Ciencias Marinas de andalucía.Each female was reared at ambient temperature and photoperiod in individual aquaria filled with water from the inlet until hatching.

Rearing techniques
immediately after hatching (F1 and F2: 07/04/2007; F3, 07/05/2007; F4, 07/06/2007), only actively swimming larvae were transferred to individual cultivation vials (10 cm 3 ) with wide-bore pipettes.three salinities (16,24,32) obtained by dilution of filtered seawater (mesh: 0.45 μm) from rio San pedro with appropriate amounts of desalinated tap water and checked with a hand refractometer to the nearest pSu were tested.From each female, 24 larvae were placed in each salinity and maintained in cultivation chambers (20 ± 1ºC) with a controlled photoperiod of 16:8 h L:d.Larvae were acclimated by transferring gradually (8 h at 24) until the final salinity to avoid osmotic shock at the lowest salinity (16).Zoeae were fed with newly-hatched Artemia nauplii and Brachionus plicatilis (previously fed with nannochloropsis gaditana).Water and food were changed daily, and larvae were checked for mortality and moults during each water change.

Statistical procedures
a log-linear model was performed to test interactions between frequency of larval death of each zoeal stage with females and salinity as factors (Norman and Streiner, 1996;Zar, 1996).Number of factors and levels were variable due to the high mortality of larvae at low salinity.
differences between mean values of development duration of each zoeal stage and the duration of the complete development were tested by a mixed-factor aNova model with unequal replicates (Zar, 1996), with salinity as the fixed factor and females as the random factor.this test was used even with nonnormal data given that aNova is robust to departure from this assumption (underwood, 1997).Significant differences among treatments after aNova were tested with a Student-Newman-keuls (SNk) test. in all statistical tests, differences were considered significant when P≤ 0.05.rESuLtS

Mortality
With few exceptions, mortality decreased with increasing salinity, but varied among clutches; it was consistently lower in F2 and higher in F4 (Fig. 1).Larvae reached the megalopa stage at 24 and 32 but died as zoea i (F4), zoea ii (F1, F2 and F3) or zoea iii (one specimen, F2) at 16. a log-linear analysis showed significant interactions between females and mortality, and between salinity and mortality (table 1), indicating that both salinity and females affected the survival of zoeae.Norman and Streiner, 1996;Zar, 1996) applied to detect interactions between proportional mortality of each zoeal stage (Fig. 1) with female and salinity as factors (***: P<0.001; * P<0.5).

Duration of development
the duration of zoeal stages increased with development from 5.59 ± 1.09 to 10.48 ± 2.40 (all females pooled).individual larvae moulted after 4-9 days in zoea i, 4-13 days in zoea ii, 4-23 days in zoea iii, 6-17 days in zoea iv, 6-20 days in zoea v and 9-10 days in the few larvae that passed through a zoea vi stage.the average values of each stage duration in larvae from the four clutches (F1-F4) reared at different salinities are summarised in Figure 2. a twoway aNova showed that salinity and females had a highly significant effect on the duration of zoea i development (table 2) but no interactions appeared between factors.a post-hoc comparison allowed the duration to be ranked among salinities (16>24>36) and females (F1 and F3>F2 and F4).
the complete zoeal development of individual larvae lasted from 23 to 51 days.the minimum and maximum average values of different clutches (F1-F4) and salinities (24 and 32 psu) ranged from 31.6 ± 4.4 to 42.7 ± 7.6 days (F4, 32 and 24, respectively, Fig. 2). the duration of the complete development was significantly affected by salinity (aNova, table 2).

Zoeal morphology
the morphology of the 5 zoeal stages and megalopa is basically equal to those described by rodríguez and Jones (1993), with the exception that the telson of zoea i to zoea v had a small dorsal spine, 2 pairs of minute lateral spines, and 2-3 minute ventral spines on each furcal arm (Figs. 4, 5).two patterns were observed regarding the morphology of the few zoeae vi obtained.Some individuals showed the typical traits of an "advanced" developmental stage, combining zoeal and megalopal characters (i.e.dorsal setae on the telson plaque, natatory setae on abdominal pleopods), as well as a general increase in size and number of setae on the appendages (i.e. 14 natatory setae on exopods of first and second maxillipeds; 10 (4+4+2) aesthetascs and 2 setae on anten-  Zar, 1996) applied to detect differences between mean values of duration of each zoeal stage (Fig. 2), developed in 16 (zoea i), 24 and 32.Female and salinity were random and fixed factors, respectively, and stage duration the variable response.nule; 16-18 setae on ventral margin of cephalotorax).Further features observed were: endopod of antenna 3-segmented (2-segmented in zoea v), terminal segment of the endopod of first maxilliped with 6 setae (5 setae from zoea i to v). other individuals were morphologically similar to a "normal" zoea v; in fact, they originated from an "intermediate" previous stage (zoea v) that had some features of a zoea iv.No significant differences in morphology were found between megalopae originating from 6th or 5th zoeal stages. diSCuSSioN

Larval tolerance to low salinities
Larvae of decapod crustacean usually tolerate narrower salinity ranges than adults of the same species (Charmantier, 1998).adults of Uca tangeri, including the ovigerous females, can live in diluted warm or temperate waters, as Ewa-oboho (1993) reported for populations of port Harcourt (southern Nigeria), which live in habitats with salinities ranging from 10.8 to 16.2.in addition, adults are hyper-hyporegulators (krippeit drews et al., 1989).However, larval development is strongly affected by low salinities. in our study, larvae of Uca tangeri from the rio San pedro population completed the development at 24 and 32 but died as zoea i or zoea ii at 16. the highest survival was observed at 32, but even in this optimal condition differences between clutches from different females were observed.the duration of the first zoeal stage and of the complete zoeal development was shorter at 32. a similar pattern was observed in larvae of Uca pugnax from the northwestern atlantic salt marshes reared at salinities of 10, 20 and 30: they were unsuccessful in moulting at 10 and had a slight delay in moulting at 20; the megalopal stage, however, was shorter at 20 (o'Connor and Epifanio, 1985).on the contrary, salinity-induced mortality was not detected in the zoea i of U. minax originating from a freshwater population during their long trip (>50 km, several days) to the sea.these larvae survived longer than those spawned by females from a brackish water population; a physiological acclimation during embryogenesis may explain the observed differences (Brodie et al., 2007).
the development of U. tangeri larvae hatched in rio San pedro is likely to occur in Cádiz Bay, and also in the open sea, via larval exportation from parental waters (rodríguez et al., 1997).the export strategy may be a consequence of different selective forces: physiological (i.e.low tolerance to diluted water) and ecological (i.e.predator avoidance).our observations show that the zoeae of U. tangeri could not tolerate retention in the mesohaline water of estuaries, and that export to oceanic waters should be optimal for their successful development.However, the larval export from rio San pedro may be better explained by ecological factors since water is euhaline or hyperhaline in this particular habitat, at least during summer (González Gordillo et al., 2003).the small and vulnerable (i.e. with small dorsal and rostral cephalothoracic spines) zoeae of several Uca species emigrate from estuaries, usually with largeamplitude nocturnal ebb tides; this strategy allows them to escape from planktivorous fish, which are especially abundant in estuaries and near-shore areas and less common in the coastal ocean ("predator avoidance hypothesis", Christy, 2003 and references therein).in fact, high densities of Uca tangeri zoeae i were found during nocturnal ebb tides (rodríguez et al., 1997).Survival at 24 suggests that U. tangeri zoeae could also develop in polyhaline conditions, if they were retained in the ocean-estuary interface.Eventually, retention of some larvae in low estuarine waters may compensate for an excessive larval dispersal due to the export strategy.
variability in survival and development duration among clutches from different females has been observed in other intertidal and estuarine crabs, such as Armases rubripes (Luppi et al., 2003).a variation among clutches in optimal salinity was also observed in the calanoid copepod eurytemora affinis (Lee and petersen, 2002;devreker et al., 2004) and it has been proposed that "the variability at the clutch level could be a 'reservoir of genetic diversity' to respond either to short-or long-term changes in salinity (i.e. in the case of fresh-or seawater inflows)" (Lee, 1999).Crabs, in fact, have been used as models to examine the complex physiological relationships between phases of the life cycle.For example, tolerance of neohelice granulata larvae to low salinities depend on two factors: 1) the salinity prevailing during the embryonic development and 2) the initial larval biomass (Giménez and anger, 2003).
the differences between zoea v and zoea vi in Uca tangeri were the number of aesthetascs and setae of the antennule, the number of setae of the maxillule, maxilla and maxillipeds 1 and 2, and, to a lesser extent, the number of dorsal setae of the first abdominal segment, as was observed in U. mordax, U. burgersi and U. uruguayensis by rieger (1996, 1997 and 1998).However, the zoea vi of Uca tangeri had an additional seta in the distal segment of the first maxilliped endopod (setation: 2,3,2,2,6), while that of U. mordax, U. burgersi, U. uruguayensis remained 2,3,2,2,5, as in zoea v (rieger, 1996, 1997 and 1998).

Larval morphology and Uca phylogeny
it has been suggested that U. tangeri should belong to a separated and monospecific subgenus on the basis of adult morphology (Crane, 1975;rosenberg, 2001); this subgenus has been named Afruca.Molecular studies also support a differentiation between Uca (Afruca) tangeri and the Uca (Uca) s. str.clade (Levinton et al., 1996;Sturmbauer et al., 1996).the spines observed in the telson of all U. tangeri zoeae (paula, 1985; this paper), as well as the presence of lateral spines in the cephalotorax (paula, 1985;rodríguez and Jones, 1993), have not been found in any other Uca species described up to this moment.these are new evidences that support the Afruca hypothesis as well as its basal position in a proto-atlantic origin of the genus Uca suggested by molecular data (Levinton et al., 1996;Sturmbauer et al., 1996).recent molecular data also point out the separation of this species in a monotypic subgenus of Uca (Landstorfer, Felder and Schubart, unpublished).the spermatozoal ultrastructure, however, did not allow U. tangeri to be separated from the other Uca species: centrioles were present in adult spermatozoa of the american species U. maracoani, U. thayeri and U. vocator, and absent in both U. tangeri and the indo-pacific U. dussumieri (Benetti et al., 2008).putting together morphological (adult and larval) and molecular evidences, we suggest that Uca tangeri should be separated from the rest of the Uca species and placed in its own genus as Afruca tangeri by moving the present status of the subgenus Afruca to the genus level.paula (1985) mentioned the dorsal spine on furcal arms of the telson of zoea i, but not the lateral and ventral ones.None of these spines was observed by rodríguez and Jones (1993), probably because of their small size.Likewise, similar small lateral spines of the furcal arms, which had also been overlooked in previous descriptions, have been recently recognised as a typical character of sesarmid zoeae (Cuesta et al. 2006a(Cuesta et al. , 2006b)).However, the spines described in U. tangeri's telson are very small and their presence in zoeae of other Uca species, including those already described, should not be discarded.in the case of Uca speciosa zoea i, only a minute scale-like lateral spine is present on each furcal arm, and dorsal and ventral spines are absent (Cuesta and Schubart, unpublished data); also, only 1 or 0-2 minute lateral spines have been described in the zoea i and ii of Uca annulipes (Hashmi, 1968) and zoea i of Uca subcylindrica (rabalais and Cameron, 1983), respectively.

Geographical distribution: Does low tolerance of zoeae to cold water explain the northern limit of U. tangeri?
temperatures also affect crustacean larval development and U. tangeri is not the exception.it has been widely demonstrated that development duration increases at lower temperatures in crustaceans (see ismael et al., 1997, and references therein).Comparing our data (salinity 32) with those by rodríguez and Jones (1993) who reared U. tangeri at 25ºC, we observed that a temperature decrease of 5ºC caused a duplication of the larval span by increasing the duration of each stage (Fig. 3).although it was often suggested that delayed development increased the possibility of dispersal, a longer larval development is correlated with a rise in mortality so its advantages are not obvious.
the northern population of U. tangeri inhabits the Mira estuary (37º43'N 8º47'W) (Costa et al., 2001).Mortality associated with an increase in duration of development at low temperatures and the latitudinal temperature diminution could limit the larval dispersal of this species, and consequently its northern distribution.also, the northern limit of Uca pugnax in the northwestern atlantic is likely maintained by the influence of cooler water temperatures on the larval phase, as reported by Sanford et al. (2006), who found that there are suitable northern habitats for this species (extensive salt marshes) and that trans-planted adult crabs could resist severe winter conditions.However, few larvae completed development at temperatures that they would regularly encounter north of their range boundary.the present distribution of U. tangeri in Europe could be explained by a combination of ecological and physiological experiments and knowledge of local hydrographical processes; this multidisciplinary study would contribute to the prediction of changes arising from an increase in sea temperature.aCkNoWLEdGEMENtS this study was carried out during a 6-month visit of EdS to the instituto de Ciencias Marinas de andalucía (iCMaN), supported by the Ministerio de Educación y Ciencia de España (MEC) grant to foreign professors and researchers visiting Spanish laboratories during their sabbatical years (SaB2005-0170).EdS would like to thank the MEC for funding the travelling and living expenses, the iCMaN for the use of the facilities, and the universidad de Mar del plata (argentina) for the sabbatical license; my special thanks to the authorities and members of the iCMaN for the excellent time spent in puerto real.thanks are due to richard Landstorfer and Christoph d.Schubart (universität regensburg) and darryl L. Felder (university of Louisiana at Lafayette) for the information provided, to dr. klaus anger, scientific editor of this paper, to the two anonymous reviewers, and to ana Cosulich for her revision of the English version.

Table 2 .
-Uca tangeri.results of a test (mixed-factor aNova model with unequal replicates;